Information processing device, information processing program, and information processing method

Abstract

To provide an index for quantitatively evaluating a change in the size of a subject without being affected by the existence / nonexistence of periodicity (regularity) on the change in the size of the subject.SOLUTION: According to the present invention, an information processing device 1 calculates an evaluation amount being an index for quantitatively evaluating a change in the size of a subject in a plurality of static images photographed at equal time intervals. The information processing device includes a storage part 12, a size calculation part 134 for calculating sizes of respective subjects in the static images stored in the storage part, and an evaluation amount calculation part 135 for calculating the evaluation amount of each subject on the basis of the size of each of two static images which are consecutively photographed. The two static images include a first static image and a second static image photographed after that. The evaluation amount calculation part calculates the evaluation amount of each subject on the basis of a ratio of the size of the subject in the second static image to the sum of the size of the subject in the first static image and the size of the subject in the second static image.SELECTED DRAWING: Figure 2

Description

[Technical field]

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

[0002] Chronic intestinal pseudo-obstruction (CIPO) is an intractable disease in which the motility of the digestive tract is inhibited despite the absence of a physical obstruction that would prevent the transport of contents in the digestive tract, including the small intestine. Patients with CIPO are diagnosed based on still images taken with a computed tomography (CT) or X-ray imaging device. However, such diagnoses are based on still images, making it difficult to evaluate the peristaltic motility of the digestive tract, and also pose the problem of exposure to radiation.

[0003] As a technique for solving such problems, a technique using a cine-MRI device that captures MRI (Magnetic Resonance Imaging) images in a time series as moving images has been proposed in recent years (see, for example, Non-Patent Document 1).

[0004] The technology disclosed in Non-Patent Document 1 calculates the size of the short axis length of the small intestine captured by MRI images, detects the movement of the intestine (change in size) as a periodic increase and decrease in the short axis length on the premise of periodic movement of the intestine, and models this movement using three parameters. This technology can evaluate the movement of the intestine when the increase and decrease in the short axis length changes periodically. However, periodic changes are rarely observed in the movement of the small intestine, and there are cases where there is no movement of the intestine. Therefore, this technology, which requires capturing the periodicity of the intestinal movement that rarely occurs as a premise, is difficult to follow up and has low reproducibility. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] A. Fuyuki, etal., Clinical importance of cine-MRI assessment of small bowel motility in patients with chronic intestinal pseudo-obstruction: a retrospective study of 33 patients, Journal of Gastroenterology, 2017, Vol.52, pp.577-584. Summary of the Invention [Problem to be solved by the invention]

[0006] An object of the present invention is to provide an index for quantitatively evaluating changes in the size of a subject, regardless of the presence or absence of periodicity (regularity) in the movement (change in size) of the subject. [Means for solving the problem]

[0007] The present invention is an information processing device that calculates an evaluation amount, which is an index for quantitatively evaluating changes in the size of a subject in a plurality of still images taken at equal time intervals, and is characterized in that it comprises a memory unit that stores the plurality of still images, a size calculation unit that calculates the size of the subject captured in each of the still images stored in the memory unit, and an evaluation amount calculation unit that calculates an evaluation amount for each subject based on the size in each of two still images taken in succession, wherein the two still images include a first still image and a second still image taken after the first still image, and the evaluation amount calculation unit calculates the evaluation amount for each subject based on the ratio of the size of the subject in the second still image to the sum of the size of the subject in the first still image and the size of the subject in the second still image. Effect of the Invention

[0008] According to the present invention, it is possible to provide an index for quantitatively evaluating changes in the size of a subject, regardless of the presence or absence of periodicity (regularity) regarding changes in the size of the subject. [Brief description of the drawings]

[0009] [Figure 1] 1 is a network connection diagram showing an embodiment of an information processing device according to the present invention; [Diagram 2] 1 is a functional block diagram showing an embodiment of an information processing apparatus according to the present invention; [Diagram 3] 1 is a flowchart illustrating an embodiment of an information processing method according to the present invention. [Figure 4] 4 is a flowchart of an evaluation amount calculation process included in the information processing method of FIG. 3. [Diagram 5] 5 is an image showing an example of a still image and an example of a sharpened image thereof used in the evaluation amount calculation process of FIG. 4. [Figure 6] 5 is a schematic diagram showing an image of calculation of the short axis length of the intestinal tract in the evaluation amount calculation process of FIG. 4. [Figure 7] 4 is a flowchart of a feature extraction process included in the information processing method of FIG. 3. [Figure 8] 4 is a flowchart of a classifier design process included in the information processing method of FIG. 3. [Figure 9] 4 is a flowchart of an identification and diagnosis process included in the information processing method of FIG. 3. [Figure 10] 1 is a graph showing the change over time in the evaluation amount calculated in an embodiment of the present invention, where (a) shows the evaluation amount of an intestine with a change in size, and (b) shows the evaluation amount of an intestine with little change in size. [Figure 11] 1 is a graph showing the intestines from 18 cases used in an embodiment of the present invention on a plane with two features as coordinates. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Below, with reference to the drawings, we will explain embodiments of an information processing device (hereinafter referred to as "this device"), an information processing program (hereinafter referred to as "this program"), and an information processing method (hereinafter referred to as "this method") according to the present invention.

[0011] The present invention calculates an evaluation amount, which is an index for quantitatively evaluating the change in size of a subject (motion of the subject) in a plurality of still images captured at equal time intervals. The present invention also classifies the subject as either a specific subject or a non-specific subject based on two features extracted based on the size and evaluation amount of the subject, and diagnoses the patient based on the classification result.

[0012] "Still images" are images of a subject captured at equal time intervals over a predetermined period of time. Still images are captured in a fixed imaging range (angle of view) by an imaging device described below. In this embodiment, the still images are cine-MRI images, the predetermined period is 17 seconds, and the time interval is 0.5 seconds. That is, in this embodiment, 35 still images are captured in chronological order for one subject as a continuous moving image (i.e., video).

[0013] In addition, the still image in the present invention is not limited to a cine-MRI image, as long as it is an image of a subject in a certain imaging range captured at equal time intervals for a predetermined period of time. That is, for example, the still image may be an image of the surface or inside of a subject captured at a fixed point. Furthermore, as long as it satisfies the above-mentioned conditions, the still image may be an X-ray transmission image or a CT image.

[0014] Furthermore, the predetermined period and time interval during which still images are captured may be any period and time interval that allows the present invention to calculate a meaningful evaluation amount in the field of application, and are not limited to the present embodiment. That is, for example, when the change in the size of the subject cannot be evaluated without long-term observation, the predetermined period may be 24 hours and the time interval may be 1 hour. Alternatively, for example, when the subject moves in an extremely short time, the predetermined period may be 17 seconds and the time interval may be 0.1 seconds.

[0015] The "subject" is an object whose change in size is to be evaluated, and whose image is captured by an imaging device described below. In this embodiment, the subject is the intestinal tract of the small intestine of a human body.

[0016] The subject in the present invention may be anything that can calculate an evaluation amount by the present invention, and is not limited to the present embodiment. That is, for example, the subject may be an animal's organ (e.g., digestive organs such as the large intestine and stomach of the human body, the bladder, the heart, the lungs, and the uterus), a blood vessel, or a pupil, or may be a machine equipped with a mechanism that moves within the range of the angle of view of the imaging device.

[0017] "Size" is a physical quantity indicating the size of a subject, such as the length (dimension), area, volume, etc. In this embodiment, the object whose size is measured is the short diameter length of the intestine, which is a part of the subject.

[0018] The "change in size" of the subject is, for example, a change in the shape (contour) of the subject, such as expansion, contraction, or expansion. Here, the change in size may include not only a non-periodic (irregular) change, but also a periodic (regular) change. In this embodiment, the change in size of the subject is a change in the short axis length of the intestinal tract of the human body.

[0019] The "evaluation amount" is an index for quantitatively evaluating the change in size of an imaged subject between two still images captured in succession. That is, the evaluation amount is an index that quantitatively indicates the degree of change in size of the subject between two still images. In other words, the evaluation amount is an index that quantitatively indicates how much the subject has moved for each time interval. Details of the evaluation amount will be described later.

[0020] "Two still images captured consecutively" are, for example, two still images adjacent to each other in a series of still images (video). That is, for example, two still images captured consecutively are a still image captured at time "t-1" and a still image captured at time "t" that is later than time "t-1".

[0021] A "specific subject" is a subject in which a specific phenomenon occurs in the change in size. In this embodiment, the specific subject means the intestine from a patient with ileus pseudo-obstruction. That is, the specific subject is a subject in which the change in size is not normal (abnormal).

[0022] A "non-specific subject" is a subject in which no specific phenomenon occurs in the change in size. In this embodiment, a non-specific subject refers to the intestines of a patient who does not have ileus pseudo-obstruction or a healthy subject. In other words, a non-specific subject is a subject in which the change in size is normal.

[0023] The "specific phenomenon" is, for example, a symptom occurring in an organ and related to a movement manifested as a change in size (such as a decrease in the change in size, an increase in the change in size, or an abnormal increase in size). In this embodiment, the specific phenomenon is the abnormal increase and decrease in the change in size seen in the intestine exhibiting the symptom of intestinal pseudo-obstruction.

[0024] ●Information processing device● First, an embodiment of the present device will be described.

[0025] FIG. 1 is a network connection diagram showing an embodiment of the present device. The figure shows that the present device 1 is connected to an imaging device 2 via a network (communication line) N that uses a wired communication method or a wireless communication method.

[0026] This device 1 calculates an evaluation amount for quantitatively evaluating the change in size of a subject in a plurality of still images taken at equal time intervals (0.5 second intervals in this embodiment). The specific configuration and operation of this device 1 will be described later.

[0027] The imaging device 2 continuously captures still images of a subject. In this embodiment, for example, the imaging device 2 is a cine-MRI device, and the subject is the intestinal tract of the small intestine.

[0028] The imaging device in the present invention is not limited to a cine-MRI device. That is, for example, the imaging device in the present invention may be an X-ray fluoroscopy device or a CT device, or may be a device with a built-in camera (for example, a smartphone or a tablet terminal), or a general-purpose digital camera.

[0029] The network N is, for example, a communication network such as the Internet, a mobile communication network, a local area network (LAN), a wide area network (WAN), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0030] Configuration of information processing device FIG. 2 is a functional block diagram showing an embodiment of the present device 1. As shown in FIG.

[0031] The present device 1 is realized, for example, by a personal computer. The present program runs on the present device 1, and the present program cooperates with the hardware resources of the present device 1 to realize the present method.

[0032] Here, by causing a computer (not shown) to execute this program, the program can cause the computer to function in the same manner as the device 1, and to execute the method.

[0033] The device 1 includes a communication unit 11, a storage unit 12, a control unit 13, an operation unit 14, and a display unit 15.

[0034] The communication unit 11 acquires still images from the imaging device 2 via the network N. The communication unit 11 is configured with, for example, a communication module and a communication interface. The still images are, for example, collected into one video (a still image group) for each predetermined period of time and stored in the storage unit 12.

[0035] The storage unit 12 stores information necessary for the device 1 to execute the method described below. The storage unit 12 is configured, for example, with a recording device such as a hard disk drive (HDD) or a solid state drive (SSD) provided in the device 1, a volatile memory such as a random access memory (RAM), and / or a portable storage medium such as a flash memory.

[0036] The control unit 13 controls the operation of the entire device 1 and executes the method described below. The control unit 13 is composed of, for example, a CPU (Central Processing Unit) provided in the device 1, a volatile memory such as a RAM that functions as a working area for the CPU, and a non-volatile memory such as a ROM (Read Only Memory) that stores various information such as the program. The control unit 13 includes an image acquisition unit 131, an image display unit 132, an image sharpening unit 133, a size calculation unit 134, an evaluation amount calculation unit 135, a feature extraction unit 136, a classifier design unit 137, a classification unit 138, and a diagnosis unit 139.

[0037] The image acquisition unit 131 acquires a video for executing the method described later, that is, a plurality of still images (a group of still images) captured continuously. Specific operations of the image acquisition unit 131 will be described later.

[0038] The image display unit 132 displays the still image stored in the storage unit 12 on the display unit 15. The specific operation of the image display unit 132 will be described later.

[0039] The image sharpening section 133 generates a sharpened image in which the subject (intestinal tract) captured in the still image is sharpened. The specific operation of the image sharpening section 133 will be described later.

[0040] Size calculation section 134 calculates the size of the subject captured in each of the videos (sharpened still images) stored in storage section 12. The specific operation of size calculation section 134 will be described later.

[0041] The evaluation amount calculation unit 135 calculates an evaluation amount for each subject based on the size of the subject in each of two still images captured consecutively among the still images stored in the storage unit 12. A specific operation of the evaluation amount calculation unit 135 will be described later.

[0042] The feature extraction unit 136 extracts a first feature based on the size of the subject, and extracts a second feature based on the evaluation amount of the subject. The specific operation of the feature extraction unit 136 and the first and second features will be described later.

[0043] The classifier design unit 137 designs a classifier (trains the classifier) ​​to be used for subject classification by the classification unit 138. A specific operation of the classifier design unit 137 will be described later.

[0044] The classification unit 138 classifies the subject into either a specific subject or a non-specific subject based on the first feature and the second feature of the subject. A specific operation of the classification unit 138 will be described later.

[0045] The diagnosis unit 139 diagnoses the patient based on the identification result of the identification unit 138. The specific operation of the diagnosis unit 139 will be described later.

[0046] The operation unit 14 is a device (eg, a keyboard, a mouse, a touch panel, etc.) that accepts operations (such as input of information or selection) from a user of the present device 1 (eg, a doctor, hereinafter simply referred to as "user").

[0047] Display unit 15 displays information (such as an information selection screen, an input screen, and a recognition result) required for carrying out the present method described below. In this embodiment, operation unit 14 and display unit 15 are, for example, a touch panel display.

[0048] The operation unit in the present invention may be an input device such as a mouse or a keyboard, and the display unit in the present invention may be a monitor or display connected to the device.

[0049] ●Information processing method● Next, the operation of the present device 1, ie, the present method executed by the present device 1, will be described.

[0050] In the following description of this method, the still images are cine-MRI images of cross sections of the abdomen of a human body captured using a cine-MRI device, and the subject is the small intestine captured in the still images. The video is 35 still images captured continuously at equal time intervals (0.5 seconds) within a predetermined period (17 seconds), and is transmitted in advance from imaging device 2 to this device 1 via network N and stored in storage unit 12. The 35 still images form one video (still image group).

[0051] In the following description of the present method, it is assumed that the intestines that are visibly captured in all still images constituting one video (still image group) are specified and selected in advance by the user as the processing target of each process of the present method. The number of intestines in one video (still image group) may be one or multiple (e.g., about five).

[0052] FIG. 3 is a flow chart illustrating an embodiment of the method.

[0053] The apparatus 1 executes an evaluation amount calculation process (S1), a feature extraction process (S2), a classifier design process (S3), and a classification and diagnosis process (S4).

[0054] ●Evaluation amount calculation process FIG. 4 is a flowchart of the evaluation amount calculation process (S1).

[0055] The "evaluation amount calculation process (S1)" is a process for calculating an evaluation amount of the intestinal tract captured in a still image.

[0056] First, the image acquisition unit 131 reads (acquires) from the storage unit 12 (S101), for example, one image of the intestinal tract that is the subject of the evaluation amount calculation process (S1), that is, a group of still images (35 still images taken continuously in chronological order). The image (group of still images) is selected by, for example, the user operating the operation unit 14.

[0057] Next, the image sharpening unit 133 performs known image processing on the still image to generate a sharpened image in which the contour of the intestinal tract is emphasized (S102). At this time, the image display unit 132 causes the display unit 15 to display the sharpened image.

[0058] The "known image processing" is, for example, a known sharpening process (unsharp mask process) that emphasizes high frequency components of an image to sharpen a subject captured in the image.

[0059] FIG. 5 is an image showing an example of a still image and its sharpened image. This figure shows that the contour of the intestine is emphasized by image processing (unsharp mask processing). In the following description, unless otherwise specified, a "still image" refers to a still image that has been subjected to sharpening processing (sharpened image).

[0060] Return to Figure 4. Next, the image display unit 132 causes the display unit 15 to display the first still image in time series from among the group of still images (sharpened images) constituting one video image, for which the short axis length described below has not been calculated (S103).

[0061] Next, the size calculation unit 134 calculates the short axis length of the intestine selected by the user from among one or more intestines captured in the still image (S104). Here, the user selects one intestine whose short axis length, which will be described later, has not been calculated from among the intestines to be processed in the evaluation amount calculation process (S1).

[0062] FIG. 6 is a schematic diagram showing an image of calculation of the short axis length of the intestinal tract.

[0063] The "minor axis length" is the diameter (minor axis) of the intestinal tract on the short side, and in this embodiment, is expressed as the length between two points (between two pixels) on a line perpendicular to the tangent to the contour of the intestinal tract. The minor axis length is an example of a size in the present invention.

[0064] Return to Figure 4. In this embodiment, the size calculation unit 134 calculates the length "L" between two pixels (first pixel, second pixel) indicating both end points of the minor axis (i.e., the Euclidean distance between the two pixels) using the following formula (1). Here, the first pixel and the second pixel are selected by, for example, the user operating the operation unit 14.

[0065]

number

[0066] Next, the control unit 13 determines whether or not there is a next still image in time series (within the video) from among the still images that make up the video (S105).

[0067] When there is a next still image in the time series ("Y" in S105), the evaluation amount calculation process (S1) returns to process S103. That is, the next still image in the time series is displayed on the display unit 15.

[0068] On the other hand, when there is no next still image in the time series, i.e., when the short axis lengths of the selected intestine have been calculated for all 35 still images ("N" in S105), the evaluation amount calculation unit 135 estimates the average value "ML" of the short axis lengths for the 35 still images (S106). The estimated average value "ML" is stored in the memory unit 12. In this way, the short axis lengths of each selected intestine are calculated for all 35 still images constituting the video, the average value "ML" is estimated, and the short axis lengths and their average value "ML" are stored.

[0069] Next, the first still image in the time series is displayed, and if there is an intestine (another intestine) whose short axis length has not been calculated among the intestines that are the subject of the evaluation amount calculation process (S1) ("Y" in S107), the evaluation amount calculation process (S1) returns to process S104.

[0070] On the other hand, when there is no intestine (another intestine) whose short axis length has not been calculated among the intestines to be processed in the evaluation amount calculation process (S1) ("N" in S107), the evaluation amount calculation unit 135 calculates the evaluation amount "J t The evaluation amount calculation unit 135 selects the intestinal tract for which the evaluation amount " is to be calculated (S108). The evaluation amount calculation unit 135 selects the intestinal tract in the order of the intestinal tracts selected in the process S104, for example.

[0071] The evaluation amount calculation unit 135 calculates the evaluation amount "J" at the time "t" of the intestine using the following formula (2). t " is calculated (S109).

[0072]

number

[0073] As shown in formula (2), the evaluation amount calculation unit 135 calculates the evaluation amount "J" based on the ratio of the short axis length of the intestine in the second still image to the sum of the short axis length of the intestine in the first still image and the short axis length of the intestine in the second still image, that is, the ratio. t In this way, the evaluation quantity "J tIn the calculation of "," the ratio of the short axis length between two still images is used to reflect the change in size (short axis length) over time, and the relative value (ratio) of the short axis length is used to make it independent of the scale of the still images of the imaging device 2, i.e., the resolution (number of pixels). As a result, 34 evaluation quantities "J" are calculated for each intestine from one image, i.e., a group of still images (35 still images). t " is calculated. The calculated evaluation quantity "J t For example, the evaluation quantity "J t " is stored in the storage unit 12 in association with information indicating the intestinal tract corresponding to " and a video (a group of still images) in which the intestinal tract is captured.

[0074] The evaluation quantity "J t The value of "J" is expressed in the range from "0" to "1". t When the value of " is 0.5", the short diameter of the intestine does not change (the size of the intestine remains unchanged), when it is greater than "0.5", the short diameter of the intestine expands, and when it is less than "0.5", the short diameter of the intestine shrinks.

[0075] Next, the control unit 13 judges the presence or absence of an intestine for which not all evaluation amounts have been calculated (uncalculated) in the video (still image group) (S110). This judgment is performed, for example, by the user selecting an option displayed on the display unit 15.

[0076] When there is an intestine for which all the evaluation amounts have not been calculated ("Y" in S110), the evaluation amount calculation process (S1) returns to process S108. That is, the evaluation amount calculation unit 135 selects the next intestine and calculates the evaluation amount of that intestine.

[0077] On the other hand, when there is no intestinal tract for which all evaluation amounts have not been calculated ("N" in S110), the device 1 ends the evaluation amount calculation process (S1).

[0078] Feature extraction processing FIG. 7 is a flowchart of the feature extraction process (S2).

[0079] The "feature extraction process (S2)" is a process for extracting features used to identify the intestine (features representing the degree of abnormal enlargement of the intestine, features representing changes in the size of the intestine).

[0080] First, the feature extraction unit 136 reads out the average value "ML" of the short axis length of each of the intestines selected from the intestines of a plurality of healthy subjects (hereinafter referred to as "intestines from a plurality of healthy subjects") from the storage unit 12 (S201). Here, the calculation of the short axis length of each of the intestines of the plurality of healthy subjects and the estimation of the average value "ML" are performed in advance, for example, by the evaluation amount calculation process (S1), and the calculated short axis length and the estimated average value "ML" are stored in the storage unit 12. Each of the intestines of the plurality of healthy subjects is an intestine that has not undergone abnormal enlargement or has not undergone a decrease in size change (intestine with normal size change). Moreover, the average value "ML" of the short axis length of each of the intestines of the plurality of healthy subjects is an example of the second average value in the present invention.

[0081] Next, the feature extraction unit 136 extracts the mean "μ" and the variance "σ" of the distribution of the average values ​​"ML" of the short axis lengths of the intestines of the multiple healthy subjects read out in step S201. 2 " and the estimated mean "μ" and variance "σ 2 " is stored in the memory unit 12 in association with information indicating the intestinal tract from a plurality of healthy individuals, for example.

[0082] The feature extraction unit 136 reads out from the memory unit 12 the average short axis length "ML" of the intestine that is the subject of the feature extraction process (S2), i.e., the intestine selected from the intestines of the diagnosed person (hereinafter referred to as the "intestine of the diagnosed person") (S203).

[0083] Next, the feature extraction unit 136 uses the following formula (3) to calculate the mean "μ" and variance "σ" of a distribution consisting of the average value "ML" of the short axis length of the intestine from the subject and the average value "ML" of the short axis length of the intestine from each of a plurality of healthy subjects read out from the storage unit 12. 2" and calculate the following statistical distance (in this embodiment, the Mahalanobis distance) "D(x)" and extract its logarithmic value as a feature value (S204). The average value "ML" of the short axis length of the intestine from the diagnosed person is an example of the first average value in the present invention.

[0084]

number

[0085] The "Mahalanobis distance" is described in, for example, Mahalanobis, Prasanta Chandra. "On the generalised distance in statistics". Proceedings of the National Institute of Sciences of India 2 (1): 49-55, 1936. The Mahalanobis distance is a distance obtained by normalizing the average value "ML" of the short diameter length of a certain intestine (the intestine from the subject) by a distribution (normal distribution) consisting of the average value "ML" of the short diameter length of each of the intestines from multiple healthy subjects. Since the Mahalanobis distance is a relative value, it has the property of being independent of the scale of the short diameter length. In this embodiment, the logarithm "D1" of the Mahalanobis distance is extracted as a feature (first feature) of the intestine from the subject. The Mahalanobis distance is an example of a statistical distance in the present invention.

[0086] Next, the feature extraction unit 136 extracts each evaluation amount of the intestine from the subject, “J t " is read from the storage unit 12 (S205).

[0087] Next, the feature extraction unit 136 extracts the intestinal tract evaluation amount “J t A 34-dimensional motion vector is defined with each of these as components (S206).

[0088] Next, the feature extraction unit 136 defines a reference vector that indicates that the size of the intestine has not changed (indicates that the intestine is not moving) (S207).

[0089] The "reference vector" is a multidimensional vector whose components are all set to "0.5". The value of "0.5" is the same as "L t " = "L t-1 ", that is, the size of the intestine is unchanged (the intestine is immobile), and is an example of a reference amount in the present invention.

[0090] Next, the feature extraction unit 136 extracts the distance "D2" between the movement vector of the intestine from the subject and the reference vector as a feature (second feature) of the intestine from the subject using the following equation (4) (S208).

[0091]

number

[0092] The larger the value of the distance "D2" extracted as a feature in this way, the further away from the reference vector it is, meaning that the size (short axis length) of the intestine is changing (the movement of the intestine is large), and the smaller the value, the closer the movement vector is to the reference vector, meaning that the size (short axis length) of the intestine is not changing (the movement of the intestine is small). The distance "D2" is stored in the memory unit 12.

[0093] In this way, the feature extraction unit 136 extracts the logarithm "D1" of the Mahalanobis distance "D(x)" of the intestine from the subject and the distance "D2" between the 34-dimensional motion vector of the intestine from the subject and the 34-dimensional reference vector as features of the intestine from the subject. That is, in the identification and diagnosis process (S4) described later, the device 1 uses the logarithm "D1" of the Mahalanobis distance as a feature representing the degree of abnormal enlargement of the intestine, and the distance "D2" as a feature representing the degree of change in the size of the intestine.

[0094] Classifier design process FIG. 8 is a flowchart of the classifier design process (S3).

[0095] In the following description, the classifier designed by the classifier design process (S3) is assumed to perform threshold processing on the feature values ​​for each intestine to determine whether the intestine selected as the classification target is an intestine from a patient with pseudo-intestinal obstruction. In this manner, by using threshold processing for classification, each feature becomes medically understandable, and classification results become easier to interpret. In addition, in the following description, it is assumed that the storage unit 12 has previously stored, in the feature extraction process (S2), the value of the logarithm "D1" of the Mahalanobis distance for each intestine from a plurality of patients with pseudo-intestinal obstruction and those without pseudo-intestinal obstruction and the value of the distance "D2" as feature values ​​for each intestine selected from the intestines from the samples for designing the classifier (hereinafter referred to as "intestines from samples").

[0096] The "classifier design process (S3)" is a process of designing a classifier used by the device 1 to classify a selected intestinal tract into either an intestinal tract from a patient with ileus pseudo-obstruction (specific subject) or an intestinal tract from a person without ileus pseudo-obstruction (a healthy person in this embodiment) (non-specific subject) in the classification / diagnosis process (S4) described later. Specifically, classification is a threshold process for each of two feature values, and the classifier design process (S3) is a process of optimizing a set of thresholds used for classification.

[0097] The "optimization process" is a process for identifying an optimal set of thresholds "α" and "β" for identifying the intestines from a patient with pseudo-intestinal obstruction with high accuracy when identifying whether the intestines from a design sample are from a patient with pseudo-intestinal obstruction. The set of thresholds "α" and "β" is, for example, parameters set in a classifier. The threshold "α" is an example of a first threshold in the present invention, and the threshold "β" is an example of a second threshold in the present invention.

[0098] First, the classifier design unit 137 reads out from the storage unit 12 the value of the logarithm of the Mahalanobis distance “D1” (first feature value) and the value of the distance “D2” (second feature value) extracted from the intestinal tract of a design sample (S301).

[0099] Next, the classifier design unit 137 executes an optimization process for the set of thresholds "α" and "β" (S302).

[0100] In the present invention, whether or not the intestine from a sample is from a patient with pseudo-intestinal obstruction is identified using the value of the logarithm of the Mahalanobis distance "D1" of the intestine from the sample and the value of the distance "D2". In this identification, when D1≧α and D2≦β for a set of thresholds "α" and "β", the intestine from the sample is identified as the intestine from a patient with pseudo-intestinal obstruction, and otherwise it is identified as the intestine from a healthy subject.

[0101] In the optimization process, the classifier design unit 137 selects "1" person as a test sample from among the available samples of "n" people including a plurality of healthy subjects and patients with pseudo-intestinal obstruction according to the division of the leave-one-out method, and uses the remaining "n-1" people as design samples. Here, the test sample is not used in the optimization process. Next, the classifier design unit 137 distinguishes whether each intestinal tract is an intestinal tract from a patient with pseudo-intestinal obstruction while changing the candidate set of thresholds "α" and "β" for the intestinal tracts from the design samples of "n-1" people. That is, each of the intestinal tracts from the design samples of "n-1" people becomes an object of classification. Next, the classifier design unit 137 selects a set of thresholds "α" and "β" that distinguishes the intestinal tract from a patient with pseudo-intestinal obstruction with high accuracy from among the candidate set of thresholds. Next, the classifier design unit 137 repeats the selection of this set until all of the samples of "n" people are selected as test samples exactly once, that is, n times independently. Next, the classifier design unit 137 identifies the most frequently selected pair, that is, the most effective pair of thresholds, from a group of n different design samples of “n-1” people used in selecting pairs “n” times, as the optimal pair of thresholds “α” and “β”.

[0102] Next, the classifier design unit 137 stores the set of optimal thresholds "α" and "β" in the storage unit 12 as parameters of a classifier for identifying the intestines of patients with pseudo-intestinal obstruction (S303).

[0103] Identification and diagnosis processing FIG. 9 is a flowchart of the identification and diagnosis process (S4).

[0104] The "identification / diagnosis process (S4)" is a process for identifying the intestinal tract captured in the video (group of still images) of the patient (diagnosed person) to be diagnosed as either the intestinal tract from a patient with pseudo-ileus or the intestinal tract from a person without pseudo-ileus, i.e., identifying the intestinal tract as either a specific subject or a non-specific subject, and diagnosing the diagnosed person based on the identification result.

[0105] The following description will be given of an example in which it is determined whether the intestines captured in the image (group of still images) of a patient are those of a patient with pseudo-intestinal obstruction or those of a person not suffering from pseudo-intestinal obstruction (e.g., a healthy person). In the following description, it is assumed that the value of the logarithm "D1" of the Mahalanobis distance and the value of the distance "D2" for each intestine in the image (group of still images) of the patient are extracted in advance by a feature extraction process (S2) and stored in the storage unit 12.

[0106] First, the identification unit 138 selects an intestinal tract to be identified from among the intestinal tracts in the video (group of still images) of the subject, that is, from among the intestinal tracts from the subject (S401).

[0107] Next, the identification unit 138 reads out the value of the logarithm "D1" of the Mahalanobis distance for each intestinal tract from the subject and the value of the distance "D2" from the storage unit 12 (S402).

[0108] Next, the classifier 138 reads out from the storage unit 12 a set of optimal thresholds "α" and "β" which are parameters of a classifier for identifying whether or not the intestine is from a patient with pseudo-intestinal obstruction (S403).

[0109] Next, the identification unit 138 identifies the intestinal tract from the subject based on the logarithm "D1" value read in process S301, the distance "D2" value, and a classifier in which the set of optimal thresholds "α" and "β" read in process S303 are set as parameters (S404).

[0110] Next, the identification unit 138 determines whether or not there is an intestinal tract that has not been identified (unidentified) (S405). When there is an intestinal tract that has not been identified ("Y" in S405), the identification and diagnosis process (S4) returns to process S401.

[0111] On the other hand, when there is no unidentified intestinal tract ("N" in S405), the diagnosing unit 139 determines whether or not there is any intestinal tract identified as being from a patient with pseudo-intestinal obstruction in process S404 among all the intestinal tracts (S406).

[0112] When there is at least one intestine identified as being from a patient with pseudo-intestinal obstruction ("Y" in S406), the diagnosis unit 139 diagnoses that the subject is a patient with pseudo-intestinal obstruction (S407). On the other hand, when there is no intestine identified as being from a patient with pseudo-intestinal obstruction ("N" in S406), the diagnosis unit 139 diagnoses that the subject is not a patient with pseudo-intestinal obstruction (S408). The diagnosis result is displayed on the display unit 15, for example, and is stored in the memory unit 12.

[0113] Example: Next, an embodiment of the present invention will be described. The samples used in the following description are 18 images (still image group) consisting of images (still image group) of the intestines of 11 healthy subjects and images (still image group) of the intestines of 7 patients with pseudo-intestinal obstruction. In this embodiment, the imaging interval is 0.5 seconds, and the imaging period is 17 seconds (17.5 seconds if no still image is captured at 0 seconds).

[0114] First, the inventor of the present invention performed an evaluation amount calculation process (S1) and a feature extraction process (S2) for each of the selected intestines in the 18 cases, and extracted the logarithm of the Mahalanobis distance "D1" and the distance between the motion vector and the reference vector "D2" as features. Here, the number of intestines in each case was "1" to "5". Next, the inventor of the present invention performed a classifier design process (S3) using the selected intestines in the 18 cases to identify an optimal set of thresholds for identifying the intestines from the pseudointestinal obstruction patients with high accuracy. At this time, the optimal set of thresholds "α" and "β" characterized by the logarithm "D1" and the distance "D2" was "α: 1.10" and "β: 0.15". Next, the inventor of the present invention performed the classification and diagnosis process (S4) using a classifier with a set of optimal thresholds "α: 1.10" and "β: 0.15" as parameters, and diagnosed 18 cases, obtaining a diagnosis result of "sensitivity of 1.00 (= 7 cases / 7 cases)" and "specificity of 0.82 (9 cases / 11 cases)". Thus, the present invention can accurately classify whether or not a subject is in a specific state even when there is no periodic movement and even when there is no change in the size of the subject, and can diagnose patients with high sensitivity using the classification result. In this case, "sensitivity" means the rate at which patients with pseudo-obstruction are correctly diagnosed as such, and "specificity" means the rate at which patients who are not patients with pseudo-obstruction (or healthy subjects) are correctly diagnosed as patients who are not patients with pseudo-obstruction (or healthy subjects).

[0115] FIG. 10 is a graph showing the change over time in the evaluation amount calculated in an embodiment of the present invention, where (a) shows the evaluation amount for an intestine with a change in size (the intestine from a healthy subject), and (b) shows the evaluation amount for an intestine with little change in size (the intestine from a patient with pseudo-intestinal obstruction). The figure shows that the intestines with varying sizes (intestines from healthy subjects) are irregularly enlarged (J t >0.5)·Reduction(J t <0.5), whereas intestines with little change in size (intestines from patients with pseudo-obstruction) show almost no movement.

[0116] FIG. 11 is a graph in which the intestines from 18 cases used in the examples of the present invention are plotted on a plane using the values ​​(coordinate values) of the feature "D1" and the values ​​(coordinate values) of the feature "D2."

[0117] The horizontal axis (X-axis) of the figure is the logarithm of the Mahalanobis distance "D1", and the vertical axis (Y-axis) of the figure is the distance "D2" between the motion vector and the reference vector. This figure shows that the smaller the coordinate value on the Y-axis, the smaller the degree of change in intestinal size, and the larger the coordinate value on the X-axis, the further away from the distribution of intestinal tracts from healthy subjects. For example, "◆" in the figure is a plot of intestinal tracts from patients with ileus pseudo-obstruction (CIPO), and "〇": volunteers" is a plot of intestinal tracts from healthy subjects.

[0118] Summary According to the embodiment described above, the evaluation amount calculation unit 135 calculates the evaluation amount for each intestinal tract of the still image based on the above-mentioned formula (2). That is, the evaluation amount calculation unit 135 calculates the evaluation amount for each intestinal tract based on the ratio of the size of the subject in the still image (second still image) captured at time "t" to the sum of the size (short axis length) of the subject (intestinal tract) in the still image (first still image) captured at time "t-1" and the size (short axis length) of the subject (intestinal tract) in the still image (second still image) captured at time "t" after time "t-1". According to this configuration, since the ratio of the short axis lengths between two adjacent still images is used, the time-series change in the size of the subject is reflected in the calculation of the evaluation amount, and since the relative value (ratio) of the short axis length is used, the effect of the resolution (number of pixels) of the imaging device 2 is theoretically eliminated. That is, the evaluation amount calculated by the present device 1 is calculated based on the change in the actual size of the subject. As a result, the present device 1 can provide an evaluation amount that serves as an index for evaluating a change in the size of a subject, regardless of the presence or absence of periodicity (regularity) in the movement of the subject (intestinal tract).

[0119] According to the embodiment described above, the device 1 includes the identification unit 138 that identifies whether or not the subject (intestine) from a patient (diagnosed person) is the intestine from a patient with pseudointestinal obstruction. In other words, the device 1 includes the identification unit 138 that identifies the subject as either a specific subject whose size change is not normal or a non-specific subject whose size change is normal. The identification unit 138 identifies the subject based on a first feature (logarithm of Mahalanobis distance "D1") extracted based on the size (minor axis length) of the subject (intestine) and a second feature (distance "D2") extracted based on an evaluation amount of the change in size of the subject (intestine). According to this configuration, the identification unit 138 can identify the intestine based on the first feature reflecting the minor axis length (degree of abnormal enlargement) of the intestine and the second feature reflecting the degree of change in size of the intestine. Therefore, the present device 1 can determine the degree of change (enlargement / reduction) in the size of the subject based on the first feature using the size of the subject, and can determine the degree of change in the size of the subject based on the second feature using the evaluation amount of the subject. As a result, the present device 1 can distinguish the subject (intestinal tract) as either a specific subject (intestinal tract from a patient with pseudo-intestinal obstruction) or a non-specific subject (intestinal tract from a person without pseudo-intestinal obstruction) from two perspectives (features), namely, the degree of enlargement / reduction in the size of the subject and the degree of change in the size of the subject.

[0120] Furthermore, according to the embodiment described above, the device 1 includes a feature extraction unit 136 that extracts the first feature and the second feature. The feature extraction unit 136 extracts the mean "μ" and the variance "σ" of a distribution consisting of the average values ​​of the sizes (minor axis lengths) of the respective non-specific subjects (intestines from healthy individuals). 2" is estimated, the statistical distance (Mahalanobis distance) between the average value of the size of the subject from the subject and the above-mentioned distribution is calculated, and the first feature is extracted based on the statistical distance. The feature extraction unit 136 also defines a 34-dimensional motion vector whose components are the evaluation amount of the subject from the subject, and a reference vector whose components of the motion vector are reference amounts indicating that the size has not changed, and extracts the distance between the motion vector and the reference vector as the second feature. When the size of the subject is calculated based on the Euclidean distance (in this embodiment, it is calculated based on the number of pixels between two points that determine both ends of the short axis length), the calculated size is affected by the resolution (number of pixels) of the still image. That is, the calculation result of the size of the subject in the still image differs depending on the setting and type of the imaging device 2. However, according to the configuration in which the Mahalanobis distance is used, the average value of the size of the subject from the subject is normalized to a distribution consisting of the average values ​​of the sizes of multiple non-specific subjects (intestines from multiple healthy individuals), and the logarithm of the normalized value is used as the feature value. As a result, the present device 1 enables versatile subject identification and patient diagnosis that are not dependent on the resolution of a still image.

[0121] The size in the present invention is not limited to the minor axis length of the intestine. That is, for example, the size in the present invention may be the major axis length of the intestine or the ratio of the minor axis length to the major axis length, such as the circularity, or the circumferential length of the intestine.

[0122] In addition, the second feature in the present invention is not limited to the distance "D2" between the motion vector and the reference vector, and various distances can be adopted. That is, for example, the second feature in the present invention may be the average of the absolute values ​​of the differences between the evaluation amount and the reference value (0.5).

[0123] Furthermore, if the still image has high image quality, the present device does not need to include an image sharpening unit. That is, for example, the present device 1 may calculate the size of the subject directly from the still image instead of the sharpened image, or may calculate the size of the subject using the sharpened image that has been sharpened by another image processing device.

[0124] Furthermore, according to the above-described embodiment, the two still images captured in succession are two still images adjacent to each other in the video. Alternatively, if the imaging device is capable of high-speed shooting, the two still images captured in succession may be two still images that are consecutive at intervals of a certain number of frames in the video. That is, for example, the two still images may be two still images that are consecutive at intervals of two frames in the video.

[0125] Furthermore, the statistical distance in the present invention is not limited to the Mahalanobis distance as long as it is a statistical distance that is not dependent on scale.

[0126] Furthermore, according to the above-described embodiment, the present device 1 is configured by one computer. Instead of this, the present device may be configured by multiple computers. That is, for example, the present device may be configured by a group of multiple computers that function as the present device. Specifically, for example, the present device (group of computers) may be configured by a computer having a storage unit and a computer having a control unit that executes the present method. Also, for example, the multiple computers may have the functions of an image acquisition unit, an image display unit, an image sharpening unit, a size calculation unit, an evaluation amount calculation unit, a feature extraction unit, a classifier design unit, a classification unit, and a diagnosis unit in a distributed manner. In this case, the multiple computers that make up the group of computers may send and receive information through a communication line, or may transfer information using a portable storage medium. [Explanation of symbols]

[0127] 1. Information processing device 12 Storage section 134 Size Calculation Section 135 Evaluation Quantity Calculation Unit 136 Feature Extraction Unit 138 Identification Unit

Claims

1. An information processing device that calculates an evaluation amount that is an index for quantitatively evaluating a change in size of a subject in a plurality of still images captured at equal time intervals, A storage unit that stores a plurality of the still images; a size calculation unit that calculates the size of the subject captured in each of the still images stored in the storage unit; an evaluation amount calculation unit that calculates the evaluation amount for each of the subjects based on the size in each of the two still images captured in succession; and the subject is an animal organ, The two still images are A first still image; A second still image captured after the first still image; Including, the evaluation amount calculation unit calculates the evaluation amount for each of the objects based on a ratio of the size of the object in the second still image to a sum of the size of the object in the first still image and the size of the object in the second still image.

23. An information processing apparatus comprising:

2. a classification unit that classifies the subject into either a specific subject in which the change in size is abnormal or a non-specific subject in which the change in size is normal; and the identification unit identifies the subject based on a first feature of the subject extracted based on the size of the subject and / or a second feature of the subject extracted based on the evaluation amount of the subject.

2. The information processing device according to claim 1.

3. the first feature is extracted based on a statistical distance between a first average value which is an average value of the size of the subject and a distribution of second average values ​​which are average values ​​of the size of each of the plurality of non-specific subjects; 3. The information processing device according to claim 2.

4. a feature extraction unit that extracts the first feature and / or the second feature; and The feature extraction unit is estimating the mean and variance of the distribution of the second mean values; Calculating the statistical distance between the first mean value and the distribution of the second mean values; Extracting the first feature based on the statistical distance; and / or A multi-dimensional motion vector having the evaluation amount of the object as a component, and a reference vector having each component of the motion vector as a reference amount indicating that the size has not changed are defined; extracting a distance between the motion vector and the reference vector as the second feature; 4. The information processing device according to claim 3.

5. The statistical distance is the Mahalanobis distance, The first feature is a logarithm of the Mahalanobis distance.

5. The information processing device according to claim 4.

6. A computer is caused to function as the information processing device according to claim 1.

23. An information processing program comprising:

7. An information processing method executed by an information processing device that calculates an evaluation amount that is an index for quantitatively evaluating a change in size of a subject in a plurality of still images captured at equal time intervals, comprising: the subject is an animal organ, The information processing device includes: A storage unit that stores a plurality of the still images; Equipped with The plurality of still images are A first still image; A second still image captured after the first still image; Including, The information processing device, a size calculation step of calculating the size of the subject captured in each of the still images stored in the storage unit; an evaluation amount calculation step of calculating the evaluation amount for each of the objects based on the size in each of the first still image and the second still image; Including, the information processing device calculates the evaluation amount based on a ratio of the size of the subject in the second still image to a sum of the size of the subject in the first still image and the size of the subject in the second still image, in the evaluation amount calculation step; 23. An information processing method comprising:

Images