JP5488229B2 - Positioning determination device, image processing device, and program - Google Patents

Description

  The present invention relates to a positioning determination apparatus, an image processing apparatus, and a program that determine whether or not a captured image can be positioned.

  In the past, many radiation images including X-ray images have been taken in medical facilities such as hospitals. These radiographic images are used, for example, when diagnosing diseases by detecting lesion tissues and abnormalities in various parts of the human body such as the lungs and head in a non-erosive manner.

  In capturing these radiographic images, it is difficult to directly confirm the site to be imaged before imaging, and a caregiver cannot accompany. Therefore, due to individual differences and movements such as the posture and physique of the person to be imaged, a situation in which the region to be diagnosed is not normally imaged occurs.

  In view of this, there has conventionally been an apparatus that reads digital data of a captured radiographic image and determines whether or not the imaging position is appropriate. For example, Patent Document 1 discloses a technique for detecting a relative position of a human body or a region to be imaged from a captured image, and determining whether or not an image has been captured at an appropriate position and issuing a warning. By incorporating such a determination system into the photographing apparatus, the risk that proper photographing is not performed and that another photographing must be performed at a later date is reduced.

Japanese Patent Laid-Open No. 4-364834

  However, there are individual differences in the size and shape of the radiographic image, and the relative or overall position based on the person's body shape is not accurate enough to determine the radiographic image. There is a problem in that the end of the part is often missing from the image.

  An object of the present invention is to provide a positioning determination device, an image processing device, and a program that can appropriately determine an image in which a part of an imaging region is missing due to a positioning error.

In order to solve the above problem, the invention according to claim 1
Positioning determination means for determining whether or not a radiographic image obtained by irradiating a specific part of a human body is an image captured by appropriate positioning;
A predetermined feature amount is extracted from image data of a partial region in contact with the boundary of the irradiation field within the irradiation field of the radiation image, and a part of the specific part is determined based on a learning result regarding the feature amount by a predetermined learning algorithm. Image determining means for determining whether or not the radiation image is missing;
With
The positioning determination means includes
The positioning determination apparatus according to claim 1, wherein the determination result of the image determination unit is included in a reference for determining whether or not the radiographic image is an image captured by appropriate positioning.

The invention according to claim 2 is the positioning determination apparatus according to claim 1,
The learning algorithm is an Adaboost or support vector machine.

The invention according to claim 3 is the positioning determination apparatus according to claim 1,
The radiographic image is a chest front image,
The specific site is a lung field.

The invention according to claim 4 is the positioning determination apparatus according to claim 1,
The image determination means
The partial region is divided into a predetermined number, and a value indicating the maximum radiation dose for each of the divided regions is used as the feature amount.

The invention according to claim 5
The positioning determination device according to any one of claims 1 to 4,
An image display means for displaying a determination result of the image determination means provided in the positioning determination apparatus, and a determination result of the positioning determination apparatus based on the determination result;
An image processing apparatus comprising:

The invention described in claim 6
Computer
Positioning determination means for determining whether or not a radiographic image obtained by irradiating a specific part of the human body is an image taken by appropriate positioning,
A predetermined feature amount is extracted from image data of a partial region in contact with the boundary of the radiation field within the radiation field of the radiation image, and the specific part is the radiation image based on a learning result regarding the feature amount by a predetermined learning algorithm. Image determining means for determining whether or not the image is missing,
Is a program for functioning as
The positioning determination means includes
The determination result of the image determination unit is included in a criterion for determining whether or not the radiographic image is an image taken by appropriate positioning.

  The positioning determination apparatus and the image processing apparatus according to the present invention include image determination means for determining whether or not the image is a part of a specific part of the human body that is missing based on a learning result obtained by a predetermined learning algorithm. Therefore, there is an effect that even a part of the missing image of the specific part can be appropriately determined.

It is a block diagram which shows the structure of the image processing apparatus of embodiment of this invention. It is a flowchart which shows the control procedure of an image process. It is a flowchart which shows the control procedure of a positioning determination process. It is the figure which showed the method of the positioning determination process concretely. It is a figure which shows the example of a display on the display part of an image processing apparatus.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus 1 according to the present embodiment.
The image processing apparatus 1 determines whether or not a specific part of the human body to be imaged is properly recorded without any defect in the acquired radiation image, and changes or saves the setting or classification of the image, or It is a device that performs disposal processing. The image processing apparatus 1 includes a control unit 11, a storage unit 12, an operation unit 13, a display unit 14, a communication unit 15, and the like, and these units are connected to each other via a bus 16.

  The control unit 11 provides the entire image processing apparatus 1 and a central processing unit (CPU) that performs control overall processing and arithmetic processing of each unit, and a RAM (temporary memory) that provides a working memory space to the CPU and stores temporary data. Random Access Memory). The CPU develops various execution programs 12a read from the storage unit 12 on the RAM and performs various processes. The CPU also receives signals and data from other components via the bus 16 and sends control signals and commands. The control unit 11 includes a determiner 11a that performs image determination processing.

  The storage unit 12 includes a storage device that stores and stores data. This storage device is, for example, a hard disk (HDD). Alternatively, the storage unit 12 may include other storage means such as an SSD (Solid State Drive) and a removable flash memory. The data stored in the storage unit 12 includes various programs 12a read by the CPU, setting data used when executing the programs, acquired image data, and image analysis data. The program 12a can be made available to the image processing apparatus 1 via a reading device using a recording medium such as a CD-ROM or DVD, or can be externally transmitted via a communication line by a carrier wave or the like. A form of obtaining and executing is also applicable.

  For example, the operation unit 13 includes a keyboard and a mouse. Alternatively, the operation unit 13 may include a touch panel and other pointing devices. The operation contents of these devices of the operation unit 13 performed by the user are converted into input signals and output to each unit of the image processing apparatus 1.

  The display unit 14 includes a display and displays an image. This display is, for example, a liquid crystal display (LCD). Alternatively, other display means such as an organic ELD (Electro-Luminescent Display) may be used for the display. The display unit 14 receives a control signal and a data signal from the control unit 11 and displays data such as a radiation image and related information on the screen.

  The communication unit 15 includes a communication interface or the like, and performs processing for communication between the image processing apparatus 1 and the outside. This communication interface is, for example, a network card. Data exchanged between the image processing apparatus 1 and the outside, for example, radiation image data captured by an external imaging apparatus, is input from the external network via the communication unit 15. On the other hand, for example, data acquired by various processes in the image processing apparatus 1 is stored in the storage unit 12 and is output to an external computer terminal via the communication unit 15.

  Next, the operation of the positioning determination process of this embodiment will be described.

  FIG. 2 is a flowchart showing a control procedure of image processing by the CPU. This image processing is started, for example, by an input signal based on a start instruction operation of the operation unit 13 by the user.

  When image processing is started, the CPU of the control unit 11 first reads image data from the storage unit 12 (step S21). In the present embodiment, the image data is stored in the storage unit 12 after initial correction such as adjustment of brightness and contrast after being captured by the radiation image capturing apparatus.

  Next, the CPU performs a positioning determination process for a specific part of the human body that is a subject photographed in the read image data (step S22). The positioning determination process by the CPU (control unit 11) constitutes a positioning determination unit, which will be described in detail later.

  Subsequently, the CPU causes the display unit 14 to display the image data together with the image related information and the operation menu (step S23). Further, the CPU performs a determination process as to whether or not the arrangement of the imaging region in the image is defective based on the result of the positioning determination process in step S22 (step S24), and determines that the result of the positioning determination is defective. If so, the fact is displayed on the image displayed on the display unit 14 and in the notification area (step S25). Then, the processing of the CPU proceeds to step S26. If it is determined that the positioning determination result is not defective, the CPU does not perform additional display on the display unit 14, and the process of the CPU proceeds directly to step S26.

  The CPU displays two types of buttons for the image on the display unit 14. That is, the CPU displays an image normality confirmation button “OK” by the user and an image failure confirmation button “NG” by the user on the display unit 14 and waits until any of the buttons is operated by the user. (Step S26).

  If it is determined that an input signal from the operation unit 13 indicating that the normal confirmation button “OK” has been operated is detected, the CPU sets a flag “OK” to the image (step S27). Then, the image determination process ends. On the other hand, if it is determined that an input signal from the operation unit 13 indicating that the failure confirmation button “NG” has been operated is detected, the CPU sets a flag “NG” to the image (step S28). Then, the image determination process ends.

  Here, in the above control procedure, the CPU only sets the flag “OK” or “NG” in the image. For example, the image data in which the flag “NG” is set is moved to a dedicated directory. Alternatively, the image data may be erased.

  Next, the positioning determination process (step S22) will be described in detail.

  FIG. 3 is a flowchart illustrating the procedure of the positioning determination process. FIG. 4 is a diagram specifically showing an example of the positioning determination process.

  When the positioning determination process is called, the CPU first performs a process of recognizing the range of the irradiation field in the image (step S221). In this processing, the CPU detects a portion where the radiation has not reached from the radiation source to the imaging surface by masking during imaging, and sets the remaining portion excluding the detected portion from the image as an irradiation field. . That is, as shown in FIG. 4 (a), the boundary line B1 indicated by the broken line is detected from the data of the chest front image indicated by the practice frame, excluding the portion not imaged by masking. The inside of the boundary line B1 is set as the irradiation field range. This irradiation field setting process can be performed, for example, by a method of detecting and excluding an area having a maximum brightness (that is, a minimum radiation dose) continuously inward from the contour of the image.

  Next, the CPU divides the identified irradiation field into a plurality of small areas. Then, the CPU performs a process of extracting a predetermined feature amount from each of the small areas included in the partial area that is a predetermined range near the boundary of the irradiation field among the plurality of divided small areas (step) S222). Specifically, as shown in FIG. 4B, the range of the irradiation field in the captured image data is not particularly limited. For example, the irradiation field range is a small area arranged in a 10 × 10 matrix. Divided. Then, feature amounts are extracted from each of the small areas included in the partial areas in two rows from the upper and lower boundary lines of the irradiation field and in two columns from the left and right boundary lines.

  Here, by dividing the partial region in contact with the boundary line into a plurality of rows or columns instead of only one row or one column, the image determination can be performed accurately even in the case of a complex image pattern. For example, when gas is reflected in the lower part of the lower end of the lung field, water is accumulated at the lower end of the lung field, or a catheter is inserted near the lower end of the lung field. Also in this case, the determination can be appropriately performed by using the determination device 11a that has been learned based on the data of the small regions of a plurality of rows and a plurality of columns.

  The feature amount extracted in the process of step S222 is, for example, the minimum value (that is, the measured radiation intensity having the maximum measured value) among the luminance values of each pixel in each small region. In addition, as other feature amounts, the maximum value of the first derivative of the luminance value in each small region, and in particular, the maximum component in each small region of the component of the first derivative that is parallel to the boundary of the irradiation field. A value and a coordinate value that takes the maximum value of the primary differential amount can be used. Alternatively, a power spectrum of luminance values can be used as the feature amount. In particular, a power spectral density in a direction parallel to the boundary line of the irradiation field and a frequency (wavelength) value that takes a maximum value can be used.

  When the above feature amount extraction is performed, the CPU performs an image determination process for each boundary line using these feature amounts and their arrangement (step S223). The image determination process performed by the CPU (control unit 11) determines whether or not a subject is missing from each boundary line, and constitutes an image determination unit. In the example of FIG. 4B, feature amounts are obtained from 2 × 10 small areas along the boundary lines at the upper end, the lower end, the left end, and the right end, and these data are obtained from the determiner 11a of the control unit 11. To determine whether there is a part where the lung field protrudes from each boundary line and is not photographed.

  The determinator 11a used for this image discrimination process is learned in advance using a predetermined learning algorithm. The learning algorithm used for learning by the determiner 11a is not particularly limited, but, for example, AdaBoost is used. Adaboost is optimal by preparing a large number of data arrays of a plurality of feature amounts extracted from images in which it is already known whether or not there is a subject missing, and combining weighted discriminators that determine subject missing. This is an algorithm for making a discriminator. As other learning algorithms, for example, a support vector machine or a neural network, or a K-nearest neighbor discriminator can be used. Statistical methods such as discriminant analysis may also be applied (reference: C. M. Bishop, “Pattern Recognition and Machine Learning”, Springer-Verlag, New Ed edition (2006)).

  In the learning by AdaBoost, a plurality of images each having a specific part missing and a plurality of images having no missing parts are prepared in advance, and a plurality of discriminators for judging the lack of a subject are prepared. As the plurality of discriminators, for example, a decision stamp that prepares a discriminator that performs discrimination based on whether or not a single feature amount exceeds a threshold value by changing the threshold value is used. Then, a discriminator effective for discrimination and its weighting amount are learned from a discrimination result of whether each of the plurality of feature amounts extracted from the image corresponds to an image having a missing portion or an image having no missing portion.

  On the other hand, the support vector machine maps data points indicated by a combination of feature quantities (feature vectors) to a high-dimensional space, and images with a specific part missing and images without a missing part in the mapped high-dimensional space. And a statistical algorithm for obtaining a separation plane that is linearly separated so as to have the maximum margin.

  The determiner 11a learned in advance by these algorithms is incorporated in the control unit 11 of the image processing apparatus 1 to perform a determination process regarding the presence or absence of a specific part.

  Here, the determination unit 11a based on the learning algorithm does not need to be used. For example, in the case of an image in which a clearly different part is photographed or an image with a different enlargement ratio, the CPU performs the conventional relative positioning determination. It is possible to determine that the image is not an image shot with an appropriate composition in combination.

  FIG. 5 shows an example in which the result of the positioning determination process is displayed on the display unit 14 of the image processing apparatus 1.

  FIG. 5A is a display example when it is determined by the positioning determination process that the image has a missing portion in the lung field as the subject. On the screen, three radiation images are called up and displayed on the image display blocks 131 to 133, respectively. In addition, on the screen, the identification information 134 of the patient whose radiographic image was taken, various operation menus, and an image processing determination notification message 135 are displayed. In each of the image display blocks 131 to 133, the called radiographic images are displayed in the upper areas 131a to 133a, and the lower areas 131b to 133b display information such as imaging regions and missing determination confirmation flags. Yes.

  In the image display blocks 131 to 132, in the lower areas 131b to 132b, a flag “OK” indicating that all of these images are chest images and that no error has been confirmed in these image data is displayed. Is displayed.

  On the other hand, in the image display block 133, an area 133b indicates that the image is a chest image, and a defect confirmation button 133d and a normal confirmation button 133e for the user to select the flag “NG” or “OK” are displayed. It is displayed. Further, in the area 133a, a warning sign 133c indicating that the position determination result is bad is displayed on the upper right together with the radiographic image of the chest. The image processing determination notification message 135 is given the same warning sign 135a corresponding to the warning sign 133c.

  The image processing determination notification message 135 is also displayed with a display 135b indicating that the image displayed in the upper area 133a may be a missing lung field due to the image processing determination. The image processing determination notification message 135 also displays an enlarge button 135c for performing an operation for enlarging the corresponding part at the same time.

  FIG. 5B is a diagram showing display contents displayed on the display screen of the display unit 14 when the enlargement button 135c is operated.

  In the upper area 133a, a part of the lung field determined to be missing is enlarged and displayed. Thus, based on the enlarged image displayed together with the image processing determination notification message 135, an operator (for example, a doctor or a laboratory technician) can select either the defect confirmation button 133d or the normal confirmation button 133e in the lower area 133b. For example, the selection is made by clicking with a mouse operation, and finally, determination is made as to whether this image data is missing data in the lung field or not.

  As described above, according to the image processing apparatus 1 of the present embodiment, a recognition pattern is learned in the determiner 11a by a predetermined learning algorithm using a radiographic image in which the presence or absence of a specific part of a human body that is a subject is known in advance. Let me. Then, since it is determined whether or not this specific part of the human body is missing from the radiographic image using the learned determination device 11a, a missing image of a part of the subject that has been difficult with the conventional relative position detection. Even so, it can be determined appropriately.

  In addition, by making a determination based on a learning result by a learning algorithm, it is possible to accurately determine whether or not a part of a subject is missing even in an image having a complicated pattern such as a lesion tissue, an abnormality, or a trace of surgery. be able to.

  In addition, by using this determiner 11a in combination with a conventional relative positioning determination device, it is possible to detect a large misalignment to a small misalignment with high accuracy without taking excessive effort for learning of the determiner 11a. can do.

  In addition, by using such a positioning determination process for a medical image such as a front image of a chest with a lung field as a subject, for example, it is possible to suppress a situation in which the patient is forced to re-photograph.

  In particular, it is possible to prevent the subject from being lost from the image by causing the determination unit 11a that has learned the image data near the boundary of the irradiation field to learn using the learning algorithm.

  The present invention is not limited to the above-described embodiment, and various modifications can be made. For example, in the above embodiment, the lung field image of the chest has been described as an example. However, for example, it can be used for imaging other parts such as the head and abdomen. In addition, it is possible to deal with a situation in which images of these plural parts are mixed.

  In addition, the positioning determination device or the image processing device 1 is incorporated in the radiation imaging system, and it is determined whether or not the image positioning is appropriate for each imaging, and whether the imaging site is completely included in the image. Can be displayed. In this case, this image processing is called and executed every time radiographic imaging is performed and image data is sent. Thus, by performing positioning determination or image determination in real time in the radiation imaging system, the operator can immediately find a positioning error and a lack of an imaging region. Therefore, the operator can prevent a situation in which a necessary image cannot be obtained while omitting the trouble of checking all images carefully every time.

  Further, in the above embodiment, the determination of the absence of a subject is performed using a pre-learned determinator. However, if it is easy depending on the type of learning algorithm, the determination result is fed back on the spot to determine It is good also as comprising.

  In addition, the details shown in the embodiment of the present invention, such as the range of partial areas used for image determination processing and the number and arrangement of small areas, can be appropriately changed without departing from the spirit of the present invention.

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 11 Control part 12 Storage part 12a Program 13 Operation part 14 Display part 15 Communication part 16 Bus

Claims (6)

Positioning determination means for determining whether or not a radiographic image obtained by irradiating a specific part of a human body is an image captured by appropriate positioning;
A predetermined feature amount is extracted from image data of a partial region in contact with the boundary of the irradiation field within the irradiation field of the radiation image, and a part of the specific part is determined based on a learning result regarding the feature amount by a predetermined learning algorithm. Image determining means for determining whether or not the radiation image is missing;
With
The positioning determination means includes
The positioning determination apparatus characterized in that the determination result of the image determination means is included in a reference for determining whether or not the radiographic image is an image taken by appropriate positioning. The positioning determination apparatus according to claim 1, wherein the learning algorithm is Adaboost or a support vector machine. The radiographic image is a chest front image,
The positioning determination apparatus according to claim 1, wherein the specific part is a lung field. The image determination means
The positioning determination apparatus according to claim 1, wherein the partial region is divided into a predetermined number, and a value indicating a maximum radiation dose for each of the divided regions is used as the feature amount. The positioning determination device according to any one of claims 1 to 4,
An image display means for displaying a determination result of the image determination means provided in the positioning determination apparatus, and a determination result of the positioning determination apparatus based on the determination result;
An image processing apparatus comprising: Computer
Positioning determination means for determining whether or not a radiographic image obtained by irradiating a specific part of the human body is an image taken by appropriate positioning,
A predetermined feature amount is extracted from image data of a partial region in contact with the boundary of the radiation field within the radiation field of the radiation image, and the specific part is the radiation image based on a learning result regarding the feature amount by a predetermined learning algorithm. Image determining means for determining whether or not the image is missing,
Is a program for functioning as
The positioning determination means includes
A program characterized in that the determination result of the image determination means is included in a criterion for determining whether or not the radiation image is an image photographed by appropriate positioning.

Images