JP2020058472A - Image processing program, image processor, and image processing method - Google Patents

Abstract

To improve identification accuracy of a CT image.SOLUTION: A computer determines whether or not a second high absorption area similar to a first high absorption area is included at a prescribed position corresponding to the direction of gravity operating on the body parts in a CT image of each of a plurality of slices including two adjacent slices of the body parts (step 301). The first high absorption area is a high absorption area appearing in a CT image of a specific case. The computer determines that the second high absorption area is a high absorption area caused by a gravitational effect when the second high absorption area is included in the prescribed position inside the CT images of the plurality of slices (step 302).SELECTED DRAWING: Figure 3

Description

  The present invention relates to an image processing program, an image processing device, and an image processing method.

  In image diagnosis for diagnosing cases using computed tomography (CT) images, it is difficult to diagnose cases in which abnormal shadows are distributed over multiple areas, such as diffuse lung disease. It is said. When diagnosing such cases, doctors narrow down the candidates for disease names with reference to similar cases for which definite diagnosis has been performed in the past.

  However, the task of retrieving similar cases in the past takes time and is a heavy burden for doctors. Therefore, there has been proposed a technique that automatically searches CT images of similar cases from past cases and presents the search results to support the diagnostic work of a doctor (for example, see Non-Patent Document 1). .

  Regarding image diagnosis using CT images, a method of quantitatively evaluating CT images of interstitial pneumonia using a computer is also known (for example, see Non-Patent Document 2). A medical image diagnosis support device that considers CT value extraction error due to gravity, an image processing device that extracts a lump-shaped observation object, an automatic analysis device that analyzes a chest CT image, and a case image search device are also known ( See, for example, Patent Documents 1 to 4.

JP, 2007-289335, A JP, 2015-29860, A JP, 2016-174773, A JP, 2011-118543, A

"Development of similar case search technology utilizing AI in CT examination", [online], June 23, 2017, Fujitsu Laboratories Ltd. press release, [search on July 13, 2018], Internet <URL: http://pr.fujitsu.com/jp/news/2017/06/23.html> Tae Iwasawa, “Quantitative evaluation of CT images of interstitial pneumonia by computer”, Journal of Tomographic Imaging Research Society, Vol. 41, No. 2, pp. 67-76, August 2014

  In cases of diffuse lung disease, multiple types of abnormal shadows spread over the entire lung field in chest CT images. Therefore, in order to search for similar cases in the past, it is necessary to identify abnormal shadows and quantify their distribution. desirable. In the current similar case image retrieval technology, there are some problems regarding the identification of abnormal shadows, but one of them is the problem of erroneously identifying normal region image areas as abnormal shadows due to the gravity effect.

  In CT examinations, the patient is often in a supine position and CT imaging is often performed. In the supine position, the venous pressure and lymphatic pressure on the dorsal side of the lung floor at the lowest position are increased by gravity, and the volume of the lung peripheral interstitium is increased, so that the lung parenchyma is excluded. This phenomenon is called the gravity effect. In this case, in the CT image, the portion where the volume of the interstitium increases appears as a whitish high absorption region. The high absorption region is a region where the CT value, which is the pixel value of the CT image, is higher than that in the normal lung field.

  As described above, although the abnormal shadow does not actually exist, a high absorption region is generated at the back side position in the lung region in the CT image due to the influence of the gravity effect. In particular, when the patient is obese or lacks inspiration during CT imaging, the effect of the gravity effect is remarkable.

  On the other hand, one of the abnormal shadows due to lung disease, Ground Glass Opacity (GGO), also contains a high absorption region, and the image feature amount of the high absorption region generated by the gravity effect is the same as that of GGO. It is similar to the image feature amount. Therefore, if the high absorption region caused by the gravity effect is identified using the image identification model generated by machine learning, it may be erroneously identified as GGO. When the accuracy of identifying an abnormal shadow by the image identification model is reduced due to the erroneous identification of the high absorption region, the accuracy of searching similar cases is also reduced.

  Note that such a problem occurs not only when identifying an abnormal shadow on a chest CT image, but also when identifying an abnormal shadow on a CT image of another body part.

  In one aspect, the present invention aims to improve the identification accuracy of CT images.

In one proposal, the image processing program causes a computer to execute the following processing.
(1) In a computer tomographic image of each of a plurality of slices including two adjacent slices of a body part, the computer defines a first high absorption region at a predetermined position corresponding to the direction of gravity acting on the body part. It is determined whether a similar second high absorption region is included. The first high-absorption region is a high-absorption region that appears in a computed tomography image of a specific case.
(2) The computer determines that the second high-absorption region is a high-absorption region caused by the gravity effect when the second high-absorption region is included in a predetermined position in the computer tomographic image of a plurality of slices. .

  According to the embodiment, the identification accuracy of CT images can be improved.

It is a figure which shows a threshold value correction curve. It is a functional block diagram of an image processing apparatus. It is a flowchart of image processing. It is a functional block diagram which shows the specific example of an image processing apparatus. It is a figure which shows a chest CT image. It is a flowchart (the 1) which shows the specific example of image processing. It is a flowchart (the 2) which shows the specific example of image processing. It is a figure which shows the CT image divided into the some block. It is a figure which shows the change of average CT value. It is a block diagram of an information processing apparatus.

  Hereinafter, embodiments will be described in detail with reference to the drawings.

  The medical image diagnosis support device of Patent Document 1 obtains the change in CT value in the direction from the ventral side to the dorsal side of the lung field in one CT image, thereby determining the influence of the gravity effect on the CT value to reduce emphysema. It is reflected in the threshold for detection. First, the medical image diagnosis support apparatus divides the CT image of the lung field into a plurality of partial spaces (partial regions) in the direction from the ventral side to the dorsal side, and obtains an average CT value in each partial region.

  Next, the medical image diagnosis support apparatus plots the average CT value of each partial region along the direction from the ventral side to the dorsal side, creates a threshold correction curve, and passes the typical CT value of emphysema. A threshold setting curve is created by translating the threshold correction curve in parallel. Then, the medical image diagnosis support apparatus changes the threshold value at each position in the direction from the ventral side to the dorsal side according to the threshold value setting curve.

  However, the average CT value of the partial region including the high absorption region due to the gravity effect may be close to the average CT value of the partial region including GGO due to lung disease. Therefore, the threshold correction curve may be similar between the CT image including the high absorption region caused by the gravity effect and the CT image including GGO.

  FIG. 1 shows an example of a threshold correction curve showing a change in average CT value. FIG. 1A shows a threshold correction curve 101 of a CT image including GGO due to interstitial pneumonia, and FIG. 1B shows a threshold correction curve of a CT image including a high absorption region due to the gravity effect. The example of 102 is shown. The horizontal axis represents the position in the direction from the ventral side to the dorsal side, and the vertical axis represents the average CT value of the partial region. The upper limit of the CT value of the normal lung of an adult is about -700HU.

  In either case of the threshold correction curve 101 or the threshold correction curve 102, since the average CT value at the position P1 on the back side exceeds -700HU, it can be seen that the high absorption region exists at the position P1. However, since the threshold correction curve 101 and the threshold correction curve 102 are very similar to each other, even if the change of the average CT value in one CT image is examined, there is a high absorption region including GGO and a high absorption region caused by the gravity effect. It is difficult to distinguish between.

  FIG. 2 shows an example of the functional configuration of the image processing apparatus according to the embodiment. The image processing apparatus 201 of FIG. 2 includes a storage unit 211 and a determination unit 212. The storage unit 211 stores CT images 221 of a plurality of slices including two adjacent slices of the body part. The determination unit 212 performs image processing on the CT image 221.

  FIG. 3 is a flowchart showing an example of image processing performed by the image processing apparatus 201 of FIG. The determination unit 212, in the CT image 221 of each of the plurality of slices, at a predetermined position corresponding to the direction of gravity acting on the body part, a second high absorption region similar to the first high absorption region that appears in the CT image of the specific case. It is determined whether or not the absorption region is included (step 301).

  When the second high absorption region is included at a predetermined position in the CT images of the plurality of slices, the determination unit 212 determines that the second high absorption region is the high absorption region caused by the gravity effect (step 302). ).

  According to the image processing apparatus 201 of FIG. 2, the accuracy of image diagnosis based on CT images can be improved.

  FIG. 4 shows a specific example of the image processing apparatus 201 of FIG. The image processing device 401 of FIG. 4 includes a storage unit 411, an acquisition unit 412, a preprocessing unit 413, a determination unit 414, and an output unit 415. The storage unit 411 and the determination unit 414 correspond to the storage unit 211 and the determination unit 212 of FIG. 2, respectively.

  The CT device 402 is installed in a medical institution and takes a CT image of a patient. The storage device 403 stores CT images of a plurality of patients captured by the CT device 402. The image processing device 401, the CT device 402, and the storage device 403 can communicate with each other via a communication network.

  The acquisition unit 412 acquires a plurality of CT images of the patient to be diagnosed from the CT device 402 or the storage device 403, and stores the acquired CT images in the storage unit 411 as CT images 421. The CT image 421 corresponds to the CT image 221 of FIG. 2 and is a CT image of a plurality of consecutive slices of a predetermined body part.

  The predetermined body part may be the head, chest, abdomen, or the like. Hereinafter, a case where the predetermined body part is the chest and each CT image 421 includes an image of the lung field will be described. When CT imaging is performed with the patient in the supine position, the predetermined position corresponding to the direction of gravity acting on the chest is the position on the back side of the lung field.

  However, it is assumed that each CT image 421 includes either a high-absorption region corresponding to an abnormal shadow due to lung disease or a high-absorption region due to gravity, and both are not included at the same time. To do. For example, in a CT image in which GGO is included on the back side of the lung field and a high-absorption region due to the gravity effect is also included, the GGO and the high-absorption region due to the gravity effect are distinguished. Is difficult to do. However, in this case, since GGO is included on the back side of the lung field, even if all the high-absorption regions are identified as GGO by the image identification model, it does not significantly affect the search accuracy of similar cases.

  As described above, since the image feature amount of the high absorption region caused by the gravity effect is similar to the image feature amount of GGO, the high absorption region caused by the gravity effect may be erroneously identified as GGO. is there. On the other hand, the image feature amount of the abnormal shadow other than GGO does not necessarily resemble the image feature amount of the high absorption region due to the gravity effect, and thus the problem of misidentification does not occur.

  Therefore, the preprocessing unit 413 checks whether or not the CT image 421 includes an abnormal shadow other than GGO on the back side of the lung field. For example, the pre-processing unit 413 can determine whether or not the CT image 421 includes an abnormal shadow other than GGO using the image identification model generated by machine learning. In this case, the image identification model is generated by performing machine learning on a plurality of CT images including abnormal shadows other than GGO using the similar case search technique of Non-Patent Document 1.

  When an abnormal shadow other than GGO is included on the back side of the lung field of the CT image 421, the preprocessing unit 413 excludes the CT image 421 from the determination targets of the determination unit 414. Thereby, the determination unit 414 can perform the image processing with only the high absorption region including GGO or the high absorption region caused by the gravity effect as the determination target.

  FIG. 5 shows an example of a chest CT image including a high absorption region. FIG. 5A shows an example of a CT image including a high absorption region due to the gravity effect. In the CT images of a plurality of slices taken continuously, when the high absorption region 501 and the high absorption region 502 are generated by the gravity effect, gravity acts equally on the upper lung field, the middle lung field, and the lower lung field. . Therefore, similar high absorption regions are observed in the CT images of the slices before and after the CT image including the high absorption region on the back side.

  FIG. 5B shows an example of a CT image including a high absorption region of GGO. In a CT image of a plurality of slices taken continuously, when the high absorption region 511 and the high absorption region 512 are generated by GGO, the positions where GGO appears in the upper lung field, the middle lung field, and the lower lung field, respectively. different. Therefore, the positions and areas of the high-absorption regions change in the CT images of the slices before and after the CT image including the high-absorption region on the back side.

  As described above, the high-absorption region due to the gravitational effect is observed continuously on the dorsal side from the upper lung field to the lower lung field, and the high-absorption region in the CT image is gravitated. It is possible to determine whether or not it is due to the effect.

  Therefore, the determination unit 414 determines whether or not a high absorption region similar to the GGO-containing high absorption region is included in the position on the back side of the lung field in the CT images 421 of each of the plurality of slices that are continuously captured. To determine. Hereinafter, a high absorption region similar to the high absorption region containing GGO may be referred to as a similar high absorption region.

  In the CT image 421 of each of the plurality of slices, when the similar high absorption region is included in the position on the back side of the lung field, the determination unit 414 determines that the similar high absorption region is the high absorption region caused by the gravity effect. Judge that there is. On the other hand, when the CT image 421 of the slice adjacent to the CT image 421 including the similar high absorption region does not include the similar high absorption region at the position on the back side of the lung field, the determination unit 414 determines that similar high absorption region. Is not a high absorption region due to the gravity effect.

  Then, the determination unit 414 generates a determination result 422 indicating whether or not the region is a high absorption region due to the gravity effect, stores the determination result 422 in the storage unit 411, and the output unit 415 outputs the determination result 422.

  According to the image processing device 401 of FIG. 4, when a high absorption region is included on the back side of the lung field in each CT image 421, it is determined whether or not the high absorption region is caused by the gravity effect. Can be determined. This reduces the possibility that the high-absorption region caused by the gravity effect will be erroneously identified as GGO by the image identification model generated by machine learning, and the CT image identification accuracy is improved. Therefore, the search accuracy of the similar case search process using the image identification model is also improved.

For example, the determination unit 414 identifies the high absorption region that may be generated by the gravity effect, using the following conditions.
(C1) In the CT image 421 of each of the plurality of slices, the image feature amount of the region existing on the back side of the lung field corresponds to the image feature amount of the high absorption region including GGO.
(C2) The statistical value of the CT value of the region existing in the direction opposite to the direction of gravity is smaller than the predetermined value with reference to the position on the back side.

  The determination unit 414 can use the image identification model generated by machine learning to check whether the image feature amount of the region existing at the position on the back side of the lung field satisfies the condition (C1). In this case, the image identification model is generated by performing machine learning on a plurality of CT images including GGO using the similar case search technique of Non-Patent Document 1 or the like.

  In the condition (C2), the region existing in the direction opposite to the direction of gravity may be a region other than the back side of the lung field, and the predetermined value is the CT value of normal adult lung. May be the upper limit value of.

  When the condition (C1) and the condition (C2) are satisfied, the region existing on the dorsal side is a similar high absorption region existing only on the dorsal side of the lung field. Therefore, by using these conditions, it is possible to identify the high absorption region that may be generated by the gravity effect.

  6A and 6B are flowcharts showing a specific example of image processing performed by the image processing apparatus 401 in FIG. 4 using the condition (C1) and the condition (C2). First, the acquisition unit 412 acquires the CT images 421 of N slices taken continuously from the CT device 402 or the storage device 403 (step 601). When the dorsal side of the lung field of the CT image 421 includes an abnormal shadow other than GGO, the preprocessing unit 413 excludes the CT image 421 from the determination targets of the determination unit 414.

  When an abnormal shadow other than GGO is not included on the back side of the lung field of the CT image 421, the determination unit 414 sets 1 to the control variable i that designates one slice (step 602), and the i-th The CT image 421 of the slice is divided into blocks of a predetermined size (step 603). The size of the block may be 16 × 16 pixels or 8 × 8 pixels. Further, the shape of the block may be square or rectangular.

  FIG. 7 shows an example of the CT image 421 divided into a plurality of blocks. In this example, the chest CT image including the lung field image is divided into 260 blocks. In this case, the horizontal direction of the CT image 421 corresponds to the left-right direction of the body, and the vertical direction corresponds to the front-back direction of the body. In the vertical direction of the CT image 421, the direction from the ventral side to the dorsal side represents the direction of gravity, and the direction from the dorsal side to the ventral side represents the direction opposite to gravity.

  Next, the determination unit 414 identifies whether or not each block corresponds to the similar high absorption region using the image identification model (step 604). When the image feature amount of the block corresponds to the image feature amount of the high absorption region including GGO, it is determined that the block corresponds to the similar high absorption region.

  Next, the determination unit 414 checks whether or not a block identified as corresponding to the similar high absorption region exists at the position on the back side in the CT image 421 (step 605). For example, as the position on the back side, a region within a predetermined length from the lower end (back) in the CT image 421 can be used. The predetermined length may be 1/3 to 1/5 of the length from the upper end (antinode) to the lower end of the CT image 421.

  In the example of FIG. 7, 1/4 of the length from the upper end to the lower end is used as the predetermined length, and 17 blocks included in the region 701 correspond to the similar high absorption region. In this example, for simplification, a similar high-absorption region is shown only in the left lung, but the high-absorption region due to the gravitational effect occurs in both left and right lungs as shown in FIG. 5 (a). I often do it.

  When there is a block corresponding to the similar high absorption region at the position on the back side (step 605, YES), the determination unit 414 determines the statistical value V of the CT value for the block at a position other than the back side in the row including the block. Is calculated (step 606).

  For example, the determination unit 414 calculates the statistical value of CT values of a plurality of pixels included in each block at positions other than the back side, and performs statistical processing on the statistical values of those blocks to obtain the statistical value V. be able to. The statistical value of each block may be an average value, a median value, a mode value, a maximum value, a minimum value, a sum, or the like. Further, the statistical processing on the statistical value may be processing for obtaining an average value, a median value, a mode value, a maximum value, a minimum value, a summation, and the like.

  FIG. 8 shows an example of changes in the average CT value in a column including blocks corresponding to similar high absorption regions. The horizontal axis represents the position in the direction from the ventral side to the dorsal side, and the vertical axis represents the average CT value of each block. A curve 801 shows the average CT value of each block included in the column 702 of FIG. 7. -700HU corresponds to the upper limit of the CT value of the normal lung of an adult.

  In this example, the average CT value of the blocks at the back side positions exceeds -700HU, and the average CT value of the blocks at positions other than the back side is approximately -700HU or less. As described above, when only the average CT value of the block on the back side exceeds the upper limit value, there is a possibility that the high absorption region is generated due to the gravity effect.

  Therefore, the determination unit 414 compares the statistical value V of the blocks at positions other than the back side with a predetermined value (step 607). For example, when the statistical value V is the average value of the average CT values for each block, the predetermined value may be -700HU. The processing of step 606 and step 607 is performed for all columns in which blocks corresponding to similar high absorption regions exist at the position on the back side.

  When the statistical value V is less than or equal to the predetermined value for all columns (step 607, YES), it can be seen that only the back side of the CT image 421 of the i-th slice includes the similar high absorption region. Therefore, the determination unit 414 determines that the CT image 421 may include a high absorption region due to the gravity effect, and increments i by 1 to determine that the slice adjacent to the current slice is The CT image 421 is designated (step 608). Then, the determination unit 414 compares i with N (step 609).

  When i is N or less (step 609, NO), the determination unit 414 repeats the processing from step 603. Thereby, it is checked whether or not the CT image 421 of the slice adjacent to the current slice may include a high absorption region due to the gravity effect.

  Then, when i exceeds N (step 609, YES), it can be seen that a similar high absorption region is included on the back side of each of the CT images 421 of N slices. Therefore, the determination unit 414 generates the determination result 422 indicating that the CT image 421 includes the high absorption region due to the gravity effect (step 610), and the output unit 415 outputs the determination result 422 ( Step 611).

  When the block corresponding to the similar high absorption region does not exist at the position on the back side (step 605, NO), the determination unit 414 determines that the CT image 421 does not include the high absorption region caused by the gravity effect. The result 422 is generated (step 612). Then, the output unit 415 outputs the determination result 422 (step 611).

  If the statistical value V is larger than the predetermined value for any of the columns (step 607, NO), the determination unit 414 determines that the similar high absorption region included in the CT image 421 is the high absorption region including GGO. . Therefore, the determination unit 414 generates the determination result 422 indicating that the CT image 421 does not include the high absorption region due to the gravity effect (step 610), and the output unit 415 outputs the determination result 422 ( Step 611).

  According to the image processing of FIGS. 6A and 6B, by dividing the CT image 421 into a plurality of blocks, it is checked for each block whether or not the condition (C1) is satisfied, and the block that satisfies the condition (C1) is checked. It is checked whether or not the condition (C2) is satisfied for each column including. Further, by confirming that the CT images 421 of the adjacent slices also satisfy the condition (C1) and the condition (C2), it can be seen that the high absorption regions due to the gravity effect continuously exist between the slices. Further, by appropriately setting the block size, it is possible to efficiently obtain the highly accurate determination result 422.

  For example, as the N slices to be image-processed, continuous slices corresponding to a portion longer than the length of the lower lung field are used. Since there are cases such as GGO that are significantly seen on the dorsal side of the lower lung field, such GGO is caused by the gravitational effect only in about 1/3 of the length from the upper lung field to the lower lung field. May be erroneously identified as a high absorption region. For example, such a misidentification can be prevented by using a slice covering a part of 2/3 of the length from the upper lung field to the lower lung field.

  The configurations of the image processing apparatus 201 in FIG. 2 and the image processing apparatus 401 in FIG. 4 are merely examples, and some of the constituent elements may be omitted or changed according to the use or conditions of the image processing apparatus. For example, in the image processing device 401 of FIG. 4, when the CT image 421 is stored in the storage unit 411 in advance, the acquisition unit 412 can be omitted. When the CT image 421 does not include an abnormal shadow other than GGO, the preprocessing unit 413 can be omitted, and when it is not necessary to output the determination result 422, the output unit 415 can be omitted.

  The flowcharts of FIGS. 3, 6A, and 6B are merely examples, and some of the processing may be omitted or changed depending on the configuration or conditions of the image processing apparatus. For example, in the image processing of FIGS. 6A and 6B, when the CT image 421 is stored in the storage unit 411 in advance, the processing of step 601 can be omitted. When it is not necessary to output the determination result 422, the process of step 611 can be omitted.

  When CT imaging is performed with the patient in the prone position, the predetermined position corresponding to the direction of gravity acting on the chest is the ventral position of the lung field. Therefore, in step 605, the determination unit 414 checks whether or not there is a block identified as corresponding to the similar high absorption region at the ventral position in the CT image 421. In this case, in step 606, the determination unit 414 calculates the statistical value V of the CT value for the block at a position other than the ventral side in the row including the block corresponding to the similar high absorption region.

  The threshold correction curves shown in FIGS. 1A and 1B are merely examples, and the threshold correction curves change according to the CT image to be taken. The CT images shown in FIGS. 5A, 5B, and 7 are merely examples, and the CT image changes depending on the patient and body part to be imaged. The change in the average CT value shown in FIG. 8 is merely an example, and the change in the average CT value changes according to the CT image to be taken.

  FIG. 9 shows a configuration example of an information processing apparatus (computer) used as the image processing apparatus 201 of FIG. 2 and the image processing apparatus 401 of FIG. The information processing apparatus of FIG. 9 includes a CPU (Central Processing Unit) 901, a memory 902, an input device 903, an output device 904, an auxiliary storage device 905, a medium drive device 906, and a network connection device 907. These components are connected to each other by a bus 908. The CT device 402 and the storage device 403 in FIG. 4 may be connected to the network connection device 907.

  The memory 902 is, for example, a semiconductor memory such as a ROM (Read Only Memory), a RAM (Random Access Memory), and a flash memory, and stores programs and data used for processing. The memory 902 can be used as the storage unit 211 of FIG. 2 or the storage unit 411 of FIG.

  The CPU 901 (processor) operates as the determination unit 212 in FIG. 2, the acquisition unit 412 in FIG. 4, the preprocessing unit 413, and the determination unit 414 by executing the program using the memory 902, for example.

  The input device 903 is, for example, a keyboard, a pointing device, or the like, and is used for inputting an instruction or information from an operator or a user. The output device 904 is, for example, a display device, a printer, a speaker, or the like, and is used for inquiring or instructing an operator or a user and outputting a processing result. The output device 904 can be used as the output unit 415 of FIG. The processing result may be the determination result 422.

  The auxiliary storage device 905 is, for example, a magnetic disk device, an optical disk device, a magneto-optical disk device, a tape device, or the like. The auxiliary storage device 905 may be a hard disk drive or a flash memory. The information processing apparatus can store the program and data in the auxiliary storage device 905 and load them into the memory 902 for use. The auxiliary storage device 905 can be used as the storage unit 211 of FIG. 2 or the storage unit 411 of FIG.

  The medium driving device 906 drives a portable recording medium 909 to access the recorded contents. The portable recording medium 909 is a memory device, a flexible disk, an optical disk, a magneto-optical disk, or the like. The portable recording medium 909 may be a CD-ROM (Compact Disk Read Only Memory), a DVD (Digital Versatile Disk), a USB (Universal Serial Bus) memory, or the like. An operator or a user can store a program and data in this portable recording medium 909 and load them into the memory 902 for use.

  As described above, a computer-readable recording medium that stores programs and data used for processing is a physical (non-transitory) recording medium such as the memory 902, the auxiliary storage device 905, or the portable recording medium 909. It is a medium.

  The network connection device 907 is a communication interface circuit that is connected to a communication network such as a LAN (Local Area Network) and a WAN (Wide Area Network) and performs data conversion accompanying communication. The information processing device can receive a program and data from an external device via the network connection device 907, load them into the memory 902, and use them. The network connection device 907 can be used as the output unit 415 of FIG.

  The information processing device can also receive the CT image 421 and the processing request from the user terminal via the network connection device 907, and can transmit the determination result 422 to the user terminal.

  Note that the information processing device does not need to include all the constituent elements in FIG. 9, and it is possible to omit some of the constituent elements depending on the application or the condition. For example, when the information processing device receives a processing request from the user terminal, the input device 903 and the output device 904 may be omitted. When the portable recording medium 909 or the communication network is not used, the medium driving device 906 or the network connection device 907 may be omitted.

  Although the disclosed embodiments and their advantages have been described in detail, those skilled in the art can make various changes, additions, and omissions without departing from the scope of the invention explicitly described in the claims. Let's do it.

Regarding the embodiment described with reference to FIGS. 1 to 9, the following supplementary notes are further disclosed.
(Appendix 1)
In the computed tomography image of each of a plurality of slices including two adjacent slices of the body part, the first to appear in the computed tomography image of a specific case at a predetermined position corresponding to the direction of gravity acting on the body part. It is determined whether or not a second high absorption region similar to the first high absorption region is included,
When the second high absorption region is included in the predetermined position in the computer tomographic image of the plurality of slices, it is determined that the second high absorption region is a high absorption region due to a gravity effect.
An image processing program that causes a computer to execute processing.
(Appendix 2)
In the computer, the image feature amount of the region existing at the predetermined position in the computer tomographic image of each of the plurality of slices corresponds to the image feature amount of the first high absorption region, and the predetermined position is used as a reference. When the statistical value of the pixel value of the region existing in the direction opposite to the direction of gravity is smaller than a predetermined value, it is determined that the second high absorption region is included in the predetermined position. The image processing program according to attachment 1.
(Appendix 3)
The computer divides the computer tomography image of each of the plurality of slices into a plurality of blocks, and the image feature amount of the block existing at the predetermined position among the plurality of blocks is the image feature amount of the first high absorption region. If the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region, in the column including the block existing at the predetermined position, A feature is that a statistic value of a pixel value of a block other than a block existing at a predetermined position is obtained, and when the statistic value is smaller than a predetermined value, it is determined that the second high absorption region is included at the predetermined position The image processing program according to Appendix 1.
(Appendix 4)
The computer may determine whether the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region using the image identification model generated by machine learning. An image processing program according to Supplementary Note 3 characterized by the above.
(Appendix 5)
The body part is the chest, the computed tomography image of each of the plurality of slices includes an image of the lung field, the predetermined position is a position on the dorsal side of the lung field, the specific case is 5. The image processing program according to any one of appendices 1 to 4, wherein the image processing program is a lung disease, and the first high absorption region is a region including a ground glass shadow.
(Appendix 6)
6. The image processing program according to appendix 5, wherein the plurality of slices are continuous slices corresponding to a portion of the lung field that is longer than the length of the lower lung field.
(Appendix 7)
A storage unit for storing computer tomographic images of a plurality of slices including two adjacent slices of the body part;
In the computed tomography image of each of the plurality of slices, at a predetermined position corresponding to the direction of gravity acting on the body part, a second high absorption region similar to the first high absorption region that appears in the computed tomography image of a specific case. It is determined whether or not an absorption region is included, and when the second high absorption region is included at the predetermined position in the computer tomographic images of the plurality of slices, the second high absorption region has a gravity effect. A determination unit that determines that the high absorption region due to,
An image processing apparatus comprising:
(Appendix 8)
The determination unit is such that the image feature amount of the region existing at the predetermined position in the computer tomographic image of each of the plurality of slices corresponds to the image feature amount of the first high absorption region, and the predetermined position is used as a reference. When the statistical value of the pixel value of the area existing in the direction opposite to the direction of gravity is smaller than a predetermined value, it is determined that the second high absorption area is included in the predetermined position. The image processing device according to attachment 7.
(Appendix 9)
The determination unit divides the computer tomographic image of each of the plurality of slices into a plurality of blocks, and the image feature amount of a block existing at the predetermined position among the plurality of blocks is an image feature of the first high absorption region. If the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region, in a column including the block existing at the predetermined position, A statistical value of pixel values of blocks other than the block existing at the predetermined position is obtained, and when the statistical value is smaller than a predetermined value, it is determined that the second high absorption region is included at the predetermined position. The image processing device according to Supplementary Note 7.
(Appendix 10)
The determination unit determines whether or not the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region, by using the image identification model generated by machine learning. The image processing apparatus according to appendix 9, characterized in that.
(Appendix 11)
An image processing method executed by a computer,
The computer is
In the computed tomography image of each of a plurality of slices including two adjacent slices of the body part, the first to appear in the computed tomography image of a specific case at a predetermined position corresponding to the direction of gravity acting on the body part. It is determined whether or not a second high absorption region similar to the first high absorption region is included,
When the second high absorption region is included in the predetermined position in the computer tomographic image of the plurality of slices, it is determined that the second high absorption region is a high absorption region due to a gravity effect.
An image processing method characterized by the above.
(Appendix 12)
In the computer, the image feature amount of the region existing at the predetermined position in the computer tomographic image of each of the plurality of slices corresponds to the image feature amount of the first high absorption region, and the predetermined position is used as a reference. When the statistical value of the pixel value of the region existing in the direction opposite to the direction of gravity is smaller than a predetermined value, it is determined that the second high absorption region is included in the predetermined position. The image processing method according to attachment 11.
(Appendix 13)
The computer divides the computer tomography image of each of the plurality of slices into a plurality of blocks, and the image feature amount of the block existing at the predetermined position among the plurality of blocks is the image feature amount of the first high absorption region. If the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region, in the column including the block existing at the predetermined position, A feature is that a statistic value of a pixel value of a block other than a block existing at a predetermined position is obtained, and when the statistic value is smaller than a predetermined value, it is determined that the second high absorption region is included at the predetermined position The image processing method according to appendix 11.
(Appendix 14)
The computer may determine whether the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region using the image identification model generated by machine learning. 14. The image processing method as described in Supplementary Note 13.

101, 102 threshold correction curve 201, 401 image processing device 211, 411 storage unit 212, 414 determination unit 221, 421 CT image 402 CT device 403 storage device 412 acquisition unit 413 pre-processing unit 415 output unit 422 determination result 501, 502, 511, 512 high absorption area 701 area 702 rows 801 curve 901 CPU
902 memory 903 input device 904 output device 905 auxiliary storage device 906 medium drive device 907 network connection device 908 bus 909 portable recording medium

Claims (5)

In the computed tomography image of each of a plurality of slices including two adjacent slices of the body part, the first to appear in the computed tomography image of a specific case at a predetermined position corresponding to the direction of gravity acting on the body part. It is determined whether or not a second high absorption region similar to the first high absorption region is included,
When the second high absorption region is included in the predetermined position in the computer tomographic image of the plurality of slices, it is determined that the second high absorption region is a high absorption region due to a gravity effect.
An image processing program that causes a computer to execute processing.   In the computer, the image feature amount of the region existing at the predetermined position in the computer tomographic image of each of the plurality of slices corresponds to the image feature amount of the first high absorption region, and the predetermined position is used as a reference. When the statistical value of the pixel value of the region existing in the direction opposite to the direction of gravity is smaller than a predetermined value, it is determined that the second high absorption region is included in the predetermined position. The image processing program according to attachment 1.   The computer divides the computer tomography image of each of the plurality of slices into a plurality of blocks, and the image feature amount of the block existing at the predetermined position among the plurality of blocks is the image feature amount of the first high absorption region. If the image feature amount of the block existing at the predetermined position corresponds to the image feature amount of the first high absorption region, in the column including the block existing at the predetermined position, A feature is that a statistic value of a pixel value of a block other than a block existing at a predetermined position is obtained, and when the statistic value is smaller than a predetermined value, it is determined that the second high absorption region is included at the predetermined position The image processing program according to Appendix 1. A storage unit for storing computer tomographic images of a plurality of slices including two adjacent slices of the body part;
In the computed tomography image of each of the plurality of slices, at a predetermined position corresponding to the direction of gravity acting on the body part, a second high absorption region similar to the first high absorption region that appears in the computed tomography image of a specific case. It is determined whether or not an absorption region is included, and when the second high absorption region is included at the predetermined position in the computer tomographic images of the plurality of slices, the second high absorption region has a gravity effect. A determination unit that determines that the high absorption region due to,
An image processing apparatus comprising: An image processing method executed by a computer,
The computer is
In the computed tomography image of each of a plurality of slices including two adjacent slices of the body part, the first to appear in the computed tomography image of a specific case at a predetermined position corresponding to the direction of gravity acting on the body part. It is determined whether or not a second high absorption region similar to the first high absorption region is included,
When the second high absorption region is included in the predetermined position in the computer tomographic image of the plurality of slices, it is determined that the second high absorption region is a high absorption region due to a gravity effect.
An image processing method characterized by the above.

Images