JP7758264B2 - Ventricular Far-Field Estimation Using Autoencoders - Google Patents

Description

The present invention relates to artificial intelligence and machine learning autoencoders related to ventricular far-field estimation, and to the identification and decomposition of near-field and far-field signals in cardiac electrical activity.

Treatment for cardiac disorders, such as cardiac arrhythmias, often requires cardiac mapping (i.e., mapping the heart tissue, chambers, veins, arteries, and/or pathways, also known as cardiac mapping). An electrocardiogram or electrocardiograph (ECG) is an example of cardiac mapping. An ECG is generated from electrical signals from the heart that describe the heart's activity.

ECGs are utilized during cardiac surgery to identify potential origin locations of cardiac disorders. Generally, when physicians examine cardiac activity, signal interference, signal artifacts, and signal noise associated with the underlying electrical signals of the ECG can significantly obscure the accuracy of the ECG. Signal interference can also result from processing regions of the signal that contain sharp transitions, peaks, and/or high-frequency and harmonic regions of the pacing signal. These interferences, artifacts, and noise prevent physicians from separating ventricular and atrial origin locations in real-time cases (e.g., during cardiac surgery), which increases the difficulty of diagnosing/treating cardiac disorders. Therefore, there is a need to provide an improved method for cardiac mapping that eliminates such interference, artifacts, and noise.

A unipolar signal is a combination of near-field and far-field signals. It is important to identify and isolate the near-field signals during an ablation procedure. When an electrode is inserted into a muscle, such as the myocardium, activation of each muscle generates an electric field. Each electrode captures all sources of electric fields at its location, including near-field signals near the electrode and far-field signals far from the electrode.

According to one embodiment, a method is provided. The method includes receiving input intracardiac signals from a monitoring and processing device. Each of the input intracardiac signals may include at least an artifact. The method includes encoding, with an autoencoder, the input intracardiac signals using an intracardiac data set to generate a latent representation. The method also includes decoding, with the autoencoder, the latent representation to generate an output intracardiac signal. The output intracardiac signal may include the input intracardiac signal reconstructed without the artifact.

According to one embodiment, a method for decomposing near-field and far-field signals is provided. Measured signals may be received. The measured signals may be encoded by an autoencoder to generate latent representations. The latent representations may be decoded by an autoencoder to decompose near-field and far-field components from the measured signals. Far-field ventricular measurements may be obtained. The measurements may be obtained using a multiple electrode catheter and a body surface ECG signal. Composite local field signals may be added. A derived far-field signal and a residual near-field signal may be detected. Decoding the latent representation may be based on the detected derived far-field signal and the residual near-field signal.

According to one or more embodiments, the above method embodiments may be implemented as an apparatus, a system, and/or a computer program product.

A more detailed understanding can be had from the following description, given by way of example in conjunction with the accompanying drawings, in which like reference numerals indicate similar elements and in which:
1 shows a schematic diagram of an exemplary system in which one or more features of the presently disclosed subject matter may be implemented, according to one or more embodiments. FIG. 1 shows a block diagram of an exemplary system for remotely monitoring and communicating patient biometric indicators, according to one or more embodiments. 1 illustrates a graphical depiction of an artificial intelligence system according to one or more embodiments. 4 shows a block diagram of a method performed by the artificial intelligence system of FIG. 3 according to one or more embodiments. 1 illustrates an example of a neural network in accordance with one or more embodiments. 1 illustrates a block diagram of a method according to one or more embodiments. 1 illustrates a graphical representation of a signal according to one or more embodiments. 1 illustrates a graphical representation of a signal according to one or more embodiments. 1 illustrates a graphical depiction of a signal in accordance with one or more embodiments. 1 illustrates a graphical depiction of signal progression according to one or more embodiments. 1 illustrates a block diagram of a method according to one or more embodiments. FIG. 2 is an exemplary flow diagram of an exemplary method for decomposing near-field and far-field signals, according to one embodiment.

Disclosed herein is an artificial intelligence and machine learning autoencoder (generically referred to herein as an autoencoder). The autoencoder may be processor-executable code or software implemented in performing processing operations by and within processing hardware of a medical device to provide improved ECGs for treating cardiac disease. According to one embodiment, the autoencoder may provide a specific encoding and decoding method for a medical device. This specific encoding and decoding method may involve multi-stage data manipulation of the cardiac electrical signal, removing signal interference, signal artifacts, and signal noise from the electrical signal.

In this regard, during operation, the autoencoder may receive input intracardiac signals (e.g., electrical signals of the heart including signal interference, signal artifacts, and signal noise). The intracardiac signals may be recorded and processed in real time by a monitoring and processing device (e.g., a catheter having an autoencoder therein) and/or may be recorded by the monitoring and processing device and transmitted to a computing device having an autoencoder therein.

The autoencoder may encode an input intracardiac signal using an intracardiac data set (e.g., predetermined and recognized electrical signals of the heart, free from signal interference, signal artifacts, and signal noise). This encoding by the autoencoder may generate a latent representation from the input intracardiac signal. The autoencoder may further decode the latent representation to generate an output intracardiac signal. The output intracardiac signal may be a reconstructed input intracardiac signal free from signal interference, signal artifacts, and signal noise. An improved ECG for treating cardiac disease is then generated from the output intracardiac signal.

The technical effect of the autoencoder includes generating its output intracardiac signal in real time, which further enables the generation of an improved ECG for a physician (e.g., during cardiac surgery) who can use the improved ECG to probe cardiac activity and identify potential origin locations of cardiac disease. The improved ECG is not obscured by signal interference, signal artifacts, and signal noise of the original input intracardiac signal, because these artifacts have been removed during decoding. Further, the technical effect of the autoencoder includes more accurately generating an improved ECG from which signal interference, signal artifacts, and signal noise have been removed, which can provide separate ventricular and atrium origin locations in real time.

In one embodiment, an autoencoder can be used to train a system to resolve near-field and far-field signals detected by the electrodes from analyzing a large number of data points. Bits may be selected as part of a training set and used to train the system to recognize the far-field signal components.

A signal may be provided and an attempt may be made to reconstruct it by providing signals with a large amount of far-field signal and signals with a large amount of both far-field and near-field signal. An autoencoder may be provided with signals to reconstruct the far-field signal. Once the network is trained, the far-field components may be output from the signals provided by the network.

FIG. 1 is a schematic diagram of an example system 100 (e.g., a medical device) in which one or more features of the presently disclosed subject matter may be implemented. All or a portion of system 100 may be used to collect information from an intracardiac dataset (e.g., a training dataset) and/or all or a portion of system 100 may be used to implement an autoencoder as described herein.

System 100 may include components such as catheter 105 configured to inflict damage on a tissue region of an internal organ. Catheter 105 may also be further configured to acquire biometric data, including cardiac electrical signals (e.g., intracardiac signals). While catheter 105 is shown to be a point catheter, it will be understood that any shape of catheter including one or more elements (e.g., electrodes) may be used to practice the embodiments disclosed herein.

The system 100 includes a probe 110 having a shaft that a physician or medical professional 115 can navigate into a body part, such as the heart 120, of a patient 125 lying on a bed (or table) 130. According to various embodiments, multiple probes may be provided, but for simplicity, a single probe 110 is described herein. It is further understood that the probe 110 may represent multiple probes.

Exemplary system 100 can be utilized to detect, diagnose, and treat cardiac disorders (e.g., using intracardiac signals). Cardiac disorders such as cardiac arrhythmias (particularly atrial fibrillation) persist as common and dangerous medical disorders, especially in the aging population. In a patient (e.g., patient 125) with normal sinus rhythm, the heart (e.g., heart 120), which includes atria, ventricles, and excitatory conduction tissue, is electrically stimulated to beat in a synchronous, patterned manner. This electrical excitation can be detected as intracardiac signals.

In a patient with cardiac arrhythmia (e.g., patient 125), abnormal regions of cardiac tissue do not follow the synchronous beating cycle associated with normal conductive tissue, as in a patient with normal sinus rhythm. Instead, the abnormal regions of cardiac tissue conduct abnormally to adjacent tissue, disrupting the cardiac cycle and resulting in an asynchronous cardiac rhythm. Asynchronous cardiac rhythms can also be detected as intracardiac signals. Such abnormal conduction has been known for some time and occurs in various regions of the heart (e.g., heart 120), along the conduction pathway of the atrioventricular (AV) node, for example, in the region of the sinoatrial (SA) node, or in the myocardial tissue forming the walls of the ventricular and atrial chambers.

Additionally, cardiac arrhythmias, including atrial arrhythmias, may be multi-wavelet reentrant, characterized by multiple asynchronous loops of electrical impulses scattered and often self-propagating around the atria (e.g., another example of intracardiac signals). Alternatively, or in addition to multi-wavelet reentrant, cardiac arrhythmias may also have a local origin, such as when isolated regions of tissue within the atria are excited autonomously in a rapid, repetitive manner (e.g., another example of intracardiac signals). Ventricular tachycardia (V-tach or VT) is a tachycardia or fast cardiac rhythm that occurs in one of the ventricles of the heart. It is a potentially fatal arrhythmia because it can lead to ventricular fibrillation and sudden death.

Atrial fibrillation, a type of arrhythmia, occurs when normal electrical impulses generated by the sinoatrial node (e.g., another example of intracardiac signaling) are overwhelmed by chaotic electrical impulses originating in the atria and pulmonary veins, causing irregular impulses to be conducted to the ventricles. The resulting irregular heartbeat can persist for minutes to weeks or even years. Atrial fibrillation (AF) is often a chronic condition that carries a small increased risk of death, often from stroke. The first-line treatment for AF is medication to slow the heart rate or restore normal heart rhythm. Additionally, patients with AF are often given anticoagulants to protect against the risk of stroke. The use of such anticoagulants carries its own risks: internal bleeding. In some patients, medication is insufficient, and their AF is deemed drug-refractory, meaning it cannot be treated with standard pharmacological interventions. Synchronized electrical cardioversion can also be used to convert AF to a normal heart rhythm. Alternatively, patients with AF can be treated with catheter ablation.

Catheter ablation-based therapy may involve mapping the electrical properties of cardiac tissue, particularly the endocardium and cardiac volumes, and selectively ablating the cardiac tissue through the application of energy. Cardiac mapping (e.g., heart mapping) may involve, for example, creating an electrical potential map (e.g., a voltage map) of wave propagation along cardiac tissue or a map of arrival times to various tissue location points (e.g., a local time activation (LAT) map). Cardiac mapping can be used to detect local cardiac tissue dysfunction. Ablation, such as that based on cardiac mapping, can stop or modify the propagation of unwanted electrical signals from one part of the heart to another.

Ablation techniques disrupt unwanted electrical pathways by creating non-conductive lesions. Various energy delivery modalities have been previously disclosed for creating lesions, including the use of microwave, laser, and more commonly radiofrequency energy to create conduction blocks along cardiac tissue walls. In a two-stage mapping-followed-by-ablation procedure, electrically active points at each intracardiac point are typically sensed and measured by advancing a catheter (e.g., catheter 105) containing one or more electrical sensors (e.g., at least one ablation electrode 134 of catheter 105) into the heart (e.g., heart 120) and acquiring data at multiple points. This data (e.g., biological data including intracardiac signals) is then used to select an endocardial target region where ablation will be performed. Due to the use of an autoencoder employed by the exemplary system 100 (e.g., medical device), this data is more accurate than the underlying electrical signals of an ECG, which contain signal interference, signal artifacts, and signal noise, and can better support the selection of endocardial target regions for ablation. Signal interference, signal artifacts, and signal noise may be collectively referred to herein as artifacts. Examples of artifacts include, but are not limited to, power noise (e.g., electrostatic and electromagnetic coupling between circuits and 50 or 60 Hz power lines), furo noise (e.g., fluorescent lighting), contact noise (e.g., collisions between catheter electrodes), and deflection noise (e.g., static discharge during catheter deflection).

Cardiac ablation and other cardiac electrophysiology procedures are becoming increasingly complex as physicians treat challenging conditions such as atrial fibrillation and ventricular tachycardia. Treatment of complex arrhythmias may currently rely on the use of three-dimensional (3D) mapping systems to reconstruct the anatomical structure of the targeted cardiac chamber. In this regard, the autoencoder employed by the exemplary system 100 (e.g., a medical device) herein provides a fundamental output signal that can result in improved 3D maps and/or ECGs for treating cardiac conditions.

For example, cardiologists rely on software such as the Complex Fractionated Atrial Electrograms (CFAE) module of the CARTO® 3 3D mapping system from Biosense Webster, Inc. (Diamond Bar, Calif.) to generate and analyze intracardiac electrograms (EGMs). The autoencoder of the exemplary system 100 (e.g., a medical device) enhances this software to generate and analyze improved intracardiac electrograms (EGMs), thereby enabling ablation site determination for the treatment of a wide range of cardiac disorders, including irregular atrial flutter and ventricular tachycardia.

Improved 3D maps supported by autoencoders can provide multiple pieces of information about the electrophysiological properties of tissues, representing the anatomical and functional substrates of these challenging arrhythmias.

Cardiomyopathy of different etiologies (e.g., ischemic, dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM), arrhythmogenic right ventricular dysplasia (ARVD), left ventricular non-compaction (LVNC)) are characterized by areas of unhealthy tissue surrounded by areas of normally functioning cardiomyocytes with a distinct substrate.

Abnormal tissue is generally characterized by low-voltage EGMs. However, initial clinical experience with endocardial-epicardial mapping has shown that low-voltage regions are not always present as the sole arrhythmogenic mechanism in these patients. Indeed, low- or intermediate-voltage regions may exhibit EGM fragmentation and delayed activity during sinus rhythm, corresponding to critical stenoses identified during sustained and coherent ventricular arrhythmias, e.g., only in intolerable ventricular tachycardia. Furthermore, EGM fragmentation and delayed activity are often observed in regions exhibiting normal or near-normal voltage amplitudes (>1-1.5 mV). While the latter regions can be evaluated according to voltage amplitude, they may not be considered normal according to the intracardiac signal and therefore represent true arrhythmogenic substrates. 3D mapping can identify the location of arrhythmogenic substrates in the endocardial and/or epicardial layers of the right and/or left ventricles, whose distribution may vary depending on the primary disease progression.

These cardiac disease-related substrates are correlated with the subdivision of the endocardial and/or epicardial layers of the ventricular chambers (right and left) and the presence of delayed EGMs. 3D mapping systems such as the CARTO® 3 can identify the location of potential arrhythmogenic substrates for cardiomyopathies in relation to abnormal EGM detection.

Electrode catheters (e.g., catheter 105) are used during medical procedures to stimulate and map electrical activity within the heart and to ablate sites of abnormal electrical activity. In use, the electrode catheter is inserted into a major vein or artery, such as the femoral artery, and then guided into a chamber of the subject's heart. A typical ablation procedure involves inserting a catheter having at least one electrode at its distal end into the heart chamber. A reference electrode is typically provided by a second catheter taped to the patient's skin or placed in or near the heart. Radio frequency (RF) current is applied to the tip electrode of the ablation catheter, causing current to flow through the medium surrounding the tip electrode, i.e., blood and tissue, toward the reference electrode. The distribution of the current depends on the amount of electrode surface in contact with the tissue compared to blood, which has a higher electrical conductivity than tissue. Tissue heating occurs due to the electrical resistance of the tissue. Sufficient tissue heating causes cell destruction in the cardiac tissue, resulting in lesions in the non-conductive cardiac tissue. During this process, the electrode also heats due to conduction from the heated tissue to the electrode itself. If the electrode temperature becomes high enough, possibly above 60 degrees Celsius, a thin, transparent film of dehydrated blood proteins can form on the surface of the electrode. If the temperature continues to rise, this dehydrated layer gradually thickens, resulting in blood coagulation on the electrode surface. Because dehydrated biological material has a higher electrical resistance than endocardial tissue, the impedance to the flow of electrical energy into the tissue also increases. If the impedance becomes high enough, an impedance rise occurs, requiring the catheter to be removed from the body and the tip electrode to be cleaned.

Treatment of cardiac disorders, such as cardiac arrhythmias, often requires obtaining detailed mapping of cardiac tissue, chambers, veins, arteries, and/or electrical pathways. For example, a prerequisite for successful catheter ablation is accurate localization of the source of the cardiac arrhythmia within a cardiac chamber. Such localization can be performed by electrophysiological studies, during which spatially resolved electrical potentials are detected by a mapping catheter introduced into the cardiac chamber. This electrophysiological study is so-called electroanatomical mapping and thus provides 3D mapping data that can be displayed on a monitor. Often, mapping and therapy functions (e.g., ablation) are provided by a single catheter or group of catheters, such that the mapping catheter also simultaneously operates as a therapy (e.g., ablation) catheter. In this case, the autoencoder can be stored and executed directly by the catheter 105.

Mapping of cardiac regions, such as cardiac regions, tissues, veins, arteries, and/or electrical pathways of a heart (e.g., 120), can result in the identification of problem areas, such as scar tissue, arrhythmia sources (e.g., electrical rotors), and healthy regions. As further disclosed herein, cardiac regions can be mapped such that a visual rendering of the mapped cardiac regions is provided using a display. Additionally, cardiac mapping can include mapping based on one or more modalities, such as, but not limited to, local activation time (LAT), electrical activity, topology, bipolar mapping, dominant frequency, or impedance. Data corresponding to multiple modalities can be captured using catheters inserted within the patient's body and provided for rendering simultaneously or at different times based on corresponding settings and/or medical professional preferences.

Cardiac mapping can be performed using one or more techniques. As an example of a first technique, cardiac mapping can be performed by sensing the electrical properties of cardiac tissue, e.g., the LAT, as a function of precise location within the heart. The corresponding data can be acquired using one or more catheters advanced into the heart using catheters having electrical and position sensors at their distal tips. As a specific example, location and electrical activity can be initially measured at approximately 10 to approximately 20 points on the inner surface of the heart. These data points can usually be sufficient to generate a preliminary reconstruction or map of the cardiac surface of satisfactory quality. This preliminary map can be combined with data acquired at additional points to generate a more comprehensive map of the cardiac electrical activity. In clinical settings, it is not uncommon to accumulate data at 100 or more sites to generate a detailed, comprehensive map of the cardiac chamber's electrical activity. The detailed map can then serve as the basis for making decisions regarding therapeutic action, e.g., tissue ablation, to alter the propagation of the cardiac electrical activity and restore normal cardiac rhythm.

Returning to FIG. 1 , to perform cardiac mapping of interest, the medical professional 115 can insert the shaft 137 through the sheath 136 while manipulating the distal end of the shaft 137 using a manipulator 138 near the proximal end of the catheter 105 and/or deflection from the sheath 136. As shown in inset 140, the catheter 105 can be attached at the distal end of the shaft 137. The catheter 105 can be inserted through the sheath 136 in a collapsed state and then expanded within the heart 120. As described further herein, the catheter 105 can include at least one ablation electrode 134 and a catheter needle.

According to several embodiments, the catheter 105 can be configured to ablate a tissue region of a chamber of the heart 120. Inset 150 shows the catheter 105 in a close-up view inside a chamber of the heart 120. As shown, the catheter 105 can include at least one ablation electrode 134 coupled to the body of the catheter. According to other embodiments, multiple elements can be connected via splines that define the shape of the catheter 105. One or more other elements (not shown) can be provided and can be any element configured to perform ablation or acquire biometric data, such as an electrode, a transducer, or one or more other elements.

According to embodiments disclosed herein, an ablation electrode, such as at least one ablation electrode 134, can be configured to deliver energy to a tissue region of an internal organ, such as the heart 120. The energy can be thermal energy and can cause damage to the tissue region starting at the surface of the tissue region and extending through the thickness of the tissue region.

According to embodiments disclosed herein, the biological data may include one or more of LAT, electrical activity, topology, bipolar mapping, dominant frequency, impedance, etc. The LAT may be a time point of threshold activity corresponding to local activation calculated based on a normalized initial starting point. The electrical activity may be any applicable electrical signal that can be measured based on one or more thresholds and detected and/or enhanced based on signal-to-noise ratio and/or other filters. The topology may correspond to the physical structure of a body part or portion of a body part, and may correspond to changes in the physical structure for different portions of the body part or for different body parts. The dominant frequency may be a frequency or range of frequencies prevalent in a portion of a body part and may be different in different portions of the same body part. For example, the dominant frequency of the pulmonary veins of a heart may be different from the dominant frequency of the right atrium of the same heart. The impedance may be a resistance measurement in a given region of a body part.

As shown in FIG. 1, the probe 110 and catheter 105 can be connected to a console 160. The console 160 may include a computing device 161 that employs an autoencoder as described herein. According to one embodiment, the console 160 and/or computing device 161 include at least a processor and a memory, where the processor executes computer instructions related to the autoencoder as described herein and the memory stores instructions for execution by the processor.

Computing device 161 may be any computing device including software and/or hardware, such as a general-purpose computer, with appropriate front-end and interface circuitry 162 for transmitting and receiving signals to and from catheter 105 and for controlling other components of system 100. Computing device 161 may include real-time noise reduction circuitry, typically configured as a field programmable gate array (FPGA), followed by an analog-to-digital (A/D) electrocardiograph or electromyogram (EMG) signal conversion integrated circuit. Computing device 161 may pass signals from the A/D ECG or EMG circuitry to a separate processor and/or may be programmed to perform one or more functions disclosed herein.

For example, the one or more functions may include receiving an input intracardiac signal, encoding the input intracardiac signal using an intracardiac data set to generate a latent representation, and decoding the latent representation to generate an output intracardiac signal. The front-end and interface circuit 162 includes an input/output (I/O) communication interface that enables the console 160 to receive signals from and/or transmit signals to at least one ablation electrode 134.

In some embodiments, computing device 161 may be further configured to receive biometric data, such as electrical activity, and determine whether a given tissue region conducts electricity. According to certain embodiments, computing device 161 may be external to console 160, for example, located in a catheter, an external device, a mobile device, a cloud-based device, or may be a stand-alone processor.

As mentioned above, computing device 161 may include a general-purpose computer that can be programmed with software to perform the functions of the autoencoder described herein. The software may be downloaded to the general-purpose computer in electronic form, for example, over a network, or alternatively or additionally, may be provided and/or stored on non-transitory tangible media such as magnetic, optical, or electronic memory (e.g., any suitable volatile and/or non-volatile memory, such as random access memory or a hard disk drive). The exemplary configuration shown in FIG. 1 may be modified to implement embodiments disclosed herein. Embodiments of the present disclosure may be similarly applied using other system components and configurations. Furthermore, system 100 may include additional components, such as elements for sensing electrical activity, wired or wireless connectors, processing and display devices, etc.

According to certain embodiments, a display 165 is connected to the computing device 161. During a procedure, the computing device 161 facilitates the presentation of body part renderings to the medical professional 115 on the display 165 and can store data representing the body part renderings in memory. In some embodiments, the medical professional 115 may be able to manipulate the body part renderings using one or more input devices, such as a touchpad, mouse, keyboard, gesture recognizer, etc. For example, the input device may be used to change the position of the catheter 105, which updates the rendering. In alternative embodiments, the display 165 may include a touchscreen that, in addition to presenting the body part renderings, may be configured to receive input from the medical professional 115. The display 165 may be located at the same location or at a remote location, such as a remote hospital, or within a separate healthcare provider network. Furthermore, the system 100 may be part of a surgical system configured to obtain anatomical and electrical measurements of a patient's organs, such as the heart 120, and to perform cardiac ablation procedures. One example of such a surgical system is the Carto® system sold by Biosense Webster.

The console 160 can be connected by a cable to body surface electrodes, which can include adhesive skin patches that are applied to the patient 125. The processor, in conjunction with the current tracking module, can determine position coordinates of the catheter 105 inside a body portion (e.g., the heart 120) of the patient 125. The position coordinates can be based on impedance or electromagnetic fields measured between the body surface electrodes and electrodes or other electromagnetic components (e.g., at least one ablation electrode 134) of the catheter 105. Additionally or alternatively, location pads can be placed on the surface of the bed 130 or can be separate from the bed 130.

System 100 can also, and optionally, acquire biometric data, such as anatomical measurements of heart 120, using ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), or other medical imaging techniques known in the art. System 100 can acquire ECG or electrical measurements using catheters or other sensors that measure electrical properties of heart 120. The biometric data, including the anatomical and electrical measurements, can then be stored in a non-transitory tangible medium of console 160. The biometric data can be transmitted from the non-transitory tangible medium to computing device 161. Alternatively, or in addition, the biometric data can be transmitted to a server, which can be local or remote, using a network as further described herein.

According to one or more embodiments, a catheter including a position sensor can be used to determine the trajectories of points on the cardiac surface. These trajectories can be used to infer motion characteristics, such as the contractile force of the tissue. A map indicative of such motion characteristics can be constructed when trajectory information is sampled at a sufficient number of points within the heart 120.

Electrical activity at a point within the heart 120 can typically be measured by advancing a catheter 105 containing an electrical sensor at or near its distal tip (e.g., at least one ablation electrode 134) to that point within the heart 120, contacting tissue with the sensor and acquiring data at that point. One drawback to mapping a heart chamber using a catheter 105 containing only a single distal tip electrode is the long period of time required to accumulate data point-by-point across the requisite number of points needed for a detailed map of the heart chamber as a whole. Therefore, multi-electrode catheters have been developed to simultaneously measure electrical activity at multiple points within a heart chamber.

The multi-electrode catheter can be implemented using any applicable shape, such as a linear catheter with multiple electrodes, a balloon catheter including electrodes arranged on multiple balloon-forming frameworks, a lasso or loop catheter with multiple electrodes, or any other applicable shape. The linear catheter may be fully or partially elastic so that it can twist, bend, or otherwise change its shape based on received signals and/or the application of an external force (e.g., cardiac tissue) to the linear catheter. The balloon catheter may be designed so that its electrodes can be held in close contact against the endocardial surface when deployed within a patient's body. As an example, the balloon catheter can be inserted into a lumen such as a pulmonary vein (PV). The balloon catheter can be inserted into the PV in a deflated state so that the balloon catheter does not occupy the full volume of the PV while inserted within the PV. The balloon catheter can be inflated inside the PV with such electrodes on the balloon catheter in contact with the entire circular area of the PV. Such contact with the entire circular portion of the PV, or any other lumen, may allow for efficient mapping and/or ablation.

According to one example, a multi-electrode catheter can be advanced into a chamber of the heart 120. Anteroposterior (AP) and lateral fluoroscopy views can be acquired to establish the position and orientation of each of the electrodes. An EGM can be recorded from each of the electrodes in contact with the cardiac surface relative to a temporal reference, such as the occurrence of a P wave in sinus rhythm from a surface ECG. As further disclosed herein, the system can distinguish between those electrodes that register electrical activity and those that do not due to their lack of proximity to the endocardial wall. After the initial EGM is recorded, the catheter can be repositioned, and fluoroscopy views and EGMs can be recorded again. An electrical map can then be constructed from a repetition of the above process.

According to one example, cardiac mapping can be generated based on the detection of intracardiac electrical fields. Non-contact techniques can be implemented to simultaneously acquire vast amounts of cardiac electrical information. For example, a catheter having a distal tip portion can include a series of sensor electrodes distributed over its entire surface and connected to insulated conductors for connection to signal sensing and processing means. The tip portion can be sized and shaped such that the electrodes are spaced a significant distance from the wall of the cardiac chamber. The intracardiac electrical fields can be detected during a single cardiac beat. According to one example, the sensor electrodes can be distributed on a series of circumferential electrodes in spaced-apart planes. These planes can be perpendicular to the longitudinal axis of the catheter tip portion. At least two additional electrodes can be provided adjacent each end of the longitudinal axis of the tip portion. As a more specific example, the catheter can include four circumferences with eight electrodes equiangularly spaced on each circumference. Thus, in this particular implementation, the catheter can include at least 34 electrodes (32 circumferential electrodes and two distal electrodes).

According to another example, electrophysiological cardiac mapping systems and techniques based on non-contact and non-expandable multi-electrode catheters can be implemented. An EGM can be obtained using a catheter with multiple electrodes (e.g., 42-122 electrodes). According to this implementation, knowledge of the relative geometry of the probe and endocardium can be obtained, for example, through an independent imaging technique such as transesophageal echocardiography. After the independent imaging, non-contact electrodes can be used to measure cardiac surface potentials, from which a map can be constructed. This technique can include the following steps (after the independent imaging step): (a) measuring potentials using multiple electrodes disposed on a probe placed within the heart 120; (b) determining the geometric relationship between the probe surface and the endocardium surface; (c) generating a matrix of coefficients representing the geometric relationship between the probe surface and the endocardium surface; and (d) determining the endocardium potentials based on the electrode potentials and the coefficient matrix.

According to another example, a technique and apparatus can be implemented for mapping the electrical potential distribution of a cardiac chamber. An intracardiac multi-electrode mapping catheter assembly can be inserted into the patient's heart 120. The mapping catheter assembly can include a multi-electrode array with an integrated reference electrode, or preferably, a companion reference catheter. These electrodes can be deployed in a substantially spherical array. The electrode array can be spatially referenced to a point on the endocardial surface by the reference electrode or by a reference catheter in contact with the endocardial surface. A preferred electrode array catheter can carry a large number of individual electrode sites (e.g., at least 24). Furthermore, this exemplary technique can be implemented knowing the location of each of the electrode sites on the array as well as the cardiac geometry. These locations are preferably determined by impedance plethysmography techniques.

According to another example, a cardiac mapping catheter assembly can include an electrode array defining multiple electrode sites. The mapping catheter assembly can also include a lumen for receiving a reference catheter having a distal tip electrode assembly that can be used to probe the heart wall. The mapping catheter can include a braid of insulated wires (e.g., having 24-64 wires within the braid), each of which can be used to form an electrode site. The catheter can be easily positioned in the heart 120 to be used to acquire electrical activity information from a first set of non-contact electrode sites and/or a second set of contact electrode sites.

According to another example, another catheter for mapping electrophysiological activity within the heart can be implemented. The catheter body can include a distal tip adapted to deliver stimulation pulses for pacing the heart or an ablation electrode for ablating tissue in contact with the tip. The catheter can further include at least a pair of orthogonal electrodes capable of generating a differential signal indicative of local cardiac electrical activity adjacent the orthogonal electrodes.

According to another embodiment, a process for measuring electrophysiological data within a heart chamber can be implemented. The method can include, in part, placing a set of active and passive electrodes within the heart 120, generating an electric field within the heart chamber by applying an electric current to the active electrodes, and measuring the electric field at the passive electrode sites. The passive electrodes are contained within an array disposed on an inflatable balloon of a balloon catheter. In a preferred embodiment, the array is said to have 60-64 electrodes.

According to another example, cardiac mapping can be performed using one or more ultrasound transducers. The ultrasound transducers can be inserted into a patient's heart 120 and can acquire multiple ultrasound slices (e.g., two-dimensional or three-dimensional slices) at various locations and orientations within the heart 120. The location and orientation of a given ultrasound transducer can be known, and the acquired ultrasound slices can be stored for later display. One or more ultrasound slices corresponding to the position of a probe (e.g., a treatment catheter) can be later displayed, and the probe can be overlaid on one or more ultrasound slices.

According to another example, the body patch and/or body surface electrodes may be positioned on or adjacent to the patient's body. A catheter having one or more electrodes may be positioned within the patient's body (e.g., within the patient's heart 120), and the position of the catheter may be determined by the system based on signals transmitted and/or received between one or more electrodes of the catheter and the body patch and/or body surface electrodes. Additionally, the catheter electrodes may sense biometric data (e.g., LAT values) from within the patient's body (e.g., within the heart 120). The biometric data may be correlated to the determined catheter position, such that a rendering of the patient's body part (e.g., heart 120) may be displayed, showing the biometric data superimposed on the body shape.

2, a block diagram of an exemplary system 200 for remotely monitoring and transmitting vital data (i.e., patient vital information, patient data, or patient biometric data) is shown. In the example shown in FIG. 2, the system 200 includes a monitoring and processing device 202 (i.e., patient data monitoring and processing device) associated with a patient 204, a local computing device 206, a remote computing system 208, a first network 210, and a second network 211. In accordance with one or more embodiments, the monitoring and processing device 202 may be an example of the catheter 105 of FIG. 1, the patient 204 may be an example of the patient 125 of FIG. 1, and the local computing device 206 may be an example of the console 160 of FIG. 1.

The monitoring and processing device 202 includes a patient biometric sensor 212, a processor 214, a user input (UI) sensor 216, a memory 218, and a transmitter-receiver (i.e., transceiver) 222. During surgery, the monitoring and processing device 202 acquires biometric data (e.g., electrical signals, blood pressure, temperature, blood glucose levels, or other biometric data) of the patient 204 and/or receives at least a portion of the biometric data representing any acquired patient biometric information, as well as additional information associated with any acquired patient biometric information, from one or more other patient biometric monitoring and processing devices. The additional information may be, for example, diagnostic information and/or additional information obtained from an additional device, such as a wearable device.

The monitoring and processing device 202 can use the autoencoder described herein to process data, including acquired biometric data, as well as any biometric data received from one or more other patient biometric monitoring and processing devices. For example, when processing data in this regard, the autoencoder may include a neural network used to learn latent representations (or data encodings) from the biometric data in an unsupervised manner. Furthermore, the autoencoder can be trained to detect particular data without being pre-programmed with specific rules by training the neural network to ignore signal interference, signal artifacts, and signal noise by considering a clean data set.

The monitoring and processing device 202 can continuously or periodically monitor, store, process, and transmit any number of different patient biometrics (e.g., acquired biometric data) via the network 210. As described herein, examples of patient biometrics include electrical signals (e.g., ECG signals and brain biometrics), blood pressure data, blood glucose data, and temperature data. Patient biometrics can be monitored and transmitted for treatment across any number of different diseases, such as cardiovascular disease (e.g., arrhythmias, cardiomyopathies, and coronary artery disease) and autoimmune diseases (e.g., type I and type II diabetes).

The patient biosensor 212 may include, for example, one or more transducers configured to convert one or more environmental conditions into electrical signals, resulting in acquisition of different types of biometric data. For example, the patient biosensor 212 may include one or more electrodes configured to acquire electrical signals (e.g., cardiac signals, brain signals, or other bioelectric signals), a temperature sensor (e.g., a thermocouple), a blood pressure sensor, a blood glucose sensor, a blood oxygen sensor, a pH sensor, an accelerometer, and a microphone.

As described in further detail herein, the monitoring and processing device 202 may be an ECG monitor for monitoring ECG signals of a heart (e.g., heart 120 in FIG. 1). In this regard, the patient biosensor 212 of the ECG monitor may include one or more electrodes (e.g., electrodes of catheter 105 in FIG. 1) for acquiring the ECG signals. The ECG signals may be used to treat various cardiovascular disorders.

In another example, the monitoring and processing device 202 may be a continuous glucose monitor (CGM) for continuously monitoring a patient's blood glucose levels on an ongoing basis to treat various diseases, such as type I and type II diabetes. In this regard, the patient biosensor 212 of the CGM may include subcutaneously disposed electrodes (e.g., electrodes of catheter 105 in FIG. 1 ), which may monitor blood glucose levels from the patient's interstitial fluid. The CGM may be a component of a closed-loop system in which blood glucose data is sent to an insulin pump, for example, for calculated delivery of insulin without user intervention.

The processor 214 may be configured to receive, process, and manage biometric data acquired by the patient biometric sensors 212 and communicate the biometric data to the memory 218 for storage and/or across the network 210 via the transceiver 222. As described in more detail herein, data from one or more other monitoring and processing devices 202 may also be received by the processor 214 through the transceiver 222. Also, as described in more detail herein, the processor 214 may be configured to selectively respond to different tapping patterns (e.g., single tap or double tap) received from the UI sensor 216 (e.g., an internal capacitive sensor), such that different tasks of the patch (e.g., data acquisition, storage, or transmission) may be initiated based on the detected pattern. In some embodiments, the processor 214 may generate audible feedback regarding detecting a gesture.

The UI sensor 216 may include, for example, a piezoelectric or capacitive sensor configured to receive user input, such as tapping or touching. For example, the UI sensor 216 may be controlled to perform capacitive coupling in response to the patient 204 tapping or touching the surface of the monitoring and processing device 202. Gesture recognition may be implemented via any one of a variety of capacitive types, such as resistive capacitive, surface capacitive, projected capacitive, surface acoustic wave, piezoelectric, and infrared touch. The capacitive sensor may be disposed over a small area of the surface or across its length, such that tapping or touching the surface activates the monitoring device.

Memory 218 may be any non-transitory tangible medium, such as magnetic, optical, or electronic memory (e.g., any suitable volatile and/or non-volatile memory, such as random access memory or a hard disk drive). According to one or more embodiments, memory 218 may store processor-executable code, software, or instructions for the training algorithm and autoencoder.

The transceiver 222 may include a separate transmitter and a separate receiver. Alternatively, the transceiver 222 may include a transmitter and receiver integrated into a single device.

According to certain embodiments, the monitoring and processing device 202 may be a device internal to the body of the patient 204 (e.g., implantable subcutaneously). The monitoring and processing device 202 may be inserted into the patient 204 via any applicable method, including oral infusion, surgical insertion via a vein or artery, an endoscopic procedure, or a laparoscopic procedure.

According to some embodiments, the monitoring and processing device 202 may be a device external to the patient 204. For example, as described in further detail herein, the monitoring and processing device 202 may include an attachable patch (e.g., attached to the patient's skin). The monitoring and processing device 202 may also include a catheter with one or more electrodes, a probe, a blood pressure cuff, a weight scale, a bracelet or smartwatch vital sign tracker, a glucose monitor, a continuous positive airway pressure (CPAP) machine, or virtually any device capable of providing input regarding the patient's health or vital signs.

In some embodiments, the monitoring and processing device 202 may include both components that are internal to the patient and components that are external to the patient.

Although a single monitoring and processing device 202 is shown in FIG. 2, an exemplary system may include multiple patient vital signs monitoring and processing devices. For example, the monitoring and processing device 202 may be in communication with one or more other patient vital signs monitoring and processing devices. Additionally or alternatively, one or more other patient vital signs monitoring and processing devices may be in communication with the network 210 and other components of the system 200.

The local computing device 206 and/or the remote computing system 208, together with the monitoring and processing device 202, may be any combination of software and/or hardware that separately or collectively stores, executes, and implements the autoencoder and its functionality. Furthermore, as described herein, the local computing device 206 and/or the remote computing system 208, together with the monitoring and processing device 202, may be an electronic computer framework that encompasses and/or uses any number and combination of computing devices and networks utilizing various communication technologies. The local computing device 206 and/or the remote computing system 208, together with the monitoring and processing device 202, may be easily scalable, extensible, and modular, with the ability to change to different services or reconfigure some features independently of other features.

According to one embodiment, the local computing device 206 and the remote computing system 208, along with the monitoring and processing unit 202, may include at least a processor and memory, where the processor executes computer instructions related to the autoencoder and the memory stores these computer instructions executed by the processor.

The local computing device 206 of the system 200 can be configured to communicate with the monitoring and processing unit 202 and to act as a gateway to the remote computing system 208 through the second network 211. The local computing device 206 may be, for example, a smartphone, smartwatch, tablet, or other portable smart device configured to communicate with other devices via the network 211. Alternatively, the local computing device 206 may be a fixed or stand-alone device, such as, for example, a fixed base station including modem and/or router functionality, a desktop or laptop computer using an executable program to communicate information between the processing unit 202 and the remote computing system 208 via a wireless module in the PC, or a USB dongle. Biometric data may be communicated between local computing device 206 and monitoring and processing device 202 using short-range wireless technology standards (e.g., Bluetooth, Wi-Fi, ZigBee, Z-wave, and other short-range wireless standards) over a short-range wireless network 210, such as a local area network (LAN) (e.g., a personal area network (PAN)). In some embodiments, local computing device 206 may also be configured to display acquired patient electrical signals and information associated with the acquired patient electrical signals, as described in further detail herein.

In some embodiments, the remote computing system 208 may be configured to receive at least one of the monitored patient biometric information and information associated with the monitored patient via network 211, which is a long-range network. For example, if the local computing device 206 is a mobile phone, network 211 may be a wireless cellular network, and information may be communicated between the local computing device 206 and the remote computing system 208 via a wireless technology standard, such as any of the wireless technologies described above. As described in further detail herein, the remote computing system 208 may be configured to provide (e.g., visually display and/or audibly provide) at least one of the patient biometric information and associated information to a medical professional, physician, healthcare professional, etc.

In FIG. 2, network 210 is an example of a short-range network (e.g., a local area network (LAN) or a personal area network (PAN)). Information can be transmitted between monitoring and processing unit 202 and local computing device 206 via short-range network 210 using any one of a variety of short-range wireless communication protocols, such as Bluetooth, Wi-Fi, ZigBee, Z-wave, near field communication (NFC), ultra-wideband, ZigBee, or infrared (IR).

Network 211 may be a wired network, a wireless network, or may include one or more wired and wireless networks, such as an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct or serial connection, a cellular network, or any other network or medium capable of facilitating communication between local computing device 206 and remote computing system 208. Information may be transmitted over network 211 using any one of a variety of long-range wireless communication protocols (e.g., TCP/IP, HTTP, 3G, 4G/LTE, or 5G/New Radio). Wired connections may be implemented using Ethernet, Universal Serial Bus (USB), RJ-11, or any other wired connection commonly known in the art. Wireless connectivity may be implemented using Wi-Fi, WiMAX, Bluetooth, infrared, cellular networks, satellite, or any other wireless connectivity technique. Additionally, several networks may function independently or communicate with each other to facilitate communication within network 211. In some cases, remote computing system 208 may be implemented as a physical server on network 211. In other cases, remote computing system 208 may be implemented as a virtual server on a public cloud computing provider of network 211 (e.g., Amazon Web Services (AWS)).

Figure 3 illustrates an artificial intelligence system 300 according to one or more embodiments. The artificial intelligence system 300 may include data 310, a machine 320, a model 330, multiple outcomes 340, and underlying hardware 350. Figure 4 illustrates a block diagram of a method 400 implemented in the artificial intelligence system of Figure 3. For ease of understanding, Figures 3-4 will be described with reference to Figure 2.

Generally, the artificial intelligence system 300 operates the method 400 by using the data 310 to train a machine 320 (e.g., the local computing device 206 of FIG. 2 ) while building a model 330 to enable multiple (predicted) outcomes 340. In such a configuration, the artificial intelligence system 300 can operate against hardware 350 (e.g., the monitoring and processing unit 202 of FIG. 2 ) to train the machine 320, build the model 330, and predict outcomes using algorithms. These algorithms can be used to solve the trained model 330 and predict outcomes 340 associated with the hardware 350. These algorithms can generally be categorized as classification algorithms, regression algorithms, and clustering algorithms.

At block 410, the method 400 may include collecting data 310 from the hardware 350. The machine 320 may act as a controller or data collector associated with the hardware 350 and/or may be associated with the hardware 350. The data 310 (e.g., biometric data that may originate from the monitoring and processing unit 202 of FIG. 2) may be associated with the hardware 350. For example, the data 310 may be ongoing data or output data associated with the hardware 350. The data 310 may also include currently collected data, historical data, or other data from the hardware 350. For example, the data 310 may include measurements taken during a surgical procedure and may be associated with the outcome of the surgical procedure. For example, cardiac temperature (e.g., of the patient 204) may be collected and correlated with the outcome of the cardiac procedure.

At block 420, the method 400 may include training the machine 320 on the hardware 350. This training may include analyzing and correlating the data 310 collected in block 410. For example, in the cardiac case, temperature and outcome data 310 may be trained to determine whether a correlation or association exists between the temperature of the heart (e.g., of the patient 204) during a cardiac procedure and the outcome.

At block 430, the method 400 may include building a model 330 on the data 310 associated with the hardware 350. Building the model 330 may include modeling of physical hardware or software, modeling of an algorithm, and/or the like. The modeling may be intended to represent the collected and trained data 310. According to some embodiments, the model 330 may be configured to model the operation of the hardware 350 and model the data 310 collected from the hardware 350 to predict outcomes achieved by the hardware 350. According to one or more embodiments, the model 330, with respect to an autoencoder, may separate the ventricular far field from the activation of the atrial system and generate separate maps for atrial and ventricular activation.

At block 440, the method 400 may include predicting a plurality of outcomes 340 of the model 330 associated with the hardware 350. This prediction of the plurality of outcomes 340 may be based on the trained model 330. For example, to enhance understanding of the present disclosure, for a heart, if a temperature between 36.5°C and 37.89°C (i.e., 97.7°F and 100.2°F) during the procedure results in a more positive outcome from the cardiac procedure, then that outcome may be predicted for a given procedure based on the temperature of the heart during the cardiac procedure. Thus, using the predicted outcome 340, the hardware 350 may be configured to provide a particular desired outcome 340 from the hardware 350.

Referring now to FIG. 5, an example of a neural network 500 is shown, according to one or more embodiments. The neural network 500 operates as an implementation of an autoencoder. The neural network 500 can be implemented in hardware, such as the machine 320 (e.g., the local computing device 206 of FIG. 2) and/or the hardware 350 (e.g., the monitoring and processing unit 202 of FIG. 2). In general, a neural network is a network or circuit of neurons, or in the modern sense, an artificial neural network (ANN), composed of artificial neurons, nodes, or cells.

For example, an ANN can include a network of processing elements (artificial neurons) that can exhibit complex global behavior determined by the connections between the processing elements and element parameters. These connections in a neuronal network or circuit can be modeled as weights. Positive weights can reflect excitatory connections, while negative values can represent inhibitory connections. Inputs can be modified by the weights and summed using a linear combination. An activation function can control the amplitude of the output. For example, the allowable range of the output is typically 0 to 1, or sometimes -1 to 1.

In most cases, ANNs are adaptive systems that change their structure based on external or internal information flowing through the network. In more practical terms, neural networks are nonlinear statistical data modeling or decision-making tools that can be used to model complex relationships between inputs and outputs or to find patterns in data. ANNs can therefore be used for predictive modeling and adaptive control applications while being trained through data sets. Self-learning arising from experience can occur in ANNs, allowing them to draw conclusions from complex, seemingly unrelated sets of information. The usefulness of artificial neural network models lies in the fact that they can be used to estimate and also use functions from observations. Unsupervised neural networks can be used to learn representations of inputs that capture salient features of the input distribution, and more recently, in deep learning algorithms that can implicitly learn distribution functions for observed data. Training with neural networks is particularly useful in applications where the complexity of the data or task prohibits the manual design of such functions.

Neural networks can be used in a variety of fields. The tasks to which ANNs are applied tend to span broad categories, including function approximation or regression analysis, including time series prediction and modeling; pattern and sequence recognition; classification, including novelty detection and sequential decision making; and data processing, including filtering, clustering, blind signal separation, and compression.

Application areas of ANNs may include identification and control of nonlinear systems (vehicle control, process control), game playing and decision-making (backgammon, chess, racing), pattern recognition (radar systems, face identification, object recognition), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, KDD), visualization, and email spam filtering. For example, it is possible to create semantic profiles of user interests resulting from photographs trained for object recognition.

Referring now to FIG. 6, a block diagram of a method 600 according to one or more embodiments is shown. Method 600 illustrates the operation of a neural network 500 (e.g., an autoencoder). Referring to FIG. 5, in neural network 500, input layer 510 may be represented by multiple inputs, such as 512 and 514. With reference to block 610 of FIG. 6, input layer 510 may receive multiple inputs (e.g., input intracardiac signals) as an initial operation. The multiple inputs may be ultrasound signals, radio signals, acoustic signals, or two-dimensional images. More specifically, the multiple inputs may be represented as input data (X), which is raw data recorded from the atria. The desired information may be within the high-frequency zone of the heart (e.g., the atria), and the autoencoder provides a better structure of the input intracardiac signals. According to one or more embodiments, the multiple inputs may be a combination of intracardiac ECG and body surface ECG (to remove far-field noise from the intracardiac signals).

At block 620 of FIG. 6 , the neural network 500 may utilize the intracardiac data set to encode the input intracardiac signal and generate a latent representation. The latent representation may include one or more intermediate images derived from the input intracardiac signal. According to one or more embodiments, the latent representation may be generated by an element-wise activation function (e.g., a sigmoid function or a rectified linear unit) of an autoencoder that applies a weighting matrix to the input intracardiac signal and appends a bias vector to the result. The weights and biases of the weighting matrix, as well as the bias vector, may be randomly initialized and then iteratively updated during training.

The intracardiac data set may be a training data set or clean data containing predetermined and approved signals that are free of interference, artifacts, and noise (i.e., an example of clean data). In one embodiment, a trained medical professional, physician, or the like may review, edit, and remove signal interference, artifacts, and noise to approve each electrical signal in the intracardiac data set. In one embodiment, the intracardiac data may contain electrical signals on the order of thousands or more, and the signal morphology of each electrical signal may be examined using template matching and blanking. For example, an intracardiac data set (e.g., a database of "clean versions" of intracardiac ECG signals) may be used to perform denoising of any IC-ECG artifacts. Given the volume of electrical signals and the complexity of reviewing, editing, and approving, the creation of the intracardiac data set may be considered the data training portion of the multi-stage data manipulation by the autoencoder.

As shown in FIG. 5 , inputs 512 and 514 may be provided (e.g., latent representation or data encoding) to a hidden layer 530, illustrated as including nodes 532, 534, 536, and 538. This encoding provides a dimensionality reduction of the input intracardiac signal. Dimensionality reduction is the process of reducing the number of random variables (of multiple inputs) under consideration by obtaining a set of key variables. For example, dimensionality reduction may be feature extraction, which transforms data (e.g., multiple inputs) from a high-dimensional space (e.g., greater than 10 dimensions) to a low-dimensional space (e.g., 2-3 dimensions). Technical effects and advantages of dimensionality reduction include reducing the time and storage space required for the data, improving data visualization, and improving parameter interpretation for machine learning. This data transformation may be linear or nonlinear. The receiving (block 610) and encoding (block 620) operations may be considered the data preparation portion of the multi-stage data manipulation performed by an autoencoder.

According to one embodiment, data preparation may further include intracardiac electrocardiogram (IC-ECG) data collection of the atria (the upper chambers where blood enters the ventricles of the heart) with simultaneous recording from the ventricles (the two lower chambers of the heart).

In block 630 of FIG. 6, the neural network 500 can decode the latent representation to generate an output intracardiac signal. The output intracardiac signal can be an estimate of the ventricular far-field in the case of IC-ECG. As shown in FIG. 5, nodes 532, 534, 536, and 538 can be combined to generate output 552 in output layer 550, which can reconstruct inputs 512 and 514 in reduced dimensionality without signal interference, signal artifacts, and signal noise. The neural network 500 performs processing through the hidden layer 530 of nodes 532, 534, 536, and 538 and can exhibit complex global behavior determined by connections between processing elements and element parameters. The target data in the output layer 550 can include ventricular activity (Y1) of target data type 1 and input data (Y2) of target data type 2 after far-field reduction. The far field can cause problems with generating and navigating 3D maps (e.g., the ventricular far field can interfere with atrial activation). Therefore, technical effects and advantages of an autoencoder using neural network 500 include improving the accuracy of 3D maps through artifact (far field) removal.

According to one or more embodiments, an autoencoder model using neural network 500 can separate the ventricular far field from activation of the atrial system, generating separate maps for atrial and ventricular activation.

According to one embodiment, the autoencoder may be a denoising autoencoder to find a mapping function (f, g) such that f(X) = Y1 and g(X) = Y2, as described further below. In this regard, the task of the autoencoder may be to learn a mapping from X to X through some dimensionality reduction of the input X (e.g., construct two neural networks (F, G) such that h = F(X) and X = G(h), where the dimensionality of h is lower than the dimensionality of X). In a denoising autoencoder, the architecture is similar, but the denoising autoencoder learns a mapping from X to Y, where Y is a denoised version of X.

Referring to FIG. 7 , a graphical representation of a signal 700 is shown in accordance with one or more embodiments. As shown by signal 700, the ECG signal includes a P wave 710 (due to atrial depolarization), a QRS complex 720 (due to atrial repolarization and ventricular depolarization), and a T wave 730 (due to ventricular repolarization). The ECG signal is generated by the contraction (depolarization) and relaxation (repolarization) of the atrial and ventricular muscles of the heart. To record the ECG signal, electrodes may be placed at specific locations on the body or may be placed within the body via a catheter. Artifacts (e.g., noise) are unwanted signals that combine with electrical signals, such as ECG signals, potentially creating obstacles to the diagnosis and/or treatment of cardiac disease. Artifacts in electrical signals may be baseline drift, power line interference, EMG noise, power line noise, etc. That is, examples of artifacts include, but are not limited to, power noise (e.g., electrostatic and electromagnetic coupling between circuits and 50 or 60 Hz power lines), fluoro noise (e.g., fluorescent lights), contact noise (e.g., collisions between catheter electrodes), and deflection noise (e.g., static discharge during catheter deflection).

Baseline drift or baseline shift occurs when the base axis (x-axis) of a signal appears to "fluctuate" or move up and down rather than being linear. This can cause the entire signal to shift from its normal base. In ECG signals, baseline drift can be caused by improper electrode contact (e.g., electrode-skin impedance), patient movement, and periodic movement (e.g., breathing).

8 illustrates a graphical representation of a signal 810 shown in plot 800, according to one or more embodiments. In this regard, signal 800 is a typical ECG signal affected by baseline drift 820. The frequency content of the baseline drift 1020 is in the 0.5 Hz range. As body movement increases during exercise or stress testing, the frequency content of the baseline drift increases. According to an implementation, considering that the baseline signal is a low-frequency signal, a finite impulse response (FIR) filter with high-pass zero-phase forward-backward filtering with a cutoff frequency of 0.5 Hz may be used to estimate and remove the baseline drift 820 in the ECG signal 810.

Electromagnetic fields generated by power lines represent a common noise source in electrical signals such as ECGs, as well as any other bioelectrical signals recorded from a patient's body. Such noise is characterized, for example, by 50 or 60 Hz sinusoidal interference, sometimes accompanied by many harmonics. Such narrow-band noise makes ECG analysis and interpretation more difficult because it can result in unreliable depiction of low-amplitude waveforms and spurious waveforms. It may be necessary to remove power line interference from ECG signals because it superimposes on low-frequency ECG waves such as the P wave 710 and T wave 730.

The presence of muscle noise can interfere with many electrical signal fields, such as the ECG field, and thus may obscure low-amplitude waveforms. In contrast to baseline drift 820 and 50/60 Hz interference, muscle noise is not removed by narrowband filtering, but presents a different filtering challenge because the spectral components of muscle activity overlap significantly with the spectral components of the PQRST complex 720. Because the ECG signal 810 is a repetitive signal, techniques can be used to reduce muscle noise in a manner similar to the processing of evoked potentials. FIG. 9 illustrates a graphical representation 900 of a signal 905 shown in accordance with one or more embodiments. In this regard, the signal 905 is an ECG signal interfered with by EMG noise 910.

Instruments for measuring electrical signals, such as ECG signals, often detect electrical interference corresponding to line, or power, frequencies. Line frequencies in most countries are nominally set at 50 Hz or 60 Hz, but can vary by a few percent from these nominal values.

Various techniques can be implemented to remove electrical interference from electrical signals. Some of these techniques use one or more low-pass or notch filters. For example, a system can be implemented for variably filtering noise in ECG signals. This system can have multiple low-pass filters, including, for example, one filter with a 3 dB point at approximately 50 Hz and a second low-pass filter with a 3 dB point at approximately 5 Hz.

According to another example, a system for rejecting line frequency components of an electrical signal may be implemented by passing the signal through two series-coupled notch filters. A system may be implemented with notch filters that can have either low-pass or high-pass coefficients or both to remove line frequency components from the ECG signal. This system may also support the removal of burst noise and may calculate the heart rate from the notch filter output.

According to another example, a system may be implemented that includes several units for removing interference. These units may include an average value unit for generating an average signal over several cardiac cycles, a subtraction unit for subtracting the average signal from the input signal to generate a residual signal, a filter unit for providing a filtered signal from the residual signal, and/or an addition unit for adding the filtered signal to the average signal.

According to another example, an analog-to-digital (A/D) converter can provide noise rejection by synchronizing the converter's clock with a phase-locked loop set to the line frequency.

Additionally, biometric (e.g., biopotential) patient monitors can use surface electrodes to make biopotential measurements, such as an ECG or electroencephalogram (EEG). The fidelity of these measurements is limited by the effectiveness of the electrode connection to the patient. The resistance of the electrode system to the flow of current is known as electrical impedance and characterizes the effectiveness of the connection. Typically, the higher the impedance, the lower the fidelity of the measurement. Several mechanisms can contribute to lower fidelity.

Signals from electrodes with high impedance are affected by thermal noise (or so-called Johnson noise), a voltage that increases with the square root of the impedance value. Furthermore, biopotential electrodes tend to have voltage noise that exceeds that predicted by Johnson. Amplifier systems making measurements from biopotential electrodes may also have degraded performance at higher electrode impedances. These degradations are characterized by poor common-mode rejection, which tends to increase contamination of bioelectric signals by noise sources such as patient movement and electronic equipment that may be used on or around the patient. These noise sources are particularly prevalent in operating rooms and can include equipment such as electrosurgical units (ESUs), cardiopulmonary bypass pumps (CPBs), electric motor-driven surgical saws, lasers, and other sources.

During cardiac surgery, it is often desirable to continuously measure electrode impedance in real time while monitoring the patient. To do this, a very small current is typically injected into the electrode and the resulting voltage is measured, thereby establishing the impedance using Ohm's Law. This current can be injected using a DC or AC power source. Often, it is impossible to separate the voltage due to electrode impedance from voltage artifacts resulting from interference. Interference tends to increase the measured voltage, and therefore the apparent measured impedance can cause the bioelectronic measurement system to erroneously detect a higher impedance than actually presented. Such monitoring systems often have maximum impedance threshold limits that can be programmed to prevent operation when they detect an impedance above these limits. This is especially true for systems that make very small voltage measurements, such as EEG. Such systems require very low electrode impedances.

The use of high-resolution intracardiac electrograms (EGMs) can guide cardiac ablation procedures. Cardiac ablation may be used, among other things, to treat ventricular tachycardia (VT), a condition in which complex electrophysiological (EP) circuits and reentry into one of the ventricles result in a rapid, irregular heartbeat. Therefore, the goal of catheter ablation is to target the origin of VT. Mapping VT circuits and identifying their source is critical for successful VT ablation. A major challenge in interpreting intracardiac EGMs in the presence of VT is that EGM signals can have complex morphologies, making it difficult to extract local activation times (LATs) with sufficiently high spatial resolution. This, in turn, makes it difficult to accurately map the complex circuits within the ventricles, which is critical for identifying the location of the associated ablation target.

Because unipolar EGM signals typically have a much lower signal-to-noise ratio compared to bipolar signals, bipolar signals are currently the primary tool for extracting LAT. However, unipolar signals potentially offer better spatial and temporal resolution, significantly improving mapping of the VT circuit. Therefore, advanced digital signal processing (DSP) methods can be applied to extract accurate LAT from noisy unipolar signals. Advanced digital signal processing methods or systems may include various digital filters aimed at reducing or attenuating noise from the signal. Linear smoothing filters, such as low-pass or high-pass filters, or any other smoothing operator that can be convolved with the signal, can be used to reduce or attenuate noise. Nonlinear filters, such as median filters that remove noise, may also be used to reduce or attenuate noise. Wavelet transforms, which can achieve both noise reduction and feature preservation, may also be used. Statistical denoising methods may also be used, which can use environmental or neighboring signals or any other patterns to reduce unwanted components in the signal.

A bipolar signal is from two adjacent unipolar electrodes. A unipolar signal is from a unipolar electrode and a reference electrode and is a combination of far-field and near-field contributions. The primary noise source in a unipolar signal is the far-field signal, which arises from voltage depolarization of distant tissues. Due to the large distance between the unipolar electrode and the reference electrode, the far-field signal is often not fully accounted for and therefore not fully removed from the signal compared to a bipolar signal. Because the two unipolar signals that form a unipolar pair have very similar far-fields, the difference between them is nearly zero except when there is local activity in each of the unipolar signals. This local activity is called the near-field signal and can appear as a small spike on the bipolar signal.

When a unipolar electrode is positioned under scar tissue that does not produce electrical activity, the near-field may have a lower amplitude than the far-field, compared to situations arising from healthy tissue. This makes it particularly difficult to apply classical DSP methods to separate the far-field signal from the near-field signal. In this case, the near-field may be very low, even negligible. Therefore, the bipolar signal may have no activation at all, or may be virtually zero. This type of unipolar signal can be represented as a pure far-field signal, since no local activation events are evident. In this case, the bipolar signal may appear flat, and the two unipolar signals may be nearly identical. These types of signals can be obtained by placing a catheter in a location without contacting the cardiac muscle, or as a body surface ECG signal, which is essentially a far-field signal. These types of unipolar signals can serve as a training data set, allowing a neural network to learn this type of activity and distinguish between the far-field component and the mixed unipolar signal.

In the present context, the far-field contribution is considered noise to be removed, but in other contexts, the far-field contribution itself may contain useful information. This can provide further motivation for separating the two contributions.

Deep learning (DL), based on deep neural networks (DNN), has emerged as a disruptive technology in the application of computer algorithms to various fields, such as computer vision and DSP. DL often enables the extraction of complex patterns and data from signals and images in cases where such extraction was previously impossible or possible only through time-consuming manual analysis. DL is therefore particularly attractive for intracardiac EGM applications, where reducing procedure time and increasing clinical success rates are important goals.

Machine learning (ML) is a group of algorithms and statistical models for data parsing used to perform specific tasks. DL is a subset of machine learning algorithms that set model parameters during the training process to enable accurate prediction of desired outputs for unknown data. ML and DL techniques enable the analysis of highly complex spatiotemporal information that classical algorithms struggle to analyze. While machine learning is typically based on feature extraction using a list of heuristics on the data, DL is based on learning from examples and typically does not require feature extraction from the data. The main difference between DL and traditional ML is the need for large amounts of data for the training process. Assuming there is a sufficient amount of data, the performance of DL-based algorithms is typically superior to traditional ML algorithms.

DL is therefore a useful tool for decomposing the near-field and far-field components in ECG signals, specifically VT signals. This allows activation detection to be limited to near-field activity. This is useful because while mapping ventricular activity, the far-field is strong and can mask near-field activity, thus misleading the annotation mechanism. It is also useful in the case of atrial fibrillation (AFIB), where the ventricular signal is strong and can be misannotated as atrial activity.

It would therefore be desirable for DL techniques to reduce overall clinical procedure time by providing medical personnel (e.g., cardiologists and electrophysiologists) with insights currently available only through manual data analysis by trained clinicians, and to identify deep data patterns that cannot currently be identified manually or using classical algorithms, thus enabling the identification of ablation targets in more complex cases that are currently untreatable.

DL training can be unsupervised. That is, while a large amount of pre-recorded unipolar EGM signals exists and additional signals can be acquired as needed, a major challenge in applying DL to remove far-field noise is the lack of ground truth data for training the DL model. The decomposition of far-field and near-field signals is an estimate and cannot necessarily be compared to the actual far-field and near-field signals at specific electrodes. Therefore, the DL approach can be unsupervised rather than supervised. Body surface ECG can be used with distal electrodes as ground truth for the far-field components.

FIG. 10 shows a graphical representation of a signal progression 1000 (10A, 10B, 10C, 10D, 10E, and 10F) for far-field cancellation in accordance with one or more embodiments. The signals in FIGS. 10A-10E are intracardiac (IC) ECG signals recorded from different locations along the coronary sinus (CS). In FIG. 10A, signal 1021 represents the body surface IC ECG signal. Boundary 1032 represents the local activation time (LAT). Boundary 1032 also exists in FIGS. 10C, 10D, 10E, and 10F. Boundary 1043 represents the location of the QRS. Boundary 1043 also exists in FIGS. 10A, 10C, 10D, 10E, and 10F. In FIG. 10, the X-axis represents time and the Y-axis represents mV.

As shown in FIG. 10, 1054 represents the far-field component of the IC ECG signal. FIGS. 10C, 10D, 10E, and 10F show the progression of signal 1054 with increasing amounts of far-field cancellation, and signal 1065 represents the IC ECG signal after far-field cancellation. Far-field cancellation may be achieved, for example, by creating blanking periods, e.g., IC ECG signal 1065 may be zero during the far-field periods.

FIG. 11 illustrates a block diagram of a method 1100 according to one or more embodiments. According to one embodiment, the method 1100 may be implemented by a denoising autoencoder. Any combination of software and/or hardware (e.g., the local computing device 206 and the remote computing system 208 in cooperation with the monitoring and processing unit 202) can store, execute, and implement the denoising autoencoder and its functionality, separately or collectively. The denoising autoencoder can train the autoencoder to reconstruct the input from a degraded version of itself, force the hidden layer (e.g., hidden layer 530 of FIG. 5) to discover more robust features (i.e., useful features that will constitute a better, higher-level representation of the input), and prevent the input from learning a particular identity (i.e., always reverting to the same value). In this regard, the denoising autoencoder can encode the input (e.g., to preserve information about the input) and reverse the effects of a corrupting process that was probabilistically applied to the autoencoder's input.

According to one or more embodiments, the denoising autoencoder may implement a long-term memory neural network architecture, a convolutional neural network architecture, or the like. The architecture of the denoising autoencoder may be configurable with respect to the number of layers, the number of connections (e.g., encoder/decoder connections), the regularization technique (e.g., dropout or BN), and the optimization features.

Long-short-term memory neural network architectures may include feedback connections, allowing them to process a single data point (such as an image) in concert with an entire sequence of data (such as audio or video). A unit of a long-short-term memory neural network architecture may consist of a cell, an input gate, an output gate, and a forget gate, where the cell remembers a value over any time interval, and the gate regulates the flow of information into and out of the cell.

The convolutional neural network architecture may be a shared-weight architecture with translation invariance, where each neuron in one layer is connected to every neuron in the next layer. Regularization methods in convolutional neural network architectures can exploit hierarchical patterns in the data and assemble more complex patterns using smaller and simpler patterns. If the denoising autoencoder implements a convolutional neural network architecture, other configurable aspects of the architecture may include the number of filters at each stage, the kernel size, and the number of kernels per layer.

Method 1100 begins at block 1105, where a denoising autoencoder can receive a "cleaned and approved" intracardiac data set from a plurality of electrical signals. As described herein, a trained medical professional, physician, or the like can review and edit the data set to remove signal interference, signal artifacts, and signal noise and approve each electrical signal in the intracardiac data set. At block 1110, the denoising autoencoder can build a model (e.g., model 330 of FIG. 3) from the cleaned and approved intracardiac data set.

In block 1115, the denoising autoencoder can receive input intracardiac signals that include at least far-field artifacts. The input intracardiac signals may be recorded by one or more monitoring and processing devices (e.g., a penta-ray catheter with 20 electrodes, a basket catheter with 64 electrodes, multiple body surface leads, etc.). The far-field may cause problems with generating and navigating 3D maps (i.e., the ventricular far-field may interfere with atrial activation).

In block 1120, the denoising autoencoder can encode the input intracardiac signal using the model (from block 1110). This encoding provides a dimensionality reduction of the input intracardiac signal according to at least how the model for removing far-field artifacts dictates that reduction. The result of the encoding is the generation of a latent representation. In block 1130, the denoising autoencoder can decode the latent representation to generate the output intracardiac signal.

In block 1135, the denoising autoencoder can map the output intracardiac signal. For example, the denoising autoencoder can use its underlying architecture to find a mapping function (f, g) such that f(X) = Y1 and g(X) = Y2.

At block 1140, an ECG may be generated from the mapped output intracardiac signal. This ECG can be generated by the computing device running the denoising autoencoder or by another device. This ECG can be improved to remove signal interference, signal noise, and signal artifacts, and then displayed to a medical professional. The improved ECG can dramatically reduce the time spent with the cardiac patient.

As shown herein, during intracardiac electrogram mapping, a mapping catheter can record both atrial and ventricular activation. In some cases, the ventricular far-field may interfere with atrial activation (e.g., signal interference), which may affect the clinical understanding and interpretation of the Carto map. According to one or more embodiments, technical effects and advantages of a denoising autoencoder may include separating the ventricular far-field from activation of the atrial system and generating separate maps for atrial and ventricular activation (e.g., the denoising autoencoder uses a model during decoding to distinguish between the ventricular far-field and activation of the atrial system in one or more output intracardiac signals).

FIG. 12 is an exemplary flow diagram of a method 1200 for decomposing near-field and far-field signals, according to one embodiment. During the training phase, far-field ventricular measurements can be acquired (1210). These may be unipolar signals. Measurements may be acquired using a multiple-electrode catheter and/or a body surface ECG. There may be numerous far-field measurements. The far-field measurements may be pure far-field signals. In one embodiment, the pure far-field signals may be from recordings where the bipolar signal is zero or near zero. Thus, the near-field of the unipolar signal may be absent or very small. In one embodiment, simulation may be used for pure far-field measurements, for example, by using specialized simulation software capable of generating pure far-field signals. This may be done by controlling the source generating the ECG signal and using only far-field sources. In one embodiment, an expert can determine the extent of the pure far-field. In one embodiment, the body surface ECG may primarily contain the far-field. In one embodiment, measurements from areas of scar tissue where the near field is ignored due to their inability to contain local activity may be used for pure far-field measurements.

A synthetic local field signal can be added (1220). The synthetic local field signal can be obtained, for example, from a simulation of an ECG signal. A number of unipolar signals can be introduced in a training phase so that the algorithm can learn to recognize unipolar signal forms. In addition, the algorithm can be exposed to far-field signals so that it can learn to detect the far-field signals.

The algorithm may be configured to evaluate or learn both pure far-field signals and combinations or mixtures of far-field and near-field signals (real or synthetic mixture signals). The algorithm can detect or predict far-field components from the mixture signal that are shared by all electrodes (far-field). While the near-field is electrode-specific because it has local activity that affects only small regions of tissue, the far-field has a much larger contribution (both in signal magnitude and distribution within tissue) and is therefore common to many electrodes, hence the term shared portion. The electrode-shared portion may be a pure near-field signal, or it may be made into a pure near-field signal by subtracting the far-field component from the original signal.

Data may be routinely collected from multiple patients during EP procedures (1240) and provided to the system. The data may include conventional unipolar signals, which are a combination of far-field and near-field signals. These signals may be collected during any VT procedure using a multiple-electrode catheter. Another approach may be to use a composite signal that combines far-field and near-field components. For example, simulated or composite data may be used to generate a pure far-field signal that can serve as an absolute reference. These signals may be generated using specialized simulation software or any other simulation program capable of controlling an ECG signal source. The data may include an intrinsic unipolar signal that includes only the far-field contribution without the near-field component. This unipolar signal may be obtained by placing a catheter at a location without contacting the myocardium. The signals may include ECG values and 3D positions (for each electrode, Signal = V(x, y, z, t)). The data may include surface ECG signals, which are essentially far-field signals. The data may also include manual annotations of underlying near-field signals, particularly unipolar signals, that identify specific features of the LAT. The surface ECG signal data and/or manual annotation data may be used to assist in the training and/or validation of any DL model.

Before processing the unipolar data to extract the near-field contribution in the training phase (1230), a pre-processing filtering step may be performed to remove irrelevant signals and artifacts. The unipolar signals may be manually annotated. The pre-processing filtering may be evaluated by a user (semi-automatically), but at a later stage the annotated data can be used to train a conventional classification convolutional neural network (CNN) to perform the filtering automatically.

The neural network training ends with far-field signal estimation (1230). By knowing the far-field contribution, the near-field signal is the residual signal that remains after removing the far-field signal from the normal unipolar signal. The bipolar signal may then be reconstructed between the two unipolar signal sets.

In the far-field reduction model (1250), what the far-field and near-field signals look like is known from the neural network training (1230), so the autoencoder can automatically decompose the near-field and far-field signals from the measured signals (1240).

Neural networks may be implemented by a number of exemplary approaches, such as autoencoders and Siamese networks, which may be applied to decompose or separate the far-field and near-field contributions (1250).

Autoencoders (AEs) are a class of unsupervised DNNs that learn to reduce the dimensionality of a given dataset, so that they can generate new data that is statistically similar to the original dataset.

In the context of the current method, if a properly selected AE is trained using an EGM signal that contains only far-field contributions without near-field contributions, the AE can be used to extract the far-field contribution from any arbitrary EGM signal. The training phase aims to equalize the input X and output X' (the pure far-field), and the prediction phase maps any arbitrary EGM signal to the far-field signal that contains it. For example, there are several AE types (e.g., variational AE, reconstruction AE, denoising AE, adversarial AE) that contain elements that can be utilized.

A Siamese network containing two identical parts is trained in a fully supervised manner to distinguish between similar and dissimilar paired signatures. Then, when presented with a reference signature and a new signature, the network predicts whether the new signature is authentic (i.e., similar to the reference). Recently, the concept of the Siamese network has been generalized to DNNs and successfully applied to face recognition and face verification. Recently, Siamese NNs have been applied to unsupervised learning of visual representations and medical diagnosis. Siamese networks can be exploited by taking advantage of the fact that two unipolar signals from closely spaced electrodes (forming a bipolar pair) typically have very similar far-field components. Therefore, when two unipolar signals are fed to each part of a Siamese network, a cost function can be constructed that tends to equalize the outputs of the two parts. To avoid obtaining a trivial solution (such as a signal that is exactly zero), a constraint term can be added to the cost function. This is, for example, a term that tends to minimize the difference between the result and the average of the two input signals.

The flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flow diagrams or block diagrams may represent a module, segment, or portion of instructions, which portion of instructions includes one or more executable instructions for implementing particular logical function(s). In some alternative implementations, the functions described in the blocks may occur out of the order described in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending on the functionality involved. It should also be noted that each block of the block diagrams and/or flow diagram illustrations, and combinations of blocks in the block diagrams and/or flow diagram illustrations, may be implemented by a special-purpose hardware-based system that performs specified functions or operations or executes a combination of special-purpose hardware and computer instructions.

While features and elements are described above in particular combinations, those skilled in the art will understand that each feature or element can be used alone or in combination with other features and elements. In addition, the methods described herein may be implemented in a computer program, software, or firmware embodied in a computer-readable medium for execution on a computer or processor. As used herein, computer-readable medium should not be construed as being a transitory signal per se, such as an electric wave or other freely propagating electromagnetic wave, a waveguide or other transmission medium (e.g., optical pulses passing through a fiber optic cable), or an electrical signal transmitted through a wire.

Examples of computer-readable media include electrical signals (transmitted via wired or wireless connections) and computer-readable storage media. Examples of computer-readable storage media include, but are not limited to, registers, cache memory, semiconductor memory devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, optical media such as compact disks (CDs) and digital versatile disks (DVDs), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), and memory sticks. A processor in association with software may be used to implement a radio frequency transceiver for use in a WTRU, UE, terminal, base station, RNC, or any host computer.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly dictates otherwise. It should be further understood that the terms "comprises" and/or "comprising," as used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, and do not exclude the presence or addition of one or more other features, integers, steps, operations, components, and/or groups thereof.

The descriptions of various embodiments herein are provided for illustrative purposes and should not be construed as being exhaustive or limiting to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein has been selected to best explain the principles, practical applications, and technical improvements of the embodiments compared to existing or commercially available technology, or to enable those skilled in the art to understand the embodiments disclosed herein.

[Embodiment]
(1) A method comprising:
receiving one or more input intracardiac signals from a monitoring and processing device, each of the one or more input intracardiac signals including one or more signal artifacts;
encoding, by an autoencoder, the one or more input intracardiac signals using an intracardiac dataset to generate latent representations;
and decoding, by the autoencoder, the latent representation to generate one or more output intracardiac signals comprising the one or more input intracardiac signals reconstructed without the one or more signal artifacts.
2. The method of claim 1, wherein decoding the latent representations to generate one or more output intracardiac signals comprises reconstructing the one or more output intracardiac signals from the latent representations having reduced dimensionality.
3. The method of claim 1, wherein the one or more input intracardiac signals are recorded by a patient biosensor of the monitoring and processing device.
4. The method of claim 1, wherein the intracardiac data set includes a predetermined and accepted signal that is free of the one or more signal artifacts.
5. The method of claim 1, further comprising generating an electrocardiogram from the one or more output intracardiac signals, the electrocardiogram being free of the one or more signal artifacts.

6. The method of claim 1, wherein the autoencoder includes a model that separates ventricular far-field and atrial system activation in the one or more output intracardiac signals during decoding.
(7) A system comprising:
a memory storing processor-executable instructions for the autoencoder;
a processor configured to execute the processor-executable instructions of the autoencoder, the processor-executable instructions providing the system with:
receiving one or more input intracardiac signals from a monitoring and processing device, each of the one or more input intracardiac signals including one or more signal artifacts;
encoding the one or more input intracardiac signals utilizing an intracardiac data set to generate a latent representation;
and a processor that causes the system to decode the latent representation to generate one or more output intracardiac signals that include the one or more input intracardiac signals reconstructed without the one or more signal artifacts.
8. The system of claim 7, wherein decoding the latent representations to generate one or more output intracardiac signals comprises reconstructing the one or more output intracardiac signals from the latent representations comprising reduced dimensionality.
9. The system of claim 7, wherein the one or more input intracardiac signals include biometric data.
10. The system of claim 7, wherein the one or more input intracardiac signals are recorded by a patient biosensor of the monitoring and processing device.

11. The system of claim 7, wherein the intracardiac data set includes a predetermined and accepted signal that is free of the one or more signal artifacts.
12. The system of claim 7, wherein the processor is further configured to execute the processor-executable instructions of the autoencoder, the processor-executable instructions causing the system to generate an electrocardiogram from the one or more output intracardiac signals, the electrocardiogram being free of the one or more signal artifacts.
13. The system of claim 7, wherein the autoencoder comprises a denoising autoencoder.
14. The system of claim 7, wherein the autoencoder includes a model that separates ventricular far-field and atrial system activation in the one or more output intracardiac signals during decoding.
(15) A method for decomposing a near-field signal and a far-field signal, comprising:
receiving a measured signal;
encoding the measured signals by an autoencoder to generate latent representations;
and decoding, by the autoencoder, the latent representation to decompose near-field and far-field components from the measured signals.

16. The method of claim 15, wherein the signal is an electrocardiogram (ECG) signal.
17. The method of claim 15, wherein the measured signal is a unipolar signal.
(18) acquiring far-field ventricular measurements with a training algorithm;
adding a composite local field signal;
16. The method of claim 15, further comprising detecting an obtained far-field signal and a residual near-field signal by the training algorithm.
19. The method of claim 15, wherein the far-field ventricular measurements are obtained using a multiple electrode catheter or a body surface ECG signal.
20. The method of claim 18, wherein decoding the latent representation to decompose the near-field and far-field components is based on the detected derived far-field signal and residual near-field signal.

Claims (12)

1. A method comprising:
receiving one or more input intracardiac signals from a monitoring and processing device, each of the one or more input intracardiac signals including one or more signal artifacts;
encoding, by an autoencoder, the one or more input intracardiac signals using an intracardiac dataset to generate latent representations;
decoding, by the autoencoder, the latent representation to generate one or more output intracardiac signals comprising the one or more input intracardiac signals reconstructed without the one or more signal artifacts;
the autoencoder includes a model that separates, during decoding, activation of the atrial system from ventricular far-field signals in the one or more output intracardiac signals;
the autoencoder further inputs the far-field ventricular measurements supplemented with the synthetic local-field signals into a training algorithm, which separates the ventricular far-field from activation of the atrial system.
method. The method of claim 1, wherein decoding the latent representation to generate one or more output intracardiac signals comprises reconstructing the one or more output intracardiac signals from the latent representation having reduced dimensionality. The method of claim 1, wherein the one or more input intracardiac signals are recorded by a patient biosensor of the monitoring and processing device. The method of claim 1, wherein the intracardiac data set includes a predetermined and accepted signal that is free of the one or more signal artifacts. The method of claim 1, further comprising generating an electrocardiogram from the one or more output intracardiac signals, the electrocardiogram being free of the one or more signal artifacts. 1. A system comprising:
a memory storing processor-executable instructions for the autoencoder;
a processor configured to execute the processor-executable instructions of the autoencoder, the processor-executable instructions providing the system with:
receiving one or more input intracardiac signals from a monitoring and processing device, each of the one or more input intracardiac signals including one or more signal artifacts;
encoding the one or more input intracardiac signals utilizing an intracardiac data set to generate a latent representation;
and decoding the latent representation to generate one or more output intracardiac signals comprising the one or more input intracardiac signals reconstructed without the one or more signal artifacts;
the autoencoder includes a model that separates, during decoding, activation of the atrial system from ventricular far-field signals in the one or more output intracardiac signals;
the autoencoder is configured to input the far-field ventricular measurements supplemented with the composite local-field signals into a training algorithm, the training algorithm being configured to separate the ventricular far-field and activation of the atrial system in the one or more output intracardiac signals.
system. 7. The system of claim 6, wherein decoding the latent representation to generate one or more output intracardiac signals comprises reconstructing the one or more output intracardiac signals from the latent representation having reduced dimensionality. The system of claim 6 , wherein the one or more input intracardiac signals include physiological data. The system of claim 6 , wherein the one or more input intracardiac signals are recorded by a patient biosensor of the monitoring and processing device. The system of claim 6 , wherein the intracardiac data set comprises a predetermined and accepted signal that is free of the one or more signal artifacts. 7. The system of claim 6, wherein the processor is further configured to execute the processor-executable instructions of the autoencoder, the processor-executable instructions causing the system to generate an electrocardiogram from the one or more output intracardiac signals, the electrocardiogram being free of the one or more signal artifacts. The system of claim 6 , wherein the autoencoder comprises a denoising autoencoder.