COMPUTER PROGRAM, INFORMATION PROCESSING APPARATUS, AND INFORMATION PROCESSING METHOD

Abstract

To provide a computer program capable of improving the detection accuracy of a narrow section, an information processing device and an information processing method.SOLUTION: A computer program makes a computer execute processing for acquiring medical image data showing a cross-sectional image of a hollow organ, inputting the acquired medical image data to a learning model for outputting position data of a prescribed portion of the hollow organ to acquire first position data of the prescribed portion in the case of inputting the medical image data showing the cross-sectional image of the hollow organ, specifying an index of the hollow organ on the basis of the acquired first position data, generating medical image data obtained by relatively expanding a region including the prescribed position in the cross-sectional data of the acquired medical image data more than a prescribed size to be inputted to the learning model if the specified index satisfies a prescribed condition, inputting the generated medical image data to the learning model to acquire second position data of the prescribed position, and calculating the internal diameter of the hollow organ on the basis of the acquired second position data.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a computer program, an information processing device, and an information processing method. [Background technology]

[0002] Interpretation of IVUS (Intra Vascular Ultrasound) images, which contain many artifacts, is generally considered difficult, and the use of object detection technology to assist in the interpretation of these images is being considered.

[0003] Patent Document 1 discloses an imaging diagnostic device that inserts a catheter into a blood vessel and generates a cross-sectional image of the vessel based on signals (ultrasound waves emitted toward the vessel tissue and reflected waves) obtained by an imaging core housed in the catheter. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2017 / 164071 Summary of the Invention [Problem to be solved by the invention]

[0005] Areas with small vascular lumen diameters, especially the minimum lumen area (MLA), are typical areas of interest for identifying vascular stenosis. It is necessary to detect where this area is located within the blood vessel and what the lumen diameter is, but there are cases where the accuracy of detecting stenosis is insufficient.

[0006] The present invention has been made in view of the above circumstances, and has an object to provide a computer program, an information processing device, and an information processing method that can improve the accuracy of detecting a stenosis. [Means for solving the problem]

[0007] The present application includes multiple means for solving the above-mentioned problems. As an example, a computer program causes a computer to execute a process of acquiring medical image data showing a cross-sectional image of a tubular organ, inputting the acquired medical image data into a learning model that outputs position data of a specific portion of the tubular organ when the medical image data showing the cross-sectional image of the tubular organ is input, acquiring first position data of the specific portion, identifying an index of the tubular organ based on the acquired first position data, and if the identified index satisfies a specific condition, generating medical image data in which an area including the specific portion in the cross-sectional image of the acquired medical image data is relatively enlarged compared to a specific size input into the learning model, inputting the generated medical image data into the learning model to acquire second position data of the specific portion, and calculating the inner diameter of the tubular organ based on the acquired second position data. Effect of the Invention

[0008] According to the present invention, it is possible to improve the accuracy of detecting a stenosis. [Brief description of the drawings]

[0009] [Figure 1] 1 is a diagram showing an example of the configuration of an image diagnostic system according to an embodiment of the present invention. [Diagram 2] FIG. 1 illustrates an example of a configuration of an information processing device. [Diagram 3] FIG. 2 is a diagram illustrating an example of a configuration of a learning model. [Figure 4] FIG. 13 is a diagram showing an example of an estimation result based on a learning model. [Diagram 5] FIG. 13 is a diagram showing an example of the agreement between an estimated value by a learning model and a true value. [Figure 6] FIG. 13 is a diagram illustrating an example of an error rate of an estimated value by a learning model. [Figure 7] 11 is a diagram showing an example of calculation of the inner diameter and the outer diameter of a hollow organ by an information processing device. FIG. [Figure 8] FIG. 1 is a diagram showing a first example of a transition of the average lumen diameter along the longitudinal axis of a blood vessel. [Figure 9]FIG. 1 is a diagram illustrating a first example of medical image generation processing by an information processing device. [Figure 10] FIG. 11 is a diagram showing an example of enlargement / reduction processing for medical image data. [Figure 11] FIG. 11 is a diagram illustrating a second example of medical image generation processing by the information processing device. [Figure 12] FIG. 13 is a diagram showing a second example of a transition of the average lumen diameter along the longitudinal axis of a blood vessel. [Figure 13] An example of a method for determining the presence or absence of a side canal based on eccentricity will be described. [Figure 14] FIG. 1 is a diagram illustrating an example of real-time processing by an information processing device. [Figure 15] FIG. 1 illustrates an example of batch processing by an information processing device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described. FIG. 1 is a diagram showing an example of the configuration of an image diagnostic system 100 according to the present embodiment. The image diagnostic system 100 is an apparatus for performing intravascular imaging (image diagnosis) used in intravascular treatment such as peripheral coronary intervention (PCI). Cardiac catheter treatment is a method of treating a narrowed portion of a coronary artery by inserting a catheter into a blood vessel in the groin, arm, wrist, etc. There are two methods for intravascular imaging: intravascular ultrasound (IVUS) and optical frequency domain imaging (OFDI, OCT). IVUS uses the reflection of ultrasound to interpret the inside of a blood vessel in a tomographic image. Specifically, a thin catheter equipped with an ultra-small sensor at its tip is inserted into a coronary artery and passed to the affected area, and then a medical image of the inside of the blood vessel can be generated by ultrasound emitted from the sensor. OFDI uses near-infrared rays to interpret the state inside the blood vessel in a high-resolution image. Specifically, like IVUS, a catheter is inserted into a blood vessel, near-infrared light is irradiated from the tip, and the cross section of the blood vessel is measured by interference to generate a medical image. OCT is an intravascular image diagnosis that applies near-infrared light and optical fiber technology. In this specification, medical images (medical image data) include those generated by IVUS, OFDI, or OCT, but the following description will mainly focus on the case where the IVUS method is used.

[0011] The image diagnostic system 100 includes a catheter 10, an MDU (Motor Drive Unit) 20, a display device 30, an input device 40, and an information processing device 50. A server 200 is connected to the information processing device 50 via a communication network 1.

[0012] The catheter 10 is an imaging diagnostic catheter for obtaining ultrasonic tomographic images of blood vessels by the IVUS method. The catheter 10 has an ultrasonic probe at its tip for obtaining ultrasonic tomographic images of blood vessels. The ultrasonic probe has an ultrasonic transducer that emits ultrasonic waves within the blood vessel, and an ultrasonic sensor that receives reflected waves (ultrasonic echoes) reflected by structures such as biological tissue of the blood vessel or medical equipment. The ultrasonic probe is configured to be capable of moving forward and backward in the longitudinal direction of the blood vessel while rotating in the circumferential direction of the blood vessel.

[0013] The MDU 20 is a drive unit to which the catheter 10 can be detachably attached, and drives a built-in motor in response to the operation of a medical professional to control the operation of the catheter 10 inserted into a blood vessel. The MDU 20 can rotate the ultrasound probe of the catheter 10 in a circumferential direction while moving it from the tip (distal) side to the base (proximal) side (pull-back operation). The ultrasound probe continuously scans the inside of the blood vessel at a predetermined time interval, and outputs reflected wave data of the detected ultrasound to the information processing device 50.

[0014] The information processing device 50 generates (acquires) multiple chronologically ordered medical images including tomographic images of blood vessels based on the reflected wave data output from the ultrasonic probe of the catheter 10. Since the ultrasonic probe scans the inside of the blood vessel while moving from the tip (distal) side to the base (proximal) side inside the blood vessel, the multiple chronologically ordered medical images are tomographic images of the blood vessels observed at multiple points from the distal to proximal ends.

[0015] The display device 30 includes a liquid crystal display panel, an organic EL display panel, or the like, and can display the results of processing by the information processing device 50. The display device 30 can also display medical images generated (acquired) by the information processing device 50.

[0016] The input device 40 is an input interface such as a keyboard or a mouse that accepts input of various setting values ​​when performing an inspection and operations of the information processing device 50. The input device 40 may be a touch panel, soft keys, hard keys, etc. provided on the display device 30.

[0017] The server 200 is, for example, a data server, and may include an image DB that stores medical image data.

[0018] 2 is a diagram showing an example of the configuration of the information processing device 50. The information processing device 50 can be configured by a computer, and includes a control unit 51 that controls the entire information processing device 50, a communication unit 52, an interface unit 53, a recording medium reading unit 54, a memory 55, and a storage unit 56.

[0019] The control unit 51 is configured by incorporating a required number of central processing units (CPUs), micro-processing units (MPUs), graphics processing units (GPUs), general-purpose computing on graphics processing units (GPGPUs), tensor processing units (TPUs), etc. The control unit 51 may also be configured by combining digital signal processors (DSPs), field-programmable gate arrays (FPGAs), quantum processors, etc. The control unit 51 has the functions of a first acquisition unit, a second acquisition unit, a third acquisition unit, an identification unit, a generation unit, and a calculation unit, and also has a function realized by a computer program 57 described later.

[0020] The memory 55 can be configured with a semiconductor memory such as a static random access memory (SRAM), a dynamic random access memory (DRAM), or a flash memory.

[0021] The communication unit 52 includes, for example, a communication module, and has a communication function with the server 200 via the communication network 1. The communication unit 52 may also have a communication function with an external device (not shown) connected to the communication network 1.

[0022] The interface unit 53 provides an interface function between the catheter 10, the display device 30, and the input device 40. The information processing device 50 (control unit 51) can transmit and receive data and information between the catheter 10, the display device 30, and the input device 40 through the interface unit 53.

[0023] The recording medium reading unit 54 can be configured, for example, by an optical disk drive, and can read a computer program (program product) recorded on a recording medium 541 (for example, an optically readable disk storage medium such as a CD-ROM) by the recording medium reading unit 54 and store it in the storage unit 56. The computer program 57 is developed in the memory 55 and executed by the control unit 51. The computer program 57 may be downloaded from an external device via the communication unit 52 and stored in the storage unit 56.

[0024] The storage unit 56 can be configured with, for example, a hard disk or a semiconductor memory, and can store required information. The storage unit 56 can store a learning model 58 in addition to the computer program 57. The learning model 58 includes a model before learning, a model in the middle of learning, or a learned model.

[0025] FIG. 3 is a diagram showing an example of the configuration of the learning model 58. The learning model 58 includes an input layer 58a, an intermediate layer 58b, and an output layer 58c, and can be configured, for example, by U-Net. The intermediate layer 58b includes a plurality of encoders and a plurality of decoders. The plurality of encoders repeatedly perform convolution processing on the medical image data input to the input layer 58a. The plurality of decoders repeatedly perform upsampling (deconvolution) processing on the image convolved by the encoder. When decoding the convolved image, a process is performed in which the feature map generated by the encoder is added to the image to be deconvolved. This makes it possible to retain position information that is lost by the convolution processing, and to output a more accurate segmentation (which pixel belongs to which class).

[0026] When medical image data is input, the learning model 58 can output position data indicating each of the lumen boundary and the blood vessel boundary. The position data is coordinate data of pixels indicating each of the lumen boundary and the blood vessel boundary. In other words, the learning model 58 can classify each pixel of the input medical image data into three classes, for example, classes 1, 2, and 3. Class 1 indicates Background and indicates the outside area of ​​the blood vessel. Class 2 indicates (Plaque + Media) and indicates the area of ​​the blood vessel including plaque. Class 3 indicates Lumen and indicates the lumen of the blood vessel. Therefore, the boundary between the pixel classified into class 2 and the pixel classified into class 3 indicates the lumen boundary, and the boundary between the pixel classified into class 1 and the pixel classified into class 2 indicates the blood vessel boundary. Note that the learning model 58 is not limited to U-Net, and may be, for example, GAN (Generative Adversarial Network), SegNet, etc.

[0027] The learning model 58 may be generated as follows. First, first training data including medical image data showing a cross-sectional image of a blood vessel and position data of the lumen boundary of the blood vessel and the blood vessel boundary of each of the blood vessels is acquired. The first training data may be collected and stored in the server 200, for example, and acquired from the server 200. Next, based on the first training data, the learning model 58 may be generated so that when medical image data showing a cross-sectional image of a blood vessel is input to the learning model 58, the learning model 58 outputs position data of the lumen boundary and the blood vessel boundary of each of the blood vessels. The first training data may include both medical image data in which a side branch is present in the cross-sectional image of the blood vessel and medical image data in which a side branch is not present in the cross-sectional image of the blood vessel.

[0028] By inputting multiple medical image data obtained along the long axis direction of the blood vessel by a pullback operation into the learning model 58 generated as described above and examining the estimation results of the learning model 58, the applicant has discovered that, from the standpoint of detection accuracy, the error rate of the blood vessel diameter is larger when the blood vessel diameter, which is one of the detection targets, is small compared to when the blood vessel diameter is not so small.

[0029] FIG. 4 is a diagram showing an example of an estimation result by the learning model 58. The example shown in FIG. 4 shows an example of an erroneous determination in a stenotic part of a blood vessel. FIG. 4A shows a medical image input to the learning model 58, FIG. 4B shows an estimated image by the learning model 58, and FIG. 4C shows a correct image. Comparing FIG. 4B and FIG. 4C, it can be seen that the lumen boundary is erroneously estimated especially when the lumen diameter of the blood vessel is small such as in a stenotic part. Note that when the lumen diameter is not very small, it is possible to extremely reduce the probability of erroneous estimation.

[0030] FIG. 5 is a diagram showing an example of the agreement between the estimated value by the learning model 58 and the true value. FIG. 5 shows a Bland-Altman plot used to evaluate the agreement between two measurement methods. In FIG. 5, the horizontal axis shows the average (mm) between the true value and the estimated value of the average lumen diameter, and the vertical axis shows the difference (mm) between the true value and the estimated value of the average lumen diameter. The average lumen diameter can be calculated by the average lumen diameter D=2×√(S / π) using the area (area up to the lumen boundary) S of the area occupied by the pixels showing the lumen. The average lumen diameter D can also be defined by the formula: average lumen diameter D=(maximum diameter D1+minimum diameter D2) / 2 using the maximum diameter D1 and minimum diameter D2 of the lumen of the blood vessel. As shown in FIG. 5, the random distribution in the upward and downward (positive and negative) directions centered on 0 on the vertical axis indicates the presence of random error. It can also be seen that there is no fixed error showing a distribution biased in a specific direction, or a proportional error showing a distribution proportional to the increase in the measured value.

[0031] From the distribution shown in FIG. 5, it can be seen that the estimation error by the learning model 58 results in approximately the same error regardless of the size of the average lumen diameter of the blood vessel. That is, since the learning model 58 performs classification and estimation on a pixel-by-pixel basis of the medical image, it can be considered that the estimation error per pixel is approximately the same anywhere on the medical image.

[0032] FIG. 6 is a diagram showing an example of the error rate of the estimated value by the learning model 58. In FIG. 6, the horizontal axis represents the average (mm) between the true value and the estimated value of the average lumen diameter, and the vertical axis represents the error rate (%). The error rate can be obtained by the formula {100×(estimated value - true value) / true value}. As shown in FIG. 6, it can be seen that in the range where the average lumen diameter is small, the error rate is large and the influence of the error becomes significant. The influence of the error becomes significant, for example, when the average lumen diameter is about 2.5 mm or less. Thus, the applicant has obtained the finding that when the blood vessel diameter is small (for example, about 2 to 3 mm), the error rate of the pixels estimated by the learning model 58 increases, affecting the accuracy of the estimation result. Since it always includes an error of the same number of pixels, a method for reducing the error at the time of calculating the final representative value (for example, the inner diameter or outer diameter of the blood vessel) by reducing the length per pixel will be described.

[0033] FIG. 7 is a diagram showing an example of the calculation of the inner diameter and outer diameter of the lumen organ by the information processing apparatus 50. Let the medical images (medical image data) obtained along the long axis direction of the blood vessel obtained by the pullback operation be G1, G2,... Gn. The medical images G1, G2,... Gn are obtained from the distal side to the proximal side inside the blood vessel. The computer program 57 inputs the medical images G1, G2,... in this order into the learning model 58 to acquire the position data (first position data) of the lumen boundary and the blood vessel boundary as a predetermined site. Note that the number of medical images input to the learning model 58 may be plural or one.

[0034] The medical image data input to the learning model 58 is reduced to a predetermined size by a size conversion process (computer program 57), and the reduced medical image data is then input to the learning model 58. If the size of the original medical image data is, for example, 512 [px] × 512 [px], the reduced size can be, for example, 128 [px] × 128 [px]. In this case, the size of the segmentation data output by the learning model 58 is 128 [px] × 128 [px], but this is enlarged to 512 [px] × 512 [px], which is the same size as the original medical image data. This allows a good balance between reducing the processing load of the learning model 58 and improving the estimation accuracy.

[0035] The index identification process (computer program 57) identifies (calculates) the index of the blood vessel (hollow organ) based on the acquired position data. The index includes, for example, the average lumen diameter (inner diameter) of the blood vessel. The index may also include the plaque area ratio (Plaque Burden) and eccentricity. The plaque area ratio can be expressed as Plaque Area Ratio = (S'-S) / S' x 100, where S' is the area up to the blood vessel boundary and S is the area up to the lumen boundary. Eccentricity will be described later.

[0036] Furthermore, the computer program 57 can specify (calculate) the outer diameter (average) of the blood vessel in addition to the inner diameter (average lumen diameter) of the blood vessel based on the position data of the lumen boundary and the blood vessel boundary output by the learning model 58. The outer diameter of the blood vessel can be calculated by 2×√(S′ / π), where S′ is the area up to the blood vessel boundary.

[0037] FIG. 8 is a diagram showing a first example of the transition of the average lumen diameter along the long axis direction of a blood vessel. It is assumed that the average lumen diameter identified based on the position data of the predetermined site output by the learning model 58 has transitioned as shown in FIG. 8 for medical images G1, G2, ... in that order. It is assumed that as a result of processing medical image Gi, the average lumen diameter falls below a threshold value. In this case, there is a possibility that the lumen boundary will be erroneously estimated. Therefore, for the unprocessed medical image (in this case, medical image G(i+1)), the processing shown in FIG. 9 described later is switched to.

[0038] FIG. 9 is a diagram showing a first example of medical image generation processing by the information processing device 50. The computer program 57 (image generation processing) acquires unprocessed medical image data G(i+1), and generates medical image data in which an area including a predetermined part in a cross-sectional image of the acquired medical image data is enlarged relatively from a predetermined size to be input to the learning model 58. Here, the predetermined part can be a blood vessel boundary (or an area of ​​the external elastic lamina (EEM)). Specifically, as shown in FIG. 9, the computer program 57 can delete an area other than the area including the predetermined part from the cross-sectional image to generate a cut-out image, reduce the generated cut-out image, and generate medical image data showing a cross-sectional image of a predetermined size to be input to the learning model 58. The size of the cut-out image is between the size of the original medical image data (512 [px] × 512 [px]) and the predetermined size (128 [px] × 128 [px]) to be input to the learning model 58. In the cut-out image, the outer region (Background) of the blood vessel that does not contribute to the estimation process by the learning model 58 is largely deleted, and the region inside the blood vessel boundary as the region of interest is relatively enlarged, which is considered to reduce the estimation error per pixel. The size of the cut-out image is not limited to a size between the size of the original medical image data (512 [px] × 512 [px]) and the predetermined size (128 [px] × 128 [px]) input to the learning model 58, and may be a size smaller than the predetermined size input to the learning model 58. In that case, the cut-out image may be enlarged to the predetermined size and input to the learning model 58.

[0039] The segmentation data (128[px]×128[px]) output by the learning model 58 is enlarged to the same size as the cut-out image, and the enlarged segmentation data is assigned to data (Background) of the same size as the original medical image data. As a result, segmentation data of the area inside the blood vessel boundary is assigned to segmentation data of the area outside the blood vessel.

[0040] The computer program 57 can calculate the inner diameter (average lumen diameter) and the average outer diameter of the blood vessel for the medical image data G(i+1) based on the position data of the lumen boundary and the blood vessel boundary output by the learning model 58.

[0041] In the first example described above, the medical image data is generated by performing a cutout process on the medical image data, but the present invention is not limited to the cutout process. For example, the medical image data may be subjected to a combination of an enlargement process and a reduction process depending on the region.

[0042] FIG. 10 is a diagram showing an example of the enlargement / reduction process for medical image data. As shown in FIG. 10, a rectangular area is considered from the center of the medical image (for example, the center of the catheter) toward the radial direction of the blood vessel. The original medical image is enlarged for the inside of the blood vessel (i.e., the area inside the blood vessel boundary), and reduced for the outside of the blood vessel. The original medical image is deleted for the area corresponding to the catheter. Note that the area corresponding to the catheter may be left as it is without being enlarged, reduced, or deleted. The rectangular area is scanned around the entire circumference and similar enlargement / reduction process is performed to generate medical image data of the same size as the original medical image data (512 [px] x 512 [px]).

[0043] The expansion rate during expansion may be a predetermined value, or may be changed according to the average lumen diameter. For example, the smaller the average lumen diameter, the larger the expansion rate (e.g., from 1.5 times to 2 times). The reduction rate during reduction may be a predetermined value, or may be changed according to the average lumen diameter. For example, the larger the average lumen diameter, the larger the reduction rate (e.g., from 1 / 3 to 1 / 2).

[0044] FIG. 11 is a diagram showing a second example of medical image generation processing by the information processing device 50. The computer program 57 (image generation processing) acquires unprocessed medical image data G(i+1) and generates medical image data in which a region including a predetermined portion in a cross-sectional image of the acquired medical image data is enlarged relatively from a predetermined size to be input to the learning model 58. Here, the predetermined portion can be a blood vessel boundary (or an EEM: region of the external elastic lamina). Specifically, as shown in FIG. 10, the computer program 57 can enlarge a region inside the blood vessel in the cross-sectional image and reduce a region outside the blood vessel to generate medical image data showing a cross-sectional image of a predetermined size to be input to the learning model 58. In the medical image after the enlargement / reduction processing, the region outside the blood vessel (Background) that does not contribute to the estimation processing by the learning model 58 is largely deleted, and the region inside the blood vessel boundary as the region of interest is enlarged, so that it is considered that the estimation error for each pixel is reduced. Furthermore, the processes in the cases of FIGS. 10 and 11 can eliminate the possibility of necessary information being cut out by cutting out the image as in the process in FIG.

[0045] When the identified indicator satisfies a specified condition, the computer program 57 may generate medical image data in which an area including a specified portion in a cross-sectional image of the acquired medical image data for which first position data of the specified portion has not yet been acquired is relatively enlarged from a specified size to be input into the learning model 58, input the generated medical image data into the learning model to acquire second position data of the specified portion, and calculate the inner diameter and outer diameter of the tubular organ based on the acquired second position data.

[0046] In the above example, as illustrated in Fig. 8, medical image data is acquired, segmentation data is output by the learning model 58 in real time, blood vessel indices are identified, and the inner and outer diameters of the blood vessels are calculated. If the average lumen diameter of the blood vessel is equal to or less than a predetermined threshold, image generation processing is performed on the next acquired medical image data as illustrated in Fig. 9 or 11, and the medical image data after the image generation processing is input to the learning model 58 to calculate the inner and outer diameters of the blood vessels. In the following, a case will be described in which, instead of real-time processing, processing is once performed by the learning model 58 on all acquired medical image data, and then processing is once again performed by the learning model 58 on medical image data whose identified indices satisfy a predetermined condition.

[0047] FIG. 12 is a diagram showing a second example of the transition of the average lumen diameter along the long axis direction of the blood vessel. It is assumed that the average lumen diameter specified based on the position data of the predetermined part output by the learning model 58 in the order of medical images G1, G2, ..., Gn transitions as shown in FIG. 12. It is assumed that the average lumen diameter becomes equal to or smaller than the threshold value for medical images Gk to Gm as a result of processing all medical images G1, G2, ..., Gn. In this case, there is a possibility that the error rate becomes large, or the influence of the error on the true value becomes large. Therefore, for medical images Gk to Gm, the process shown in FIG. 9 or FIG. 11 can be switched to and the estimation process can be performed again by the learning model 58. For convenience, the process shown in FIG. 8 is referred to as real-time processing, and the process shown in FIG. 12 is referred to as batch processing.

[0048] When the identified indicator satisfies a specified condition, the computer program 57 may generate medical image data in which an area including a specified portion in a cross-sectional image of medical image data for which first position data identifying the indicator has already been acquired is relatively enlarged from a specified size to be input into the learning model 58, input the generated medical image data into the learning model to acquire second position data of the specified portion, and calculate the inner and outer diameters of the tubular organ based on the acquired second position data.

[0049] As described above, the information processing device 50 (computer program 57) of this embodiment acquires medical image data showing a cross-sectional image of a hollow organ, inputs the acquired medical image data to a learning model 58 that outputs position data of a predetermined part of the hollow organ when the medical image data showing the cross-sectional image of the hollow organ is input, acquires first position data of the predetermined part, identifies an index of the hollow organ based on the acquired first position data, generates medical image data in which an area including the predetermined part in the cross-sectional image of the acquired medical image data is relatively enlarged from a predetermined size input to the learning model 58 when the identified index satisfies a predetermined condition, inputs the generated medical image data to the learning model 58 to acquire second position data of the predetermined part, and calculates the inner diameter and outer diameter of the hollow organ based on the acquired second position data. With this configuration, the outer area (Background) of the blood vessel that does not contribute to the estimation process by the learning model 58 is largely deleted, and the area inside the blood vessel boundary as the area of ​​interest is directly or indirectly enlarged, which is considered to reduce the estimation error for each pixel, and improves the detection accuracy of the stenosis.

[0050] Next, a method for determining the presence or absence of a side branch will be described using the learning model 58. A side branch is a branching portion that branches off from a blood vessel (main trunk). The presence or absence of a side branch can be determined by the eccentricity of the cross-sectional shape of the blood vessel.

[0051] FIG. 13 shows an example of a method for determining the presence or absence of a side branch based on the eccentricity. As described above, when medical image data is input, the learning model 58 outputs position data (first position data) of the boundary of the lumen and the boundary of the blood vessel. The computer program 57 can calculate the eccentricity of the blood vessel cross-sectional shape based on the position data. The eccentricity can be calculated by the formula: eccentricity=(maximum diameter D1-minimum diameter D2) / maximum diameter D1. As shown in FIG. 13A, the maximum diameter D1 and the minimum diameter D2 of the lumen diameter can be calculated based on the boundary of the lumen output by the learning model 58 to obtain the eccentricity. In this case, the maximum diameter D1 and the minimum diameter D2 can be calculated from a line segment passing through the center of gravity of the blood vessel. In the case of FIG. 13A, the eccentricity is equal to or greater than a predetermined threshold, and it is determined that a side branch is present. In the case of FIG. 13B, the eccentricity is less than a predetermined threshold, and it is determined that a side branch is not present. The eccentricity may be calculated based on the boundary of the blood vessel. In addition, instead of using the maximum and minimum diameters, for example, the distance from the center (which may be the center of gravity) of all pixels (picture elements) present within the boundary of the lumen may be calculated, the distribution of distances within the boundary of the lumen may be obtained, and the presence or absence of side branches may be determined based on the number of pixels whose distance is equal to or greater than a predetermined threshold value.

[0052] The computer program 57 may determine the presence or absence of a side branch of the hollow organ based on the calculated eccentricity, and when it is determined that a side branch is present, may stop the generation of medical image data in which an area including a predetermined portion is enlarged relatively from a predetermined size to be input to the learning model (for example, the process shown in FIG. 9 or FIG. 11). This makes it possible to prevent the loss of necessary information, for example, by deleting all or part of the side branch from the medical image by performing the enlargement process of the predetermined portion.

[0053] Next, the processing procedure of the information processing device 50 will be described.

[0054] Fig. 14 is a diagram showing an example of real-time processing by the information processing device 50. The processing shown in Fig. 14 corresponds to the processing exemplified in Fig. 8. For convenience, the processing will be described below as being performed by the control unit 51. The control unit 51 acquires medical image data (S11), and inputs the acquired medical image data into a learning model 58 to acquire position data (first position data) of a predetermined site (S12). The predetermined site includes the boundary of the lumen and the boundary of the blood vessel.

[0055] The control unit 51 calculates the inner diameter (average) and the outer diameter (average) of the blood vessel based on the position data (S13), and judges the presence or absence of a side branch (S14). The presence or absence of a side branch can be judged based on the eccentricity of the cross-sectional shape of the blood vessel. If a side branch is present (YES in S14), the control unit 51 judges whether or not the processing of all medical image data has been completed (S15). If the processing of all medical image data has not been completed (NO in S15), the control unit 51 continues the processing from step S11 onwards, and if the processing of all medical image data has been completed (YES in S15), the processing ends.

[0056] If there is no side branch (NO in S14), the control unit 51 determines whether the inner diameter of the blood vessel is equal to or smaller than the threshold (S16), and if the inner diameter of the blood vessel is not equal to or smaller than the threshold (NO in S16), the control unit 51 performs the process of step S15. If the inner diameter of the blood vessel is equal to or smaller than the threshold (YES in S16), the control unit 51 acquires unprocessed medical image data (S17), and generates medical image data by enlarging an area including a predetermined portion of the acquired medical image data relatively from a predetermined size to be input to the learning model 58 (S18). The generation of medical image data is, for example, the process illustrated in FIG. 9 or FIG. 11.

[0057] The control unit 51 inputs the generated medical image data to the learning model 58 to obtain position data (second position data) of the predetermined part (S19). The predetermined part includes the boundary of the lumen and the boundary of the blood vessel. The control unit 51 calculates the inner diameter (average) and the outer diameter (average) of the blood vessel based on the position data (S20), and continues the processing from step S11 onwards.

[0058] Fig. 15 is a diagram showing an example of batch processing by the information processing device 50. The processing shown in Fig. 15 corresponds to the processing in the case shown in Fig. 12. The control unit 51 acquires medical image data (S31), and inputs the acquired medical image data to the learning model 58 to acquire position data (first position data) of a predetermined region (S32). The predetermined region includes the boundary of the lumen and the boundary of the blood vessel.

[0059] The control unit 51 calculates the inner diameter (average) and the outer diameter (average) of the blood vessel based on the position data (S33), and determines whether or not there is a side branch (S34). The control unit 51 determines whether or not the inner diameter of the blood vessel is equal to or smaller than a threshold (S35). The determination results of whether or not there is a side branch and whether or not the inner diameter of the blood vessel is equal to or smaller than a threshold may be stored in the storage unit 56 in association with each medical image data.

[0060] The control unit 51 determines whether processing of all medical image data has been completed (S36), and if processing of all medical image data has not been completed (NO in S36), it continues processing from step S31 onwards, and if processing of all medical image data has been completed (YES in S36), it extracts medical image data that has no side branches and whose inner diameter of the blood vessel is below a threshold value from the medical image data acquired in step S31 (S37).

[0061] The control unit 51 generates medical image data by enlarging an area including the specified portion of the extracted medical image data relatively from a specified size to be input to the learning model 58 (S38). The generation of medical image data is, for example, the process illustrated in FIG. 9 or FIG. 11. The control unit 51 inputs the generated medical image data to the learning model 58 to obtain position data of the specified portion (second position data) (S39). The specified portion includes the boundary of the lumen and the boundary of the blood vessel. The control unit 51 calculates the inner diameter (average) and outer diameter (average) of the blood vessel based on the position data (S40), and ends the process.

[0062] In the above embodiment, the information processing device 50 is configured to calculate the inner diameter and outer diameter of the blood vessel and generate the learning model, but is not limited to this. For example, the information processing device 50 may be a client device, and the generation process of the learning model may be performed by an external server, and the learning model may be acquired from the server. Also, the calculation of the inner diameter and outer diameter of the blood vessel and the generation of the learning model may be performed by an external server, and the information processing device may acquire the calculation result from the server. [Explanation of symbols]

[0063] 1. Communication Network 10 Catheter 20 MDU 30 Display device 40 Input Devices 50 Information processing device 51 Control section 52 Communications Department 53 Interface section 54 Recording medium reading unit 541 Recording media 55 Memory 56 Memory section 57 Computer Programs 58 Learning Model 58a Input layer 58b Middle class 58c Output layer

Claims

1. On the computer, Acquiring medical image data representing a cross-sectional image of a hollow organ; inputting the acquired medical image data into a learning model that outputs position data of a predetermined part of a hollow organ when the medical image data showing a cross-sectional image of the hollow organ is inputted, and acquiring first position data of the predetermined part; Identifying an indicator of the hollow organ based on the acquired first position data; If the identified index satisfies a predetermined condition, a region including the predetermined portion in the cross-sectional image of the acquired medical image data is enlarged or reduced relatively to a predetermined size to be input to the learning model to generate cut-out medical image data; Reducing the enlarged cut-out medical image data or enlarging the reduced cut-out medical image data to generate medical image data showing a cross-sectional image of the predetermined size to be input to the learning model; inputting the generated medical image data into the learning model to obtain second position data of the predetermined part; calculating an inner diameter of the hollow organ based on the acquired second position data; A computer program that executes a process.

2. On the computer, calculating an outer diameter of the hollow organ based on the acquired second position data; 2. A computer program product as claimed in claim 1, which is adapted to carry out a process.

3. On the computer, and generating medical image data representing a cross-sectional image of a predetermined size to be input to the learning model by deleting an area other than the area including the predetermined portion from the cross-sectional image.

3. A computer program product according to claim 1 or 2, which causes a process to be executed.

4. the predetermined region includes a blood vessel boundary; On the computer, enlarging an area inside the blood vessel and reducing an area outside the blood vessel in the cross-sectional image to generate medical image data showing a cross-sectional image of a predetermined size to be input to the learning model; 3. A computer program product according to claim 1 or 2, which causes a process to be executed.

5. On the computer, Calculating the eccentricity of the blood vessel cross section based on the acquired first position data; determining the presence or absence of a side branch of the hollow organ based on the calculated eccentricity; If a side branch is present, generation of cut-out medical image data in which a region including the predetermined portion is enlarged or reduced relatively from a predetermined size to be input to the learning model is stopped. A computer program product according to any one of claims 1 to 4, which is used to execute a process.

6. The indicator includes an inner diameter of a hollow organ, The predetermined condition includes that the inner diameter is equal to or less than a predetermined threshold value. A computer program according to any one of claims 1 to 5.

7. On the computer, When the specified index satisfies a predetermined condition, a region including the specified portion in a cross-sectional image of the acquired medical image data from which first position data of the specified portion has not yet been acquired is enlarged or reduced relatively from a predetermined size to be input to the learning model to generate cut-out medical image data. A computer program product according to any one of claims 1 to 6, which is used to execute a process.

8. On the computer, When the identified marker satisfies a predetermined condition, a region including the predetermined portion in the cross-sectional image of the medical image data from which the first position data from which the marker is identified has already been acquired is enlarged or reduced relatively to a predetermined size to be input to the learning model to generate cut-out medical image data. A computer program product according to any one of claims 1 to 6, which is used to execute a process.

9. a first acquisition unit that acquires medical image data representing a cross-sectional image of a hollow organ; a second acquisition unit that inputs acquired medical image data into a learning model that outputs position data of a predetermined part of a hollow organ when the medical image data showing a cross-sectional image of the hollow organ is input, and acquires first position data of the predetermined part; an identification unit that identifies an indicator of the hollow organ based on the acquired first position data; a first generation unit that generates cut-out medical image data by enlarging or reducing an area including the predetermined portion in the cross-sectional image of the acquired medical image data relatively to a predetermined size to be input to the learning model when the identified index satisfies a predetermined condition; A second generation unit that reduces the enlarged cut-out medical image data or enlarges the reduced cut-out medical image data to generate medical image data showing a cross-sectional image of the predetermined size to be input to the learning model; a third acquisition unit that inputs the generated medical image data into the learning model to acquire second position data of the predetermined part; a calculation unit that calculates an inner diameter of the hollow organ based on the acquired second position data; An information processing device comprising:

10. A first acquisition unit acquires medical image data representing a cross-sectional image of a hollow organ; a second acquisition unit acquires first position data of a predetermined portion by inputting the acquired medical image data into a learning model that outputs position data of a predetermined portion of the hollow organ when medical image data showing a cross-sectional image of the hollow organ is input; an identification unit identifies an indicator of the hollow organ based on the acquired first position data; When the identified index satisfies a predetermined condition, a first generation unit generates cut-out medical image data by enlarging or reducing an area including the predetermined portion in the cross-sectional image of the acquired medical image data relatively to a predetermined size to be input to the learning model, A second generation unit generates medical image data indicating a cross-sectional image of the predetermined size to be input to the learning model by reducing the enlarged cut-out medical image data or by enlarging the reduced cut-out medical image data; The generated medical image data is input to the learning model, and a third acquisition unit acquires second position data of the predetermined part; A calculation unit calculates an inner diameter of the tubular organ based on the acquired second position data. Information processing methods.

Images