JP7568412B2 - Radiation equipment and failure prediction equipment - Google Patents

Description

The embodiments disclosed herein relate to a radiation device and a failure prediction device.

Various types of radiation devices, such as radiation therapy devices and radiation diagnostic devices, typically include movable parts that can perform mechanical operations such as movement and rotation. For example, radiation devices include leaves that move to control the radiation irradiation range. Radiation devices also include a top plate on which a subject is placed, which also moves and rotates (tilts).

In such movable devices that include moving parts, wear accumulates with repeated use, and malfunctions may occur even when the device is used appropriately. Furthermore, if a malfunction occurs in the movable device, the radiation device cannot be used until the part is replaced or repaired. In other words, malfunction of the movable device causes the radiation device to go down. If the radiation device goes down, scheduled radiation therapy or image collection cannot be performed, and treatment and diagnosis will be delayed. Furthermore, depending on the timing of the malfunction, patients who visit the hospital may have to return at a later date, which is a burden on the patients.

JP 2015-130908 A JP 2018-192256 A JP 2005-110794 A

One of the problems that the embodiments disclosed in this specification are intended to solve is to avoid a system downtime of the radiation device. However, the problems solved by the embodiments disclosed in this specification are not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be considered as other problems solved by the embodiments disclosed in this specification.

The data sorting device of the embodiment includes a movable device, a control unit, a model storage unit, a prediction unit, and an output unit. The movable device includes a movable part capable of performing mechanical operations, and a drive unit that operates the movable part. The control unit controls the movable device to perform radiation therapy or image collection. The model storage unit stores a trained model that is functionalized to predict a failure related to the movable device. The prediction unit predicts a failure related to the movable device by inputting at least one of first information related to the operation of the movable part and second information indicating the environment when the movable part operates into the trained model. The output unit outputs the failure prediction result by the prediction unit.

FIG. 1 is a diagram illustrating an example of a radiation therapy apparatus according to the first embodiment. FIG. 2 is a block diagram showing an example of the configuration of the radiation therapy apparatus according to the first embodiment. FIG. 3 is a diagram illustrating an example of a radiation constrictor according to the first embodiment. FIG. 4 is a diagram illustrating an example of a process for generating a trained model according to the first embodiment. FIG. 5 is a diagram illustrating an example of use of the trained model according to the first embodiment. FIG. 6 is a flowchart for explaining a series of processing steps of the radiotherapy apparatus according to the first embodiment. FIG. 7 is a block diagram showing an example of the configuration of an X-ray CT apparatus according to the second embodiment. FIG. 8 is a block diagram showing an example of the configuration of a radiation system according to the third embodiment.

Below, we will explain in detail the embodiments of the radiation device and the failure prediction device with reference to the drawings.

(First embodiment)
In the first embodiment, a radiation therapy apparatus 10 capable of performing radiation therapy will be described as an example of a radiation apparatus.

First, an overview of the radiation therapy device 10 will be described using FIG. 1. FIG. 1 is a diagram showing an example of the radiation therapy device 10 according to the first embodiment. For example, as shown in FIG. 1, the radiation therapy device 10 has a bed 16, a radiation generator 171 that generates therapeutic radiation, a radiation aperture 172, and a rotating gantry 19.

For example, the radiation therapy device 10 controls the operation of the bed 16, radiation generator 171, radiation aperture 172, rotating gantry 19, etc., in accordance with the treatment plan to perform radiation therapy. As one example, the radiation therapy device 10 is connected to a radiation therapy planning CT (Computed Tomography) device (not shown), and performs radiation therapy by irradiating radiation from the radiation generator 171 to the area to be treated in accordance with the treatment plan transmitted from the radiation therapy planning CT device.

As shown in FIG. 1, the radiation therapy device 10 further includes a radiation generator 181 that irradiates radiation for imaging and a detector 182 that detects radiation for imaging. For example, before the start of radiation therapy, the radiation therapy device 10 irradiates radiation from the radiation generator 181 to the patient P1 while rotating the rotating gantry 19, and detects the radiation that has passed through the patient P1 with the detector 182. The radiation therapy device 10 also generates a cone beam CT image based on the detection result by the detector 182 and displays it on the display 12. This allows the user of the radiation therapy device 10 to confirm the position of the treatment target site at the start of radiation therapy. The radiation generator 181 and the detector 182 can also be used to confirm the position of the treatment target site during radiation therapy. That is, while the radiation generator 171 is irradiating therapeutic radiation, the radiation generator 181 irradiates radiation to the patient P1 and collects X-ray images, making it possible to confirm the position of the treatment target site during treatment.

Next, the configuration of the radiation therapy device 10 will be described with reference to FIG. 2. FIG. 2 is a block diagram showing an example of the configuration of the radiation therapy device 10 according to the first embodiment. As shown in FIG. 2, the radiation therapy device 10 has an input interface 11, a display 12, a memory circuitry 13, a model memory circuitry 14, a processing circuitry 15, a bed 16, and a radiation generating device 17.

Although not shown in FIG. 2, the radiation therapy device 10 further includes the rotating gantry 19 shown in FIG. 1, and the radiation generation device 17 is held by the rotating gantry 19 so that it can rotate around the patient P1. Similarly, the radiation therapy device 10 further includes a radiation generator 181 that irradiates radiation for imaging, and a detector 182 that detects radiation for imaging.

The input interface 11 accepts various input operations from the user, converts the accepted input operations into electrical signals, and outputs the electrical signals to the processing circuit 15. For example, the input interface 11 is realized by a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touchpad that performs input operations by touching the operation surface, a touch screen in which a display screen and a touchpad are integrated, a non-contact input circuit using an optical sensor, a voice input circuit, and the like. The input interface 11 may be configured as a tablet terminal or the like that can wirelessly communicate with the main body of the radiation therapy device 10. In addition, the input interface 11 is not limited to only those that have physical operating parts such as a mouse and a keyboard. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the radiation therapy device 10 and outputs the electrical signal to the processing circuit 15 is also included as an example of the input interface 11.

The display 12 displays various information. For example, the display 12 displays a cone beam CT image collected using the radiation generator 181 and the detector 182 under the control of the processing circuit 15. For example, the display 12 displays a GUI (Graphical User Interface) for receiving various instructions and settings from the user via the input interface 11. For example, the display 12 displays the results of a failure prediction made by the prediction function 153. Note that failure prediction made by the prediction function 153 will be described later. For example, the display 12 is a liquid crystal display or a CRT (Cathode Ray Tube) display. The display 12 may be a desktop type, or may be configured as a tablet terminal capable of wireless communication with the radiation therapy device 10 main body.

The memory circuitry 13 and the model memory circuitry 14 are realized, for example, by semiconductor memory elements such as RAM (Random Access Memory) and flash memory, a hard disk, an optical disk, etc. For example, the memory circuitry 13 stores a program for the circuits included in the radiation therapy device 10 to realize their functions. Also, for example, the model memory circuitry 14 records the trained model M1 generated by the model generation function 152. The trained model M1 will be described later. The memory circuitry 13 and the model memory circuitry 14 may be integrated.

The processing circuitry 15 controls the operation of the entire radiation therapy device 10 by executing a control function 151, a model generation function 152, a prediction function 153, and an output function 154. Here, the control function 151 is an example of a control unit. The model generation function 152 is an example of a model generation unit. The prediction function 153 is an example of a prediction unit. The output function 154 is an example of an output unit.

For example, the processing circuitry 15 performs radiation therapy by reading out a program corresponding to the control function 151 from the storage circuitry 13 and executing it. For example, after the patient P1 is placed on the top plate 161 of the bed 16, the control function 151 controls the bed 16 to adjust the position and angle of the patient P1 relative to the radiation generator 17. That is, the control function 151 controls the bed 16 to move and rotate (tilt) the top plate 161 with the patient P1 placed on it. Also, for example, the control function 151 controls the radiation aperture 172 to move the leaf (aperture blade) 1721 of the radiation aperture 172 to control the size and shape of the radiation irradiation range. Also, for example, the control function 151 controls the supply of high voltage to the radiation generator 171 to generate radiation for treatment. As an example, the control function 151 controls a high voltage generator (not shown) to control the voltage and application time applied to the radiation generator 171 so that a dose of radiation in accordance with the treatment plan is generated.

The processing circuit 15 also generates a trained model M1 by reading out a program corresponding to the model generation function 152 from the storage circuit 13 and executing it. The processing circuit 15 also predicts a failure related to the movable device using the trained model M1 by reading out a program corresponding to the prediction function 153 from the storage circuit 13 and executing it. The processing circuit 15 also outputs the failure prediction result by the prediction function 153 by reading out a program corresponding to the output function 154 from the storage circuit 13 and executing it. The processing by the model generation function 152, the prediction function 153, and the output function 154 will be described later.

In the radiation therapy device 10 shown in FIG. 1, each processing function is stored in the memory circuitry 13 in the form of a program executable by a computer. The processing circuitry 15 is a processor that realizes the function corresponding to each program by reading and executing the program from the memory circuitry 13. In other words, when each program is read, the processing circuitry 15 has the function corresponding to the read program.

In FIG. 1, the control function 151, model generation function 152, prediction function 153, and output function 154 are realized by a single processing circuit 15. However, the processing circuit 15 may be configured by combining multiple independent processors, and each processor may execute a program to realize the functions.

In addition, in FIG. 1, a single memory circuit 13 is described as storing a program corresponding to each processing function. However, the embodiment is not limited to this. For example, a configuration may be adopted in which multiple memory circuits 13 are distributed and the processing circuit 15 reads out the corresponding program from each individual memory circuit 13. Also, instead of storing the program in the memory circuit 13, the program may be directly built into the circuit of the processor. In this case, the processor realizes the function by reading and executing the program built into the circuit.

The bed 16 is a bed having a top plate 161 on which the patient P1 is placed. Note that the patient P1 is not included in the radiation therapy device 10. For example, the bed 16 includes a bed driving device 162 (not shown), which moves and rotates (tilts) the top plate 161 with the patient P1 placed on it. For example, the bed driving device 162 moves or rotates the top plate 161 by driving an actuator such as a motor under the control of the control function 151.

In other words, the tabletop 161 is a part that can perform mechanical operations such as movement and rotation, and is an example of a movable part. The bed drive device 162 is a device that operates the tabletop 161, and is an example of a drive device. The bed 16, which includes the tabletop 161 and the bed drive device 162, is an example of a movable device.

For example, the bed 16 is provided with a sensor that detects the movement of the top plate 161. As an example, the bed 16 is provided with sensors such as a rotary encoder and a potentiometer. Here, the rotary encoder makes it possible to detect the amount of movement and rotation of the top plate 161. That is, the rotary encoder makes it possible to detect the relative position of the top plate 161 after movement with respect to the initial position, the inclination of the top plate 161 with respect to the direction of gravity, etc. Also, the potentiometer makes it possible to detect the absolute position of the top plate 161. For example, the control function 151 controls the bed 16 while checking the current position and inclination of the top plate 161 with the sensor, and moves and rotates the top plate 161 so as to achieve the target position and inclination.

As shown in FIG. 2, the radiation generating device 17 has a radiation generator 171 and a radiation aperture 172.

The radiation generator 171 includes an electron gun and an acceleration tube. The electron gun generates thermoelectrons. The acceleration tube accelerates the thermoelectrons generated by the electron gun. For example, under the control of the control function 151, the radiation generator 171 receives a high voltage from a high-voltage generator (not shown) to accelerate the thermoelectrons. The radiation generator 171 then causes the accelerated thermoelectrons to collide with a tungsten target to generate radiation for treatment.

The radiation constrictor 172 has leaves 1721 that control the irradiation range of the therapeutic radiation, and a leaf drive device 1722 that operates the leaves 1721. For example, the radiation constrictor 172 has a plurality of leaves 1721 that are movable relative to the radiation generator 171. Furthermore, the leaf drive device 1722 moves each of the plurality of leaves 1721 by driving an actuator such as a motor under the control of the control function 151. In other words, the control function 151 controls the radiation constrictor 172 to move each of the plurality of leaves 1721, thereby controlling the size and shape of the radiation irradiation range.

In other words, the leaf 1721 is a part capable of performing a movement operation, and is an example of a movable part. Furthermore, the leaf drive device 1722 is a device that operates the leaf 1721, and is an example of a drive device. Furthermore, the radiation squeezer 172 including the leaf 1721 and the leaf drive device 1722 is an example of a movable device.

Here, an example of the radiation narrowing device 172 will be described with reference to FIG. 3. FIG. 3 is a diagram showing an example of the radiation narrowing device 172 according to the first embodiment. Note that in FIG. 3, a case will be described in which the radiation narrowing device 172 is a multi-leaf collimator (MLC) having multiple leaves 1721. Also, in FIG. 3, the X-ray irradiation direction is defined as the Z-axis direction, and the X-axis direction and the Y-axis direction are defined to be perpendicular to the Z-axis direction.

As shown in FIG. 3, the radiation constrictor 172 has a total of 26 leaves, consisting of two leaves, leaf La01 to leaf La02, and 24 leaves, leaf Lb01 to leaf Lb24, as the plurality of leaves 1721. Leaf La01 to leaf La02 are movable in the Y-axis direction in FIG. 3, and control the radiation irradiation range in the Y-axis direction. Leaf Lb01 to leaf Lb24 are movable in the X-axis direction in FIG. 3, and control the radiation irradiation range in the X-axis direction. Here, the control function 151 can control the shape of the radiation irradiation range by moving leaves Lb01 to leaf Lb24 individually, as shown by the dot pattern in FIG. 3. For example, the control function 151 controls the shape of the radiation irradiation range in accordance with the shape of the treatment target site of patient P1.

For example, the radiation constrictor 172 shown in FIG. 3 is provided with a sensor that detects the movement of the leaves. Such a sensor is provided for each of the leaves shown in FIG. 3. As an example, the leaves are provided with sensors such as rotary encoders and potentiometers. For example, the control function 151 controls the radiation constrictor 172 while checking the current position of each of the multiple leaves using the sensors, and moves each of the multiple leaves to a target position.

Note that in FIG. 3, the radiation irradiation range is described as being controlled by a total of 26 leaves, leaves La01 to La02 and leaves Lb01 to Lb24, but the number of leaves is not particularly limited. For example, the radiation constrictor 172 may be provided with 160 leaves in order to control the radiation irradiation range. The greater the number of leaves provided in the radiation constrictor 172, the more detailed the shape of the radiation irradiation range can be controlled.

An example of the configuration of the radiation therapy device 10 has been described above. Here, movable devices such as the bed 16 and the radiation constrictor 172 may malfunction during use. For example, drive devices such as the bed drive device 162 that moves the tabletop 161 and the leaf drive device 1722 that moves the leaf 1721 may wear out with each use, increasing mechanical play and leading to malfunction. Or, even if the movable device itself does not malfunction, a sensor that detects the operation of the movable part may malfunction. Hereinafter, malfunctions of the movable device and malfunctions of the sensor will be collectively referred to as malfunctions related to the movable device.

In particular, when the radiation constrictor 172 is a multi-leaf collimator, the radiation constrictor 172 has a large number of leaves 1721, which are movable parts. The greater the number of leaves 1721, the greater the number of leaf drive devices 1722 required, and the greater the risk of malfunction of the movable device. Alternatively, in a configuration in which one leaf drive device 1722 operates multiple leaves 1721, the greater the number of leaves 1721, the more complex the leaf drive device 1722 becomes, and the greater the risk of malfunction of the movable device. In addition, sensors are often provided individually for each leaf. Therefore, the greater the number of leaves, the greater the risk of malfunction of the sensor.

Furthermore, in some cases, multiple sensors are provided for one moving part to detect sensor failure. In other words, if the detection results from multiple sensors do not match, it is easy to determine that one of the multiple sensors is faulty, reducing the risk of overlooking a failure and improving the safety of the radiation therapy device 10. However, in this case, the number of sensors increases, and so does the risk of sensor failure.

In order to deal with malfunctions related to movable devices, it is possible to record an error log. That is, in order to shorten the time it takes for the radiation therapy device 10 to recover from a system failure, it is possible to keep an error log as a hint for recovery. However, even if the time to recovery can be shortened, as long as the radiation therapy device 10 is in a system failure state, it is unavoidable that some effect will be caused on the treatment plan.

Therefore, in the radiotherapy device 10, malfunctions related to movable devices are predicted by the processing circuit 15, which will be described in detail below, to prevent the radiotherapy device 10 from going down.

First, the model generation function 152 generates a trained model M1 and stores it in the model storage circuit 14. Here, the trained model M1 is a type of artificial intelligence (AI) and is functionalized to predict failures related to movable devices. The generation process of the trained model M1 by the model generation function 152 will be described below with reference to FIG. 4. FIG. 4 is a diagram showing an example of the generation process of the trained model M1 according to the first embodiment.

In FIG. 4, a case where the trained model M1 is configured by a neural network will be described. A neural network is a network that has a structure in which adjacent layers arranged in layers are connected, and information propagates from the input layer side to the output layer side. For example, the model generation function 152 generates the trained model M1 by performing deep learning on a multi-layered neural network using the learning data (input side teacher data and output side teacher data) shown in FIG. 4. Note that the multi-layered neural network is configured, for example, by an input layer, multiple intermediate layers (hidden layers), and an output layer.

Here, the various types of learning data shown in FIG. 4 will be described. In FIG. 4, sensor information, additional information, and fault history information are exemplified as learning data. The sensor information and additional information are collected by the control function 151, for example, each time the radiation therapy device 10 is used, and stored in the memory circuitry 13. Furthermore, the fault history information is collected by the control function 151, for example, each time a fault occurs in a movable device, and stored in the memory circuitry 13.

The sensor information shown in FIG. 4 is information acquired based on the detection results obtained by a sensor detecting the operation of the movable part. The sensor information is an example of the first information regarding the operation of the movable part. For example, when operating a movable part such as the top plate 161 or the leaf 1721, the control function 151 moves or rotates the movable part while checking the current position and inclination of the movable part using a sensor such as a rotary encoder or a potentiometer. Here, the control function 151 acquires the sensor information based on the detection results by the sensor and stores it in the memory circuit 13.

More specifically, an example of sensor information is the accumulated movement amount shown in FIG. 4. That is, if the position data of the movable part is detected by a sensor when the movable part is moved, the control function 151 can calculate the distance moved by the movable part based on this position data. For example, the control function 151 can calculate the amount of displacement between two consecutive pieces of position data among a plurality of pieces of position data collected over time, and calculate the sum of the amounts of displacement as the distance moved by the movable part. In addition, the control function 151 can calculate the distance moved by the movable part each time the radiation therapy device 10 is used, and calculate the accumulated amount of movement of the movable part by adding it to the distance moved by the movable part in the past. Then, the control function 151 stores the calculated accumulated amount of movement in the memory circuitry 13 as sensor information.

Another example of sensor information is the difference data shown in FIG. 4. For example, when multiple sensors are provided for one moving part, the control function 151 calculates difference data between the detection results of the multiple sensors and determines whether the difference data exceeds a threshold value, thereby monitoring sensor failure.

As an example, in a case where sensors B1 and B2 are provided for movable part A1, control function 151 calculates distance C1 traveled by movable part A1 based on the detection result of sensor B1 detecting the operation of movable part A1, and calculates distance C2 traveled by movable part A1 based on the detection result of sensor B2 detecting the operation of movable part A1. Next, control function 151 calculates difference data D1 by subtracting distance C1 from distance C2. Then, control function 151 determines whether difference data D1 exceeds a threshold value, thereby determining whether a failure has occurred in sensors B1 and B2. Furthermore, control function 151 stores difference data D1 in memory circuit 13 as sensor information.

Note that sensors B1 and B2 may be the same type of sensor or different types of sensors. For example, control function 151 may detect the operation of one moving part using sensor B1, which is a rotary encoder, and sensor B2, which is a potentiometer.

Another example of sensor information is the image information shown in FIG. 4. Here, the image information is, for example, position data of the movable part acquired based on image data of the movable part. As an example, the control function 151 uses a camera as a sensor, continuously images the movable part when moving the movable part, and collects multiple frames of image data. The control function 151 also extracts the movable part from each of the collected image data to acquire position data of the movable part in each frame, and stores the data in the memory circuit 13 as sensor information. Here, the control function 151 can also calculate the cumulative movement amount of the movable part from the position data of the movable part in each frame. Alternatively, the control function 151 can acquire multiple position data based on the image data captured by each camera by capturing an image of the movable part using multiple cameras, and calculate difference data indicating the difference between the multiple acquired position data.

The additional information shown in FIG. 4 is information indicating the environment when the movable part operates. The additional information is an example of second information. That is, the movable part operates under various environments regardless of the operation of the movable part itself. The control function 151 acquires information indicating the environment when the movable part operates, for example, based on a treatment plan, and stores it in the memory circuitry 13.

For example, the radiation generating device 17 rotates around the patient P1 depending on the radiation irradiation direction for the patient P1, and the application of gravity to the leaf 1721 changes accordingly. Therefore, the control function 151 acquires tilt information indicating the tilt of the leaf 1721 with respect to the direction of gravity, and stores this in the memory circuitry 13 as additional information. Note that the radiation irradiation direction is set in advance as a treatment plan. Therefore, the control function 151 can acquire the radiation irradiation direction based on the treatment plan, and acquire tilt information indicating the tilt of the leaf 1721 with respect to the direction of gravity based on the radiation irradiation direction.

For example, there are various types of radiation therapy, and the degree of wear that occurs in the movable device varies depending on the type of therapy. For example, in radiation therapy using multiple portal irradiation, the radiation generating device 17 is rotated around the patient P1 to change the radiation irradiation direction, and a predetermined amount of radiation is irradiated from multiple irradiation directions. In other words, in radiation therapy using multiple portal irradiation, there is no need to move the leaves 1721 in accordance with the change in the radiation irradiation direction, so the degree of wear that occurs in the leaf drive device 1722 that operates the leaves 1721 and the sensor that detects the movement of the leaves 1721 is relatively small.

On the other hand, in Intensity Modulated Radiation Therapy (IMRT), the radiation generating device 17 is rotated around the patient P1 to change the radiation irradiation direction, and the leaves 1721 are moved according to the radiation irradiation range set for each irradiation direction to irradiate radiation. In Intensity Modulated Arc Therapy (VMAT), the radiation generating device 17 is rotated around the patient P1 to change the radiation irradiation direction, and the leaves 1721 are moved to change the radiation irradiation range to irradiate radiation. That is, in IMRT and VMAT, the leaves 1721 need to move in accordance with the change in the radiation irradiation direction, so the degree of wear that occurs in the leaf drive device 1722 and the sensor is relatively large.

The control function 151 then acquires surgical procedure information indicating the surgical procedure used when performing radiation therapy, and stores it in the memory circuitry 13 as additional information. Note that the surgical procedure is set in advance as a treatment plan. Therefore, the control function 151 can acquire surgical procedure information based on the treatment plan, and store it in the memory circuitry 13.

In addition, for example, when radiation therapy is performed, the radiation emitted from the radiation generator 171 or scattered rays based on this radiation may be irradiated onto the movable device or sensor. In some cases, the radiation may deteriorate various components in the movable device or sensor, causing malfunctions of the movable device.

For example, in the radiation constrictor 172, the leaves 1721 are generally made of metal and therefore do not deteriorate due to radiation. However, the leaf drive device 1722 that operates the leaves 1721 and the sensor that detects the operation of the leaves 1721 may use semiconductor components. Furthermore, semiconductor components may deteriorate when exposed to radiation, which may be a cause of failure of the radiation constrictor 172.

The control function 151 then acquires radiation information indicating the amount of radiation used when performing radiation therapy, and stores this in the memory circuitry 13 as additional information. Note that the intensity and exposure time of the radiation used in radiation therapy are set in advance as a treatment plan. Therefore, the control function 151 can acquire the intensity and exposure time of the radiation based on the treatment plan, calculate the amount of radiation, and store this in the memory circuitry 13 as radiation information.

The failure history information shown in FIG. 4 is an error log. When a failure occurs in a movable device, the control function 151 creates an error log that includes at least the time of the failure and stores it in the memory circuit 13.

In addition, the control function 151 stores information about the malfunction of the movable device and information about the malfunction of the sensor in the memory circuit 13, along with the malfunction time.

For example, when a malfunction occurs in a movable device, the control function 151 stores information indicating which of the multiple movable devices equipped in the radiation therapy device 10 has malfunctioned as information regarding the malfunction in the memory circuitry 13 in association with the malfunction time. For example, when the radiation therapy device 10 has multiple movable devices such as the bed 16 and the radiation generator 17, the control function 151 stores information indicating which of the multiple movable devices equipped in the radiation therapy device 10 has malfunctioned in the memory circuitry 13 in association with the malfunction time.

For example, the control function 151 stores information indicating the location of the movable device where the failure occurred in the memory circuit 13 as information regarding the malfunction of the movable device in association with the malfunction time. For example, when the radiation constrictor 172 includes a plurality of leaves 1721 and a plurality of leaf drive devices 1722 corresponding to each of the leaves 1721, the control function 151 stores information indicating which of the plurality of leaf drive devices 1722 has malfunctioned in association with the malfunction time in the memory circuit 13. For example, the bed drive device 162 that operates the top plate 161 may have an axis for moving the top plate 161 in the horizontal direction and an axis for moving the top plate 161 in the vertical direction. Here, when a malfunction occurs in the bed 16, the control function 151 stores information indicating which of the axis for moving the top plate 161 in the horizontal direction and the axis for moving the top plate 161 in the vertical direction has malfunctioned in association with the malfunction time in the memory circuit 13.

In addition, for example, if a sensor failure occurs, the control function 151 stores information about the sensor failure, which indicates the sensor that has failed among multiple sensors provided to detect the operation of movable parts such as the top plate 161 and the leaf 1721, in the memory circuit 13 in association with the time of the failure.

The above describes an example of the learning data used in the generation process of the trained model M1. Note that FIG. 4 lists examples of learning data, and it is not necessary to use all of the learning data in FIG. 4 when generating the trained model M1. For example, if no malfunction of the movable device has occurred and only a malfunction of the sensor has occurred, the model generation function 152 can omit the use of the "information regarding malfunction of the movable device" shown in FIG. 4 as an example of output side teacher data. Similarly, the model generation function 152 can appropriately omit the use of some of the accumulated movement amount, difference data, image information, tilt information, surgical procedure information, and radiation information shown in FIG. 4 as an example of input side teacher data.

Also, FIG. 4 shows an example of learning data, and learning data that can be used to generate the trained model M1 is not limited to this. For example, the model generation function 152 may further use speed information as input side teacher data. Here, speed information is information indicating the speed at which the movable part moves. For example, if the position data of the movable part is detected by a sensor when it is moved, the control function 151 can calculate the speed at which the movable part moves by differentiating this position data with respect to time. In other words, speed information is sensor information that can be obtained based on the detection result of the operation of the movable part detected by the sensor. In general, the higher the speed, the greater the wear that occurs, and the higher the risk of malfunction of the movable device.

The speed at which the leaves 1721 are moved may be constant or may change dynamically depending on the procedure. For example, in multi-port irradiation and IMRT, the leaves 1721 are often moved at a constant speed so that a radiation irradiation range of a desired size and shape can be formed. As an example, in multi-port irradiation and IMRT, the leaves 1721 are moved at the maximum speed that can be achieved by the leaf drive device 1722 in order to save time. On the other hand, in VMAT, the size and shape of the radiation irradiation range are continuously changed during radiation therapy, so each of the multiple leaves 1721 is moved individually while dynamically changing the speed.

For example, the model generation function 152 acquires multiple pairs of input side teacher data and output side teacher data shown in FIG. 4, and generates a learning dataset. Note that multiple pairs of input side teacher data and output side teacher data can also be generated each time a malfunction related to the movable device occurs. For example, the model generation function 152 first creates a pair of input side teacher data acquired from the memory circuitry 13 and output side teacher data, and adds it to the learning dataset. Here, the model generation function 152 can further create a pair of data obtained from the memory circuitry 13 by omitting a portion of the input side teacher data, and the output side teacher data, and add it to the learning dataset.

Next, the model generation function 152 executes machine learning based on the generated learning data set to generate a trained model M1. Specifically, the model generation function 152 first inputs the input side teacher data to the neural network. Here, in the neural network, information propagates in one direction from the input layer side to the output layer side while connecting only between adjacent layers, and the output layer outputs an estimation result of a malfunction related to the movable device. For example, the output layer of the neural network outputs an estimation result of which of the movable devices and sensors included in the radiation therapy device 10 will malfunction, and an estimation result of the time when the malfunction will occur. Note that a neural network in which information propagates in one direction from the input layer side to the output layer side is also called a convolutional neural network (CNN).

The model generation function 152 generates a trained model M1 by adjusting the parameters of the neural network so that the neural network can output a favorable result when input side teacher data is input. For example, the model generation function 152 repeats the process while adjusting the parameters of the neural network so that the estimation result by the neural network regarding which of the movable devices and sensors included in the radiation therapy device 10 will cause a malfunction coincides with the output side teacher data (correct answer data). Furthermore, the model generation function 152 repeats the process while adjusting the parameters of the neural network until the difference between the estimation result by the neural network regarding the time when a malfunction will occur and the output side teacher data (correct answer data) falls below a threshold value. In this way, the model generation function 152 generates a trained model M1 that is functionalized to predict malfunctions related to the movable devices.

Note that although the trained model M1 has been described as being configured using a convolutional neural network (CNN), the model generation function 152 may configure the trained model M1 using other types of neural networks, such as a fully connected neural network or a recurrent neural network (RNN).

The model generation function 152 may also generate the trained model M1 using a machine learning method other than a neural network. For example, the model generation function 152 may perform machine learning using algorithms such as logistic regression analysis, nonlinear discriminant analysis, a support vector machine (SVM), a random forest, and a naive Bayes to generate the trained model M1.

The model generation function 152 may also generate multiple trained models M1. For example, the model generation function 152 may generate a trained model M11 that is functionalized to predict a fault related to the bed 16, and a trained model M12 that is functionalized to predict a fault related to the radiation throttle 172.

The model generation function 152 stores the generated trained model M1 in the model storage circuit 14. Thereafter, the prediction function 153 reads out the trained model M1 from the model storage circuit 14 and predicts a malfunction related to the movable device, for example, each time the radiation therapy device 10 is used.

For example, each time the radiation therapy device 10 is used, the control function 151 first acquires sensor information such as accumulated movement amount, difference data, and image information, as well as additional information such as tilt information, surgical procedure information, and radiation information, and stores the information in the memory circuitry 13.

When a drive device, sensor, etc. is replaced or repaired due to a failure, the control function 151 may delete the sensor information and additional information related to the replaced or repaired part from the memory circuit 13. That is, the control function 151 may reset the sensor information and additional information collected up until the replacement or repair, and may control so as to start accumulating new sensor information and additional information after the replacement or repair. Similarly, when a drive device, sensor, etc. is replaced or adjusted due to a predicted failure, the control function 151 may delete the sensor information and additional information related to the replaced or adjusted part from the memory circuit 13. Adjustment of the drive device, sensor, etc. is, for example, maintenance work such as cleaning and calibration.

As shown in FIG. 5, the prediction function 153 acquires sensor information and auxiliary information from the memory circuit 13 and inputs them to the trained model M1 read from the model memory circuit 14. Here, the trained model M1 predicts a failure related to the movable device based on these input data. That is, the prediction function 153 predicts a failure related to the movable device by inputting first information related to the operation of the movable part and second information indicating the environment when the movable part operates to the trained model M1. Specifically, as shown in FIG. 5, the trained model M1 generates and outputs failure prediction information including the time of failure, information related to the failure of the movable device, and information related to the failure of the sensor. Note that FIG. 5 is a diagram showing an example of use of the trained model M1 according to the first embodiment.

Then, the output function 154 outputs the failure prediction result by the prediction function 153. For example, the output function 154 causes the display 12 to display the failure prediction result by the prediction function 153. Note that, if it is predicted that no failure will occur, the output function 154 may cause the display 12 to display information indicating that no failure will occur as the failure prediction result, or may omit displaying the failure prediction result.

For example, the output function 154 causes the display 12 to display, as a prediction result, information indicating which of the multiple movable devices included in the radiation therapy device 10 will malfunction. As an example, in a case where the radiation therapy device 10 is equipped with multiple movable devices such as the bed 16 and the radiation generator 17, the output function 154 causes the display 12 to display, as a prediction result, information indicating which of the multiple movable devices will malfunction. The output function 154 also causes the display 12 to display, as a prediction result, the time when the malfunction will occur.

Also, for example, the output function 154 displays on the display 12, as a prediction result, information indicating a location of the movable device where a failure will occur. As an example, in a case where the radiation constrictor 172 has a plurality of leaves 1721 and a plurality of leaf drive devices 1722 corresponding to each of the leaves 1721, the output function 154 displays on the display 12, as a prediction result, information indicating which of the plurality of leaf drive devices 1722 will fail. As another example, the output function 154 displays on the display 12 information indicating which of the plurality of axes (e.g., an axis for moving the tabletop 161 in the horizontal direction and an axis for moving the tabletop 161 in the vertical direction) of the bed drive device 162 will fail. As a prediction result, the output function 154 displays on the display 12 the time when the failure will occur.

In addition, for example, the output function 154 causes the display 12 to display, as a prediction result, information indicating which sensor will fail among a plurality of sensors provided to detect the operation of movable parts such as the top plate 161 and the leaf 1721. In addition, the output function 154 causes the display 12 to display, as a prediction result, the time when the failure will occur.

Note that, although the failure prediction results by the prediction function 153 are described as being displayed on the display 12, the embodiment is not limited to this. For example, the output function 154 may transmit the failure prediction results by the prediction function 153 to another device via a network, and the other device may display the failure prediction results by the prediction function 153.

That is, the output function 154 notifies the user of the failure prediction result by the prediction function 153. Here, the user can appropriately manage the state of the radiation therapy device 10 by referring to the failure prediction result. For example, if a failure is predicted to occur in the near future, the user can request a contractor to replace or adjust the drive device, sensor, etc., which is predicted to fail, and prevent the occurrence of a failure related to the movable device in advance. That is, the processing circuit 15 can avoid a system down of the radiation therapy device 10 by predicting a failure related to the movable device and notifying the user. On the other hand, if a failure is predicted to occur but is far away, or if it is predicted that no failure will occur, the user can continue to use the radiation therapy device 10 with peace of mind.

The user may be, for example, a doctor or technician at a medical facility where the radiation therapy device 10 is installed, or a serviceman dispatched to the medical facility by the manufacturer of the radiation therapy device 10.

Next, an example of the processing procedure by the radiation therapy device 10 will be described with reference to FIG. 6. FIG. 6 is a flowchart for explaining a series of processing steps by the radiation therapy device 10 according to the first embodiment. Steps S101, S102, S103, S107, S108, S109, S112, and S113 correspond to the control function 151. Steps S110 and S111 correspond to the model generation function 152. Step S104 corresponds to the prediction function 153. Steps S105 and S106 correspond to the output function 154.

First, the processing circuitry 15 controls movable devices such as the bed 16 and the radiation constrictor 172 to perform radiation therapy (step S101). The processing circuitry 15 also stores the first information and the second information in the memory circuitry 13 (step S102). For example, during the radiation therapy in step S101, the processing circuitry 15 acquires various sensor information shown in FIG. 4 based on the detection results of the movement of the movable part detected by the sensor, and stores the information in the memory circuitry 13 as the first information. For example, the processing circuitry 15 also acquires various supplementary information shown in FIG. 4 based on the treatment plan for radiation therapy in step S101, and stores the information in the memory circuitry 13 as the second information.

Next, the processing circuitry 15 determines whether or not a malfunction has occurred in the movable device during the radiation therapy in step S101 (step S103). If no malfunction has occurred (step S103: No), the processing circuitry 15 reads out the trained model M1 from the model storage circuitry 14, and inputs the first information and the second information to the trained model M1 to predict a malfunction in the movable device (step S104). Here, the processing circuitry 15 determines whether or not a malfunction has been predicted (step S105), and if a malfunction has been predicted (step S105: Yes), it outputs the result of the malfunction prediction (step S106).

When a user is notified of the predicted results of a breakdown, the user can request a contractor to replace or adjust the drive unit, sensor, etc. that is predicted to break down. However, if the predicted time for a breakdown is far away, the user can postpone the replacement or adjustment. The processing circuit 15 then determines whether or not a breakdown or adjustment has actually occurred (step S107). If replacement or adjustment has been performed (Yes in step S107), the processing circuit 15 deletes the first information and second information related to the replaced or adjusted part from the memory circuit 13 (step S108).

On the other hand, if a failure occurs during radiation therapy in step S101 (Yes in step S103), the processing circuitry 15 acquires various types of failure history information shown in FIG. 4 and stores it in the memory circuitry 13 (step S109). For example, if a malfunction occurs in a movable device, the processing circuitry 15 acquires failure history information that associates information about the malfunction of the movable device with the time when the malfunction occurred, and stores it in the memory circuitry 13. Also, for example, if a malfunction occurs in a sensor, the processing circuitry 15 acquires failure history information that associates information about the sensor malfunction with the time when the malfunction occurred, and stores it in the memory circuitry 13.

Next, the processing circuitry 15 generates a learning dataset including multiple pairs of input-side teacher data and output-side teacher data shown in FIG. 4. For example, the processing circuitry 15 adds a pair of the first information and the second information stored in the memory circuitry 13 in step S102 and the fault history information stored in the memory circuitry 13 in step S109 to a previously generated learning dataset to generate a new learning dataset (step S110).

Next, the processing circuit 15 updates the trained model M1 stored in the model storage circuit 14 (step S111). That is, the processing circuit 15 executes machine learning based on the learning dataset generated in step S110 to generate a new trained model M1, and stores it in the model storage circuit 14. After updating the trained model M1, the processing circuit 15 determines whether the faulty part has been replaced or repaired (step S112), and enters a standby state if the part has not been replaced or repaired (step S112: No).

After step S108, if it is predicted in step S105 that no failure will occur, or if it is determined in step S112 that replacement or repair has been performed, the processing circuit 15 determines whether or not the next radiation therapy is scheduled (step S113), and if the next radiation therapy is scheduled, proceeds again to step S101 (step S113: Yes). On the other hand, if the next radiation therapy is not scheduled (step S113: No), the processing circuit 15 ends the process.

The flowchart in FIG. 6 can be modified as appropriate. For example, the processing circuit 15 may omit step S105 and output the failure prediction result regardless of whether or not a failure is predicted to occur. Also, for example, the processing circuit 15 may omit step S108 and store the first information and the second information in the memory circuit 13.

In addition, although the flowchart in FIG. 6 describes a case where failure prediction is performed using both the first information and the second information, the processing circuit 15 can also perform failure prediction using either the first information or the second information. For example, the processing circuit 15 can input only the accumulated movement amount shown in FIG. 5 as the first information into the trained model M1 to predict a failure related to a movable device.

As described above, according to the first embodiment, the control function 151 controls movable devices such as the bed 16 and the radiation generator 17 to perform radiation therapy. The model storage circuit 14 also stores a trained model M1 that is functionalized to predict malfunctions of the movable devices. The prediction function 153 also predicts malfunctions of the movable devices by inputting at least one of the first information regarding the operation of the movable parts such as the leaf 1721 and the tabletop 161, and the second information indicating the environment when the movable parts operate, into the trained model M1. The output function 154 also outputs the result of the malfunction prediction by the prediction function 153. Therefore, the radiation therapy device 10 according to the first embodiment predicts malfunctions of the movable devices and notifies the user, thereby making it possible to repair or adjust parts before a malfunction occurs, and thus avoiding a system downtime.

Although the bed 16 and the radiation constrictor 172 have been described as examples of movable devices, the embodiment is not limited thereto. For example, the rotating gantry 19 of the radiation therapy device 10 is an example of a movable part capable of rotating. The radiation therapy device 10 also has a rotating gantry drive device 191 (not shown), which drives an actuator such as a motor to rotate the rotating gantry 19.
Here, the radiation therapy apparatus 10 can predict failures related to the movable devices, including the rotating gantry 19 and the rotating gantry drive device 191, as in the case of the bed 16 and the radiation constrictor 172, and notify the user of the failures.

Although the above description concerns a case where the generation process of the trained model M1 is executed in the radiation therapy device 10, the embodiment is not limited to this. For example, instead of the processing circuitry 15 executing the model generation function 152, the radiation therapy device 10 may acquire the trained model M1 generated in an external device and store it in the model storage circuitry 14.

Second Embodiment
In the above-described first embodiment, a radiation therapy apparatus capable of performing radiation therapy has been described as an example of a radiation apparatus, whereas in the second embodiment, a radiation diagnostic apparatus capable of performing image acquisition using radiation will be described.

Below, an X-ray CT device 20 shown in FIG. 7 will be described as an example of a radiation diagnostic device. FIG. 7 is a block diagram showing an example of the configuration of the X-ray CT device 20 according to the second embodiment. As shown in FIG. 7, the X-ray CT device 20 has a gantry device 210, a bed 230, and a console device 240.

In FIG. 7, the rotation axis of the rotating frame 213 when not tilted or the longitudinal direction of the top plate 233 of the bed 230 is the Z-axis direction. The axis direction that is perpendicular to the Z-axis direction and horizontal to the floor surface is the X-axis direction. The axis direction that is perpendicular to the Z-axis direction and perpendicular to the floor surface is the Y-axis direction. For the sake of explanation, FIG. 7 shows the gantry device 210 from multiple directions, and shows a case where the X-ray CT device 20 has one gantry device 210.

The gantry device 210 has an X-ray tube 211, an X-ray detector 212, a rotating frame 213, an X-ray high voltage device 214, a control device 215, a wedge device 216, a collimator 217, and a DAS 218.

The X-ray tube 211 is a vacuum tube having a cathode (filament) that generates thermoelectrons and an anode (target) that generates X-rays when the thermoelectrons collide with it. The X-ray tube 211 generates X-rays to be irradiated to the patient P2 by irradiating thermoelectrons from the cathode to the anode through application of high voltage from the X-ray high voltage device 214.

The X-ray detector 212 detects X-rays emitted from the X-ray tube 211 and passing through the patient P2, and outputs a signal corresponding to the detected X-ray amount to the DAS 218. The X-ray detector 212 has, for example, multiple detection element rows in which multiple detection elements are arranged in the channel direction along an arc centered on the focus of the X-ray tube 211. The X-ray detector 212 has a structure in which, for example, multiple detection element rows in which multiple detection elements are arranged in the channel direction are arranged in the row direction. The row direction is also referred to as the slice direction, row direction, or segment direction. The row direction corresponds to the body axis direction (Z-axis direction) of the patient P2 shown in FIG. 7.

For example, the X-ray detector 212 is an indirect conversion type detector having a grid, a scintillator array, and a photosensor array. The scintillator array has multiple scintillators. The scintillator has scintillator crystals that output light with a photon amount corresponding to the amount of incident X-rays. The grid is arranged on the X-ray incident side of the scintillator array and has an X-ray shielding plate that absorbs scattered X-rays. The grid is sometimes called a collimator (one-dimensional collimator or two-dimensional collimator). The photosensor array has a function of converting the amount of light from the scintillator into an electrical signal corresponding to the amount of light, and has a photosensor such as a photodiode. The X-ray detector 212 may be a direct conversion type detector having a semiconductor element that converts the incident X-rays into an electrical signal.

The rotating frame 213 is an annular frame that supports the X-ray tube 211 and the X-ray detector 212 facing each other and rotates the X-ray tube 211 and the X-ray detector 212 by the control device 215. For example, the rotating frame 213 is a casting made of aluminum. In addition to the X-ray tube 211 and the X-ray detector 212, the rotating frame 213 can also support the X-ray high voltage device 214, the wedge device 216, the collimator 217, the DAS 218, etc. Furthermore, the rotating frame 213 can also support various components not shown in FIG. 7. Hereinafter, the rotating frame 213 and the part of the gantry device 210 that rotates together with the rotating frame 213 are also referred to as the rotating part.

The X-ray high voltage device 214 has electrical circuits such as a transformer and a rectifier, and includes a high voltage generator that generates a high voltage to be applied to the X-ray tube 211, and an X-ray control device that controls the output voltage according to the X-rays generated by the X-ray tube 211. The high voltage generator may be of a transformer type or an inverter type. The X-ray high voltage device 214 may be provided on the rotating frame 213, or on a fixed frame (not shown).

The control device 215 operates the rotating part of the gantry 210. For example, the control device 215 rotates the rotating part of the gantry 210 by driving an actuator such as a motor under the control of the processing circuit 245. Also, for example, the control device 215 tilts the rotating part under the control of the processing circuit 245. As an example, the control device 215 tilts the rotating part around an axis parallel to the X-axis direction according to tilt angle information input via the input interface 241 as a control to tilt the rotating part.

In other words, the rotating part of the gantry 210 is a part capable of performing mechanical operations such as rotation and tilt, and is an example of a movable part. Furthermore, the control device 215 is a device that operates the rotating part, and is an example of a drive device. Furthermore, the gantry 210 including the rotating part and the control device 215 is an example of a movable device.

The wedge device 216 includes, for example, a wedge 2161 capable of moving or rotating, and a wedge driving device 2162 for operating the wedge 2161. Here, the wedge 2161 is an X-ray filter for adjusting the amount of X-rays irradiated from the X-ray tube 211. Specifically, the wedge 2161 is an X-ray filter for attenuating the X-rays irradiated from the X-ray tube 211 so that the X-rays irradiated from the X-ray tube 211 to the patient P2 have a predetermined distribution. For example, the wedge 2161 is a wedge filter or a bow-tie filter, and is manufactured by processing aluminum or the like to have a predetermined target angle and a predetermined thickness. In addition, the wedge driving device 2162 operates the wedge 2161 by driving an actuator such as a motor.

In other words, the wedge 2161 is a part that can perform mechanical operations such as movement and rotation, and is an example of a movable part. Furthermore, the wedge driving device 2162 is a device that operates the wedge 2161, and is an example of a driving device. Furthermore, the wedge device 216, which includes the wedge 2161 and the wedge driving device 2162, is an example of a movable device.

The collimator 217 includes, for example, aperture blades 2171 capable of performing a moving operation, and an aperture blade drive device 2172 that operates the aperture blades 2171. Here, the aperture blades 2171 are lead plates or the like for narrowing the irradiation range of the X-rays that have passed through the wedge 2161. For example, the collimator 217 includes multiple aperture blades 2171, and a slit is formed by combining these multiple aperture blades 2171. In addition, the aperture blade drive device 2172 operates the aperture blades 2171 by driving an actuator such as a motor.

In other words, the aperture blade 2171 is a movable part and is an example of a movable part. The aperture blade drive device 2172 is a device that operates the aperture blade 2171 and is an example of a drive device. The collimator 217 including the aperture blade 2171 and the aperture blade drive device 2172 is an example of a movable device.

Note that, although FIG. 7 shows a case where the wedge device 216 is disposed between the X-ray tube 211 and the collimator 217, the collimator 217 may be disposed between the X-ray tube 211 and the wedge device 216. In this case, the wedge 2161 transmits and attenuates the X-rays irradiated from the X-ray tube 211 and whose irradiation range is limited by the collimator 217.

DAS218 collects X-ray signals detected by each detection element of X-ray detector 212. For example, DAS218 has an amplifier that amplifies the electrical signal output from each detection element and an A/D converter that converts the electrical signal into a digital signal, and generates detection data. DAS218 is realized, for example, by a processor.

The data generated by the DAS 218 is transmitted by optical communication from a transmitter having a light emitting diode (LED) provided on the rotating frame 213 to a receiver having a photodiode provided on a non-rotating part of the gantry 210 (e.g., a fixed frame, etc. (not shown in FIG. 1)), and then transferred to the console device 240. Here, the non-rotating part is, for example, a fixed frame that rotatably supports the rotating frame 213. Note that the method of transmitting data from the rotating frame 213 to the non-rotating part of the gantry 210 is not limited to optical communication, and any non-contact data transmission method or a contact data transmission method may be adopted.

The bed 230 is a bed on which the patient P2 is placed, and includes a base 231, a bed drive device 232, a top plate 233, and a support frame 234. Note that the patient P2 is not included in the X-ray CT device 20. The base 231 is a housing that supports the support frame 234 so that it can move in the vertical direction. The support frame 234 is a frame that supports the top plate 233 so that it can move in the longitudinal direction.

For example, the bed driving device 232 drives an actuator such as a motor under the control of the processing circuit 245 to move the support frame 234 in the vertical direction. Also, for example, the bed driving device 232 moves the tabletop 233 in the longitudinal direction. Note that the bed driving device 232 may move the support frame 234 in the longitudinal direction of the tabletop 233 in addition to the tabletop 233.

In other words, the tabletop 233 and the support frame 234 are movable parts and are an example of a movable part. The bed driving device 232 is a device that operates the tabletop 233 and the support frame 234 and is an example of a driving device. The bed 230 including the table driving device 232, the tabletop 233, and the support frame 234 is an example of a movable device.

The console device 240 has an input interface 241, a display 242, a memory circuit 243, a model memory circuit 244, and a processing circuit 245. Note that although the console device 240 is described as being separate from the gantry device 210, the gantry device 210 may include the console device 240 or some of the components of the console device 240.

The input interface 241 accepts various input operations from the user, converts the accepted input operations into electrical signals, and outputs the electrical signals to the processing circuit 245. For example, the input interface 241 is realized by a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touchpad that performs input operations by touching the operation surface, a touch screen in which a display screen and a touchpad are integrated, a non-contact input circuit using an optical sensor, a voice input circuit, or the like. The input interface 241 may be provided in the pedestal device 210. The input interface 241 may also be configured as a tablet terminal or the like that can wirelessly communicate with the console device 240 main body. The input interface 241 is not limited to only those that have physical operating parts such as a mouse and a keyboard. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the console device 240 and outputs the electrical signal to the processing circuit 245 is also included as an example of the input interface 241.

The display 242 displays various information. For example, the display 242 displays CT images collected using the X-ray tube 211 and the X-ray detector 212 under the control of the processing circuit 245. Also, for example, the display 242 displays a GUI for accepting various instructions and settings from the user via the input interface 241. Also, for example, the display 242 displays the results of a failure prediction made by the prediction function 245c. For example, the display 242 is a liquid crystal display or a CRT display. The display 242 may be a desktop type, or may be configured as a tablet terminal capable of wireless communication with the console device 240 main body.

The memory circuitry 243 and the model memory circuitry 244 are realized, for example, by semiconductor memory elements such as RAM and flash memory, a hard disk, an optical disk, etc. For example, the memory circuitry 243 stores a program for enabling the circuits included in the X-ray CT device 20 to realize their functions. Also, for example, the model memory circuitry 244 records the trained model M2 generated by the model generation function 245b. Note that the memory circuitry 243 and the model memory circuitry 244 may be integrated.

The processing circuitry 245 controls the operation of the entire X-ray CT device 20 by executing a control function 245a, a model generation function 245b, a prediction function 245c, and an output function 245d. Here, the control function 245a is an example of a control unit. The model generation function 245b is an example of a model generation unit. The prediction function 245c is an example of a prediction unit. The output function 245d is an example of an output unit.

For example, the processing circuitry 245 executes image collection by reading a program corresponding to the control function 245a from the storage circuitry 243 and executing the program. Note that the control function 245a is the same function as the control function 151 described above, except for whether it executes image collection or radiation therapy.

For example, the control function 245a controls the wedge device 216 to move and rotate the wedge 2161. Also, for example, the control function 245a controls the collimator 217 to move the aperture blades 2171 to adjust the aperture degree, etc. Also, for example, the control function 245a controls the gantry device 210 to tilt and rotate the rotating part. Also, for example, the control function 245a controls the bed 230 to move the top plate 233 and the support frame 234.

For example, the control function 245a controls the X-ray high voltage device 214 to supply a high voltage to the X-ray tube 211. This causes the X-ray tube 211 to generate X-rays to be irradiated to the patient P2. While the patient P2 is being irradiated with X-rays, the DAS 218 collects X-ray signals from each detection element in the X-ray detector 212 and generates detection data. The control function 245a also performs preprocessing on the detection data output from the DAS 218. For example, the control function 245a performs preprocessing such as logarithmic conversion processing, offset correction processing, sensitivity correction processing between channels, and beam hardening correction on the detection data output from the DAS 218. Note that data after preprocessing is also referred to as raw data. The detection data before preprocessing and the raw data after preprocessing are also collectively referred to as projection data.

The control function 245a also generates image data based on the pre-processed projection data. For example, the control function 245a generates CT image data (volume data) by performing reconstruction processing on the projection data using a filtered backprojection method, an iterative reconstruction method, an iterative reconstruction method, or the like. The control function 245a can also generate CT image data by performing reconstruction processing using a machine learning technique. For example, the control function 245a generates CT image data using a DLR (Deep Learning Reconstruction) method.

The processing circuitry 245 also generates a trained model M2 by reading out a program corresponding to the model generation function 245b from the storage circuitry 243 and executing it. The model generation function 245b is the same function as the model generation function 152 described above. The trained model M2 is an AI generated in the same manner as the trained model M1 described above, and is functionalized to predict failures related to the movable device. That is, the trained model M1 is an AI functionalized to predict failures related to the movable device provided in the radiation therapy device 10, whereas the trained model M2 is an AI functionalized to predict failures related to the movable device provided in the X-ray CT device 20.

The processing circuitry 245 also reads out a program corresponding to the prediction function 245c from the memory circuitry 243 and executes it to predict a failure related to the movable device using the trained model M2. Note that the prediction function 245c is the same function as the prediction function 153 described above.

The processing circuitry 245 also reads out a program corresponding to the output function 245d from the storage circuitry 243 and executes it to output the failure prediction result by the prediction function 245c. The output function 245d is the same function as the output function 154 described above.

In the X-ray CT device 20 shown in FIG. 7, each processing function is stored in the memory circuitry 243 in the form of a program executable by a computer. The processing circuitry 245 is a processor that realizes the function corresponding to each program by reading the program from the memory circuitry 243 and executing it. In other words, when each program has been read from the processing circuitry 245, it has the function corresponding to the read program.

In FIG. 7, the control function 245a, the model generation function 245b, the prediction function 245c, and the output function 245d are described as being realized by a single processing circuit 245, but the processing circuit 245 may be configured by combining multiple independent processors, and each processor may execute a program to realize the functions.

In addition, in FIG. 7, a single memory circuit 243 is described as storing a program corresponding to each processing function. However, the embodiment is not limited to this. For example, a configuration may be adopted in which multiple memory circuits 243 are distributed and the processing circuit 245 reads out the corresponding program from each individual memory circuit 243. Also, instead of storing the program in the memory circuit 243, the program may be directly built into the circuit of the processor. In this case, the processor realizes the function by reading and executing the program built into the circuit.

An example of the configuration of the X-ray CT device 20 has been described above. As in the case of the radiation therapy device 10, movable devices such as the wedge device 216, collimator 217, gantry device 210, and bed 230 may malfunction during use. Or, even if the movable device itself does not malfunction, a sensor that detects the operation of the movable part may malfunction. Therefore, in the X-ray CT device 20, malfunctions related to the movable devices are predicted by processing of the processing circuit 245 described below, and a system downtime of the X-ray CT device 20 is avoided.

First, the model generation function 245b generates a trained model M2 that is functionalized to predict failures related to movable devices, and stores the trained model M2 in the model storage circuit 244. Here, the model generation function 245b can generate the trained model M2 in a manner similar to the generation process of the trained model M1 by the model generation function 152.

For example, the control function 245a acquires sensor information based on the detection results of the movements of movable parts such as the wedge 2161 of the wedge device 216, the aperture blades 2171 of the collimator 217, the rotating part of the gantry device 210, and the top plate 233 and support frame 234 of the bed 230 detected by a sensor, and stores the sensor information in the memory circuitry 243. Examples of the sensor information include an accumulated amount of movement, difference data, image information, etc. The sensor information is also an example of the first information.

For example, the control function 245a acquires various types of additional information based on the imaging conditions and stores them in the memory circuitry 243. Examples of the additional information include tilt information, surgical procedure information, and radiation information. The additional information is an example of the second information.

As an example, when image collection is performed by tilting the rotating part of the mount device 210 around an axis parallel to the X-axis direction, the tilt angle of the rotating part is set in advance as an imaging condition. Therefore, the control function 245a can obtain tilt information indicating the tilt of the movable part relative to the direction of gravity based on the imaging conditions, and store the tilt information in the memory circuitry 243.

As another example, when performing image collection, various techniques such as conventional scan, helical scan, and step-and-shoot are used. Here, the technique is set in advance as an imaging condition. Therefore, the control function 245a can obtain technique information indicating the technique used when performing image collection based on the imaging condition, and store the information in the memory circuitry 243.

As another example, the intensity and exposure time of the X-rays used to collect images are set in advance as imaging conditions. Therefore, the control function 245a can obtain the intensity and exposure time of the X-rays based on the imaging conditions, calculate the amount of X-rays, and store them in the memory circuitry 243 as radiation information.

For example, the control function 245a acquires failure history information each time a failure occurs in the movable device, and stores the information in the memory circuit 243. Examples of failure history information include information regarding the time of failure, information regarding the failure of the movable device, information regarding the failure of the sensor, etc.

For example, the model generation function 245b acquires the above-mentioned sensor information and auxiliary information as input side teacher data, and acquires fault history information as output side teacher data. The model generation function 245b also acquires multiple pairs of such input side teacher data and output side teacher data, and generates a learning dataset. Furthermore, the model generation function 245b executes machine learning based on the generated learning dataset to generate a trained model M2, and stores it in the model storage circuit 244.

After the model generation function 245b generates the trained model M2 and stores it in the model storage circuit 244, the prediction function 245c reads out the trained model M2 from the model storage circuit 244 and predicts a fault related to the movable device, for example, each time the X-ray CT device 20 is used.

For example, first, each time the X-ray CT device 20 is used, the control function 245a acquires sensor information such as accumulated movement amount, difference data, image information, and associated information such as tilt information, surgical procedure information, and radiation information, and stores them in the memory circuitry 243. The prediction function 245c also acquires the sensor information and associated information from the memory circuitry 243 and inputs them to the trained model M2 read from the model memory circuitry 244. Here, the trained model M2 predicts a failure related to the movable device based on these input data. That is, the prediction function 245c predicts a failure related to the movable device by inputting first information related to the operation of the movable part and second information indicating the environment when the movable part operates into the trained model M2.

Then, the output function 245d outputs the failure prediction result by the prediction function 245c. For example, the output function 245d causes the failure prediction result by the prediction function 245c to be displayed on the display 242. Alternatively, the output function 245d may transmit the failure prediction result by the prediction function 245c to another device via a network, and the other device may display the failure prediction result by the prediction function 245c.

The processing by the processing circuit 245 described above allows the user of the X-ray CT device 20 to appropriately manage the state of the X-ray CT device 20 while referring to the failure prediction results. For example, if a failure is predicted to occur in the near future, the user can request a contractor to replace or adjust the drive device, sensor, etc., which is predicted to fail, and prevent the occurrence of a failure related to the movable device in advance. In other words, the processing circuit 245 can avoid a system downtime of the X-ray CT device 20 by predicting a failure related to the movable device and notifying the user. On the other hand, if a failure is predicted to occur but is far away, or if it is predicted that no failure will occur, the user can continue to use the X-ray CT device 20 with peace of mind.

Third Embodiment
In the above-mentioned first and second embodiments, the prediction function 153 or the prediction function 245c is described as predicting a malfunction related to the movable device. That is, in the above-mentioned first and second embodiments, the radiation device such as the radiation therapy device 10 or the X-ray CT device 20 is described as predicting a malfunction related to the movable device by itself. In contrast, in the third embodiment, a case where a device other than the radiation device predicts a malfunction related to the movable device included in the radiation device is described. Hereinafter, the same reference numerals as those in FIG. 2 or FIG. 7 are used for the components having the same configuration as those described in the first and second embodiments, and the description thereof will be omitted.

In the third embodiment, as shown in FIG. 8, a radiation system 1 including a radiation therapy device 10, an X-ray CT device 20, and a failure prediction device 30 will be described. FIG. 8 is a block diagram showing an example of the configuration of the radiation system 1 according to the third embodiment. As shown in FIG. 8, the radiation therapy device 10, the X-ray CT device 20, and the failure prediction device 30 are connected to each other via a network NW. The network NW may be configured as a closed local network within a facility, or may be a network via the Internet.

The failure prediction device 30 is a device that predicts failures of movable devices provided in the radiation therapy device 10 or the X-ray CT device 20. For example, as shown in FIG. 8, the failure prediction device 30 has an input interface 31, a display 32, a memory circuit 33, a model memory circuit 34, and a processing circuit 35.

The input interface 31 accepts various input operations from the user, converts the accepted input operations into electrical signals, and outputs the electrical signals to the processing circuit 35. For example, the input interface 31 is realized by a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touchpad that performs input operations by touching the operation surface, a touch screen in which a display screen and a touchpad are integrated, a non-contact input circuit using an optical sensor, a voice input circuit, and the like. The input interface 31 may be configured as a tablet terminal or the like that can wirelessly communicate with the main body of the failure prediction device 30. The input interface 31 is not limited to only those that have physical operating parts such as a mouse and a keyboard. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the failure prediction device 30 and outputs the electrical signal to the processing circuit 35 is also included as an example of the input interface 31.

The display 32 displays various information. For example, under the control of the processing circuit 35, the display 32 displays a GUI for accepting various instructions and settings from the user via the input interface 31. Also, for example, the display 32 displays the results of failure prediction by the prediction function 352. For example, the display 32 is an LCD display or a CRT display. The display 32 may be a desktop type, or may be configured as a tablet terminal or the like capable of wireless communication with the failure prediction device 30 main body.

The memory circuit 33 and the model memory circuit 34 are realized, for example, by semiconductor memory elements such as RAM and flash memory, a hard disk, an optical disk, etc. For example, the memory circuit 33 stores a program for the circuits included in the failure prediction device 30 to realize their functions. Also, for example, the model memory circuit 34 records the trained model M1 or trained model M2 generated by the model generation function 351. Note that the memory circuit 33 and the model memory circuit 34 may be integrated.

The processing circuit 35 controls the operation of the entire failure prediction device 30 by executing the model generation function 351, the prediction function 352, and the output function 353. Here, the model generation function 351 is an example of a model generation unit. Also, the prediction function 352 is an example of a prediction unit. Also, the output function 353 is an example of an output unit.

For example, the processing circuitry 35 generates the trained model M1 or the trained model M2 by reading out a program corresponding to the model generation function 351 from the storage circuitry 33 and executing it. The model generation function 351 is the same function as the model generation function 152 or the model generation function 245b described above. As described above, the trained model M1 is an AI that is functionalized to predict failures related to the movable device provided in the radiation therapy device 10. As described above, the trained model M2 is an AI that is functionalized to predict failures related to the movable device provided in the X-ray CT device 20.

The processing circuitry 35 also reads out from the memory circuitry 33 a program corresponding to the prediction function 352 and executes it to predict failures related to the movable devices of the radiation therapy device 10 or the X-ray CT device 20 using the learned model M1 or the learned model M2. Note that the prediction function 352 is the same function as the prediction function 153 or the prediction function 245c described above.

The processing circuitry 35 also reads out from the storage circuitry 33 a program corresponding to the output function 353 and executes it to output the failure prediction result by the prediction function 352. The output function 353 is the same function as the output function 154 or the output function 245d described above.

In the failure prediction device 30 shown in FIG. 8, each processing function is stored in the memory circuitry 33 in the form of a program executable by a computer. The processing circuitry 35 is a processor that realizes the function corresponding to each program by reading the program from the memory circuitry 33 and executing it. In other words, when each program has been read, the processing circuitry 35 has the function corresponding to the read program.

In FIG. 8, the model generation function 351, prediction function 352, and output function 353 are described as being realized by a single processing circuit 35, but the processing circuit 35 may be configured by combining multiple independent processors, and each processor may execute a program to realize the functions.

Also, in FIG. 8, a single memory circuit 33 is described as storing a program corresponding to each processing function. However, the embodiment is not limited to this. For example, a configuration may be adopted in which a plurality of memory circuits 33 are distributed and the processing circuit 35 reads out the corresponding program from each individual memory circuit 33. Also, instead of storing the program in the memory circuit 33, the program may be directly built into the circuit of the processor. In this case, the processor realizes the function by reading and executing the program built into the circuit.

First, the failure prediction device 30 generates the trained model M1 or trained model M2 and stores it in the model storage circuit 34. For example, the model generation function 351 acquires training data used in the generation process of the trained model M1 from the radiation therapy device 10. As an example, the model generation function 351 acquires the sensor information, auxiliary information, failure history information, etc. shown in FIG. 4 as training data via the network NW. The model generation function 351 also generates a training data set based on the training data acquired from the radiation therapy device 10. Then, the model generation function 351 executes machine learning based on the generated training data set to generate the trained model M1. Similarly, the model generation function 351 can also acquire training data used in the generation process of the trained model M2 from the X-ray CT device 20 and generate the trained model M2. The model generation function 351 also stores the generated trained model M1 or trained model M2 in the model storage circuit 34.

Although the case where the model generation function 351 generates the trained model M1 or trained model M2 has been described, the embodiment is not limited to this. For example, instead of the processing circuit 35 executing the model generation function 351, the failure prediction device 30 may acquire the trained model M1 or trained model M2 generated in an external device and store it in the model storage circuit 34. As an example, the failure prediction device 30 may acquire the trained model M1 generated in the radiation therapy device 10 via the network NW and store it in the model storage circuit 34. As another example, the failure prediction device 30 may acquire the trained model M2 generated in the X-ray CT device 20 via the network NW and store it in the model storage circuit 34.

Next, the failure prediction device 30 predicts a failure related to the movable device equipped in the radiation therapy device 10 or the X-ray CT device 20. For example, the prediction function 352 predicts a failure related to the movable device equipped in the radiation therapy device 10 using the trained model M1. As an example, the prediction function 352 acquires sensor information and additional information from the radiation therapy device 10 via the network NW each time radiation therapy is performed in the radiation therapy device 10. Next, the prediction function 352 reads out the trained model M1 from the model storage circuit 34 and inputs the acquired sensor information and additional information to the trained model M1. Here, the trained model M1 predicts a failure related to the movable device based on these input data. That is, the prediction function 153 can predict a failure related to the movable device equipped in the radiation therapy device 10 by acquiring the sensor information and additional information from the radiation therapy device 10 and inputting them to the trained model M1. Similarly, the prediction function 153 can predict failures related to movable devices equipped in the X-ray CT scanner 20 by acquiring sensor information and auxiliary information from the X-ray CT scanner 20 and inputting the information into the trained model M2.

Then, the output function 353 outputs the failure prediction result obtained by the prediction function 352. For example, the output function 353 causes the failure prediction result obtained by the prediction function 352 to be displayed on the display 32. Alternatively, the output function 353 may transmit the failure prediction result obtained by the prediction function 352 to another device via the network NW, and the other device may display the failure prediction result obtained by the prediction function 352.

For example, when the prediction function 352 predicts a failure of a movable device included in the radiation therapy device 10, the output function 353 transmits the failure prediction result by the prediction function 352 to the radiation therapy device 10 via the network NW. In this case, the output function 154 in the radiation therapy device 10 can display the failure prediction result by the prediction function 352 on the display 12.

For example, when the prediction function 352 predicts a failure of a movable device equipped in the X-ray CT device 20, the output function 353 transmits the failure prediction result by the prediction function 352 to the X-ray CT device 20 via the network NW. In this case, the output function 245d in the X-ray CT device 20 can display the failure prediction result by the prediction function 352 on the display 242.

Note that while FIG. 8 shows the radiation therapy device 10 and the X-ray CT device 20, one of these radiation devices may be omitted. In other words, the failure prediction device 30 may be connected to only one of the radiation therapy device 10 and the X-ray CT device 20.

Although only one radiation therapy device 10 is shown in FIG. 8, the failure prediction device 30 may be connected to multiple radiation therapy devices 10. Similarly, the failure prediction device 30 may be connected to multiple X-ray CT devices 20.

7 and 8, the X-ray CT device 20 has been described as an example of a radiation diagnostic device capable of performing image collection. However, the embodiment is not limited to this. For example, in various radiation diagnostic devices such as a PET (Positron Emission Tomography) device, a SPECT (Single Photon Emission Computed Tomography) device, and an X-ray diagnostic device, it is possible to predict failures related to movable devices as in the case of the X-ray CT device 20. Furthermore, the failure prediction device 30 is capable of predicting failures related to movable devices provided in various radiation diagnostic devices such as a PET device, a SPECT device, and an X-ray diagnostic device. The failure prediction device 30 may be connected to multiple types of radiation diagnostic devices.

The term "processor" used in the above description refers to circuits such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an Application Specific Integrated Circuit (ASIC), a programmable logic device (e.g., a Simple Programmable Logic Device (SPLD), a Complex Programmable Logic Device (CPLD), and a Field Programmable Gate Array (FPGA)). If the processor is, for example, a CPU, the processor realizes the function by reading and executing a program stored in a memory circuit. On the other hand, if the processor is, for example, an ASIC, instead of storing a program in a memory circuit, the function is directly incorporated as a logic circuit in the circuit of the processor. Note that each processor in the embodiments is not limited to being configured as a single circuit for each processor, but may be configured as a single processor by combining multiple independent circuits to realize the function. Furthermore, multiple components in each figure may be integrated into a single processor to realize the function.

The components of each device according to the above-described embodiments are conceptual and functional, and do not necessarily need to be physically configured as shown in the figures. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figures, and all or part of the devices can be functionally or physically distributed and integrated in any unit depending on various loads and usage conditions. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic.

The failure prediction method described in the above embodiment can be realized by executing a prepared failure prediction program on a computer such as a personal computer or a workstation. This failure prediction program can be distributed via a network such as the Internet. This failure prediction program can also be recorded on a non-transitory recording medium that can be read by a computer, such as a hard disk, flexible disk (FD), CD-ROM, MO, or DVD, and executed by being read from the recording medium by a computer.

According to at least one of the embodiments described above, it is possible to avoid a system downtime of the radiation device.

Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, modifications, and combinations of embodiments can be made without departing from the spirit of the invention. These embodiments and their modifications are within the scope of the invention and its equivalents as set forth in the claims, as well as the scope and spirit of the invention.

REFERENCE SIGNS LIST 1 Radiation system 10 Radiation therapy device 14 Model storage circuit 15 Processing circuit 151 Control function 152 Model generation function 153 Prediction function 154 Output function 16 Bed 161 Top plate 162 Bed drive device 17 Radiation generator 171 Radiation generator 172 Radiation aperture 20 X-ray CT device 210 Stand device 211 X-ray tube 212 X-ray detector 213 Rotating frame 214 X-ray high voltage device 215 Control device 216 Wedge device 2161 Wedge 2162 Wedge drive device 217 Collimator 2171 Aperture blade 2172 Aperture blade drive device 218 DAS
230 Bed 231 Base 232 Bed drive device 233 Top plate 234 Support frame 240 Console device 244 Model storage circuit 245 Processing circuit 245a Control function 245b Model generation function 245c Prediction function 245d Output function 30 Fault prediction device 34 Model storage circuit 35 Processing circuit 351 Model generation function 352 Prediction function 353 Output function

Claims (19)

A movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part;
A control unit that controls the movable device to perform radiation therapy or image acquisition;
A model storage unit that stores a trained model configured to predict failure of the movable device;
a prediction unit that predicts a failure of the movable device by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The control unit records image information based on image data of the movable part and surgical procedure information indicating the surgical procedure used when performing the radiation therapy or the image collection in a recording unit,
The prediction unit inputs the image information recorded in the recording unit as the first information into the trained model, and inputs the operative procedure information as the second information into the trained model. A radiation apparatus. A movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part;
A control unit that controls the movable device to perform radiation therapy or image acquisition;
A model storage unit that stores a trained model configured to predict failure of the movable device;
a prediction unit that predicts a failure of the movable device by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The control unit records image information based on image data of the movable part and radiation information indicating the amount of radiation used when performing the radiation therapy or the image collection in a recording unit,
A radiation device, wherein the prediction unit inputs the image information recorded in the recording unit as the first information into the trained model, and inputs the radiation information as the second information into the trained model. The radiation device according to claim 2, wherein the control unit further records, in a recording unit, surgical procedure information indicating a surgical procedure used when performing the radiation therapy or the image acquisition, and the prediction unit further inputs the surgical procedure information into the trained model as the second information. The control unit further records tilt information indicating a tilt of the movable unit with respect to a gravity direction in a recording unit,
The radiation device according to claim 1 , wherein the prediction unit further inputs the gradient information as the second information into the trained model. The radiation device according to any one of claims 1 to 4, wherein the prediction unit predicts which of the multiple movable devices will have a malfunction, or predicts a location of the movable device where a malfunction will occur. The radiation device according to claim 5, wherein the prediction unit further predicts when a failure of the movable device will occur. A movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part;
A control unit that controls the movable device to perform radiation therapy or image acquisition;
A model storage unit that stores a trained model that is functionalized to predict a failure of a sensor that detects the operation of the movable part;
a prediction unit that predicts a failure of the sensor by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The control unit records, in a recording unit, difference data indicating a difference between the detection results of the movement of the movable unit detected by the plurality of sensors and radiation information indicating the amount of radiation used when performing the radiation therapy or the image acquisition;
A radiation device, wherein the prediction unit inputs the difference data recorded in the recording unit into the trained model as the first information, and inputs the radiation information into the trained model as the second information. A movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part;
A control unit that controls the movable device to perform radiation therapy or image acquisition;
A model storage unit that stores a trained model that is functionalized to predict a failure of a sensor that detects the operation of the movable part;
a prediction unit that predicts a failure of the sensor by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The control unit records, in a recording unit, difference data indicating a difference between the detection results of the plurality of sensors, the detection results being obtained by detecting the operation of the movable unit by the plurality of sensors;
The prediction unit predicts, as a sensor failure, one of the multiple sensors that detect the operation of the movable part, by inputting the difference data recorded in the recording unit into the trained model as the first information. The radiation device according to claim 7, wherein the prediction unit predicts, as a sensor failure, a sensor that will fail among the plurality of sensors that detect the operation of the movable part. The radiation device according to claim 8 or 9, wherein the prediction unit further predicts when the sensor will fail. The radiation device according to any one of claims 7 to 10, wherein the difference data is data indicating the difference in detection results between a plurality of sensors of different types. The radiation device according to any one of claims 1 to 11, wherein the movable part performs a movement or a rotation as the mechanical movement. The radiation device according to any one of claims 1 to 12, further comprising a model generation unit that generates the trained model and stores it in the model storage unit. Further comprising a model generation unit that generates the trained model and stores it in the model storage unit;
The radiation device according to any one of claims 1 to 6, wherein the model generation unit generates the trained model using the time when a failure occurred in the movable device as output side teacher data. Further comprising a model generation unit that generates the trained model and stores it in the model storage unit;
The radiation device according to any one of claims 7 to 11, wherein the model generation unit generates the trained model by using the time when the sensor failure occurred as output side teacher data. A failure prediction device connected to a radiation device including a movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part, and a control unit for controlling the movable device to perform radiation therapy or image acquisition,
A model storage unit that stores a trained model configured to predict failure of the movable device;
a prediction unit that predicts a failure of the movable device by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The prediction unit inputs image information based on image data of the moving part into the trained model as the first information, and inputs surgical procedure information indicating the surgical procedure used when performing the radiation therapy or the image collection into the trained model as the second information. A failure prediction device connected to a radiation device including a movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part, and a control unit for controlling the movable device to perform radiation therapy or image acquisition,
A model storage unit that stores a trained model configured to predict failure of the movable device;
a prediction unit that predicts a failure of the movable device by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
A failure prediction device in which the prediction unit inputs image information based on image data of the moving part into the trained model as the first information, and inputs radiation information indicating the amount of radiation used when performing the radiation therapy or the image collection into the trained model as the second information. A failure prediction device connected to a radiation device including a movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part, and a control unit for controlling the movable device to perform radiation therapy or image acquisition,
A model storage unit that stores a trained model that is functionalized to predict a failure of a sensor that detects the operation of the movable part;
a prediction unit that predicts a failure of the sensor by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The prediction unit inputs difference data indicating differences between the detection results of the operation of the movable part detected by the multiple sensors into the trained model as the first information, and inputs radiation information indicating the amount of radiation used when performing the radiation therapy or the image collection into the trained model as the second information. A failure prediction device connected to a radiation device including a movable device including a movable part capable of performing a mechanical operation and a drive device for operating the movable part, and a control unit for controlling the movable device to perform radiation therapy or image acquisition,
A model storage unit that stores a trained model that is functionalized to predict a failure of a sensor that detects the operation of the movable part;
a prediction unit that predicts a failure of the sensor by inputting first information regarding the operation of the movable unit and second information indicating an environment when the movable unit operates into the trained model;
an output unit that outputs a failure prediction result by the prediction unit,
The prediction unit predicts which sensor among the multiple sensors that detect the operation of the moving part will fail as a sensor failure by inputting difference data indicating differences between the detection results of the multiple sensors detecting the operation of the moving part as the first information into the trained model.

Images