JP2019092542A - Medical image processor and medical image processing method - Google Patents

Abstract

To provide a medical image processor capable of collecting easily many medical images for learning; and to provide a medical image processing method.SOLUTION: A medical image processor in an embodiment includes an area extraction part for extracting an area to which finding information is imparted from a three-dimensional medical image, and an acquisition part for acquiring a medical image cut in a direction crossing the area extracted by the area extraction part, and including a part of the finding information from the three-dimensional medical image.SELECTED DRAWING: Figure 1

Description

  Embodiments of the present invention relate to a medical image processing apparatus and a medical image processing method.

  In the field of image recognition and speech recognition, machine learning such as deep learning has been increasingly used. Deep learning is machine learning using multi-layered neural networks. For example, in the field of image recognition, it is known that a learning method using a convolutional neural network (hereinafter referred to as CNN), which is one of deep learning, shows much higher performance than conventional methods. ing. CNN (also called ConvNet) automatically extracts feature structures by learning image data for learning, and determines whether or not the input image has the same characteristics as the image data for learning Can.

  By the way, in order to obtain highly accurate determination results in deep learning, a huge number of learning image data are required. For example, it is said that about 10,000 sheets of image data are required to achieve an accuracy of about 80%, and about 10,000 sheets to achieve an accuracy of 90%. For example, if the learning target is an easily obtainable image such as an animal or a car, a large number of images can be easily collected via a public electronic network such as the Internet.

  Here, it is considered to use deep learning to detect an affected area included in a medical image. In this case, a large number of medical images including the affected area are collected and learned by the CNN. However, regardless of the presence or absence of the affected area, the medical image is not a common image on the network. For this reason, it is difficult to collect sufficient numbers of medical images including the affected area in deep learning to obtain a highly accurate determination result. Further, in order to extract the feature structure with high accuracy, it is preferable to finely classify medical image data to be learned on the basis of a region of a medical image, a physique and an age of a subject, and to learn for each classification. However, when classifying medical image data to be learned, a large amount of learning medical image data must be collected for each classification. For this reason, the more finely they are classified, the more difficult it becomes to collect medical image data for learning.

Unexamined-Japanese-Patent No. 2015-80720

  The problem to be solved by the present invention is to easily collect many medical images for learning.

  A medical image processing apparatus according to an embodiment includes an area extraction unit and an acquisition unit. The area extraction unit extracts an area to which the finding information is added from the three-dimensional medical image. The acquisition unit is cut in a direction intersecting the area extracted by the area extraction unit, and acquires a medical image including a part of the finding information from the three-dimensional medical image.

FIG. 1 is a block diagram showing a configuration example of a medical image processing apparatus according to a first embodiment. The figure for demonstrating an example of the original data of the image data for learning. (A) is explanatory drawing which shows an example of a DICOM (Digital Imaging and Communications in Medicine) tag and RIS information, (b) is a figure for demonstrating an example of an interpretation report. FIG. 6 is a flowchart showing an example of a schematic procedure when collecting a large number of medical image data including an affected part for machine learning easily by the processor of the processing circuit shown in FIG. 1; FIG. Explanatory drawing which shows an example of a mode that annotation attached tomographic image data are extracted from several tomographic image data which comprise a volume image. Explanatory drawing which shows an example of the cutting-out direction z1 set with respect to the annotation attachment tomographic image data extracted in FIG. Explanatory drawing which shows an example of several MPR image data containing the annotation produced | generated along extraction direction z1 shown in FIG. The flowchart which shows an example of the procedure at the time of collecting many images for study while setting common cutting-out direction z1 to a plurality of original data sets. The figure for demonstrating a mode that common cutting-out direction z1 is set to several source data sets. FIG. 10 is a block diagram showing an example of the configuration of a medical image processing apparatus according to a third embodiment. 5 is a flowchart showing an example of a procedure for classifying MPR image data including an annotation 34 according to the size of an organ and the size of an annotation. Explanatory drawing which shows an example of the classification by a classification function.

  A medical image processing apparatus and a medical image processing method according to an embodiment of the present invention will be described with reference to the attached drawings.

  Medical image data for learning generated by the medical image processing apparatus 10 according to the present embodiment is, for example, CNN (Convolutional Neural Network), Convolutional Deep Belief Network (CDBN), Reconfigurable Topographic Independent Component It is used for machine learning such as analysis (TICA: Topographic Independent Component Analysis).

(Embodiment 1)
The medical image processing apparatus 10 includes an input circuit 11, a display 12, a memory circuit 13, a network connection circuit 14, and a processing circuit 15, as shown in FIG.

  The input circuit 11 is constituted by a general input device such as a track ball, a switch button, a mouse, a keyboard, and a ten key, and outputs an operation input signal corresponding to the user's operation to the processing circuit 15. The display 12 is configured by a general display output device such as a liquid crystal display or an OLED (Organic Light Emitting Diode) display.

  The storage circuit 13 has a configuration including a processor-readable storage medium such as a magnetic or optical storage medium or a semiconductor memory.

  For example, the storage circuit 13 stores a volume image (medical three-dimensional image) supplied from the modality 100. The modality 100 is an apparatus capable of generating a volume image based on projection data obtained by imaging a subject (patient). The modality 100 includes, for example, a medical image diagnostic apparatus such as an X-ray computed tomography (CT) apparatus, a magnetic resonance imaging (MRI) apparatus, an X-ray diagnostic apparatus, and a nuclear medicine diagnostic apparatus.

  The network connection circuit 14 implements various information communication protocols according to the form of the network 101. The network connection circuit 14 connects with other electric devices via the network 101 in accordance with the various protocols. The network 101 generally means an information communication network using telecommunications technology, and in addition to a wireless / wired LAN such as a hospital backbone LAN (Local Area Network) and the Internet network, a telephone communication network, an optical fiber communication network, and cable communication. Including networks and satellite communication networks etc.

  The medical image processing apparatus 10 may receive medical image data from the modality 100 or the medical image storage apparatus 102 connected via the network 101. The volume image received via the network 101 is stored in the storage circuit 13. The medical image processing apparatus 10 may be included in the modality 100.

  The medical image storage apparatus 102 is, for example, a server for image storage provided in a PACS (Picture Archiving and Communication System), and is a volume image generated by the modality 100 connected via the network 101. Remember.

  Further, the medical image storage apparatus 102 stores the information of the annotation and the volume image in association with each other. Further, it is assumed that the medical image storage apparatus 102 stores RIS information handled by a Radiology Information System (RIS). The medical image processing apparatus 10 acquires these data via the network connection circuit 14. Further, the medical image storage apparatus 102 stores medical image data for learning (hereinafter referred to as learning image data) generated by the medical image processing apparatus 10 based on the volume image, according to the control of the medical image processing apparatus 10.

  The medical image observation apparatus 103 is used, for example, by a radiologist as a viewer of various systems such as PACS, hospital information system (HIS), and RIS. The image interpretation doctor or the like creates an image interpretation report as a result of the image interpretation. In the interpretation report, the anatomical part of the affected area found by the interpretation and the aspect of the affected part are recorded as findings.

  The original data of the learning image data generated by the medical image processing apparatus 10 shown in FIG. 1 will be described with reference to FIG. In the following description, the z direction is taken as the normal direction. The volume image 30 has a plurality of tomographic image data 32 along the z direction.

  When the volume image 30 conforms to the DICOM standard, as shown in FIG. 3A, DICOM incidental data includes information of the modality 100 which generated the volume image 30, maker information of the modality 100, and contrast agent It includes various character information such as information on photographing conditions such as use / incidence and photographing protocol, and information on a subject such as height and weight. The medical image processing apparatus 10 acquires information included in the DICOM incidental data from the medical image storage apparatus 102. Note that the information on the imaging conditions and the information on the subject may be stored in the medical image storage apparatus 102 as RIS information.

  In the volume image 30, an annotation 34 is attached to the area 31a of the affected part of the organ 31 represented by the volume image 30.

  Specifically, the image interpreting doctor selects one representative tomogram data 33 in which the affected part 31a can be seen among the plurality of pieces of tomogram data 32 constituting the volume image 30 as a key image, and the region of the affected part 31a is selected. An annotation 34 is attached so as to surround it. The medical image storage apparatus 102 stores the information of the annotation 34 given by the image interpretation doctor in association with the volume image 30. The information of the annotation 34 includes the position and size of the annotation 34, the position and size of the affected area 31 a surrounded by the annotation 34, and additional information of the tomographic image data 33 to which the annotation 34 is added. The annotations 34 are attached by the operation of a radiologist. In addition, the image interpretation doctor may attach the annotation 34 by freely performing handwriting so as to surround the affected part 31a, or may use a template figure such as an ellipse or a rectangle. Further, the radiologist may attach a three-dimensional annotation 34 such as a sphere using a three-dimensional viewer.

  The radiologist creates an interpretation report as a result of the interpretation (see FIG. 3 (b)). The interpretation report includes the findings of the affected area 31a found by the interpretation. Further, the image interpretation doctor assigns the tomographic image data (hereinafter referred to as annotation-added tomographic image data) 33 to which the annotation 34 is added to the image interpretation report as representative image data in which a finding is observed.

  The processing circuit 15 of the medical image processing apparatus 10 adopts the annotation-attached tomogram data 33 as original data of learning image data for machine learning (see FIG. 2).

  Next, an example of the configuration and operation of the processing circuit 15 shown in FIG. 1 will be described.

  The processing circuit 15 is a processor that reads and executes a program stored in the storage circuit 13.

  The processing circuit 15 realizes the original data acquisition function 21, the area extraction function 22, the cutting function 23, the learning image acquisition function 24, and the storage instruction function 25 by reading out and executing the program stored in the storage circuit 13. .

  The original data acquisition function 21 acquires the volume image 30 and the information of the annotation 34. The medical image processing apparatus 10 may not have the original data acquisition function 21.

  The area extraction function 22 extracts the area 31 a to which the annotation 34 is attached from the volume image 30.

  The cutting function 23 sets the direction in which the affected area 31 a is cut out based on the affected area 31 a extracted by the area extracting function 22. For example, the cutting function 23 sets, in the volume image 30, a clipping direction z1 used in multi-planner reconstruction (MPR) processing. The cutout direction z1 is set to a direction different from the z direction (see FIG. 2), which is a direction in which at least a plurality of tomographic image data 32 are continuously arranged. Specifically, when a plurality of tomographic images are generated from the volume image 30 along the clipping direction z1, the clipping direction z1 is set so that a plurality of tomographic image data including the annotation 34 is included in the plurality of tomographic image data. Ru.

  The cutting function 23 cuts the volume image 30 in the direction intersecting the affected part 31 a along the set cutting direction z1. Specifically, the cutting function 23 generates a plurality of medical image data (hereinafter referred to as MPR image data) by subjecting the volume image 30 to MPR processing (cutting) along the cutting direction z1. The MPR method is a method of cutting three-dimensional data at a plane in an arbitrary direction and reconstructing a cross-sectional image viewed from the direction perpendicular to the plane. The medical image processing apparatus 10 may not have the cutting function 23.

  The learning image acquisition function 24 is cut in the direction intersecting the affected area 31 a extracted by the area extraction function 22, and acquires a medical image including a part of the annotation 34 from the volume image. When the medical image processing apparatus 10 includes the cutting function 23, the learning image acquisition function 24 acquires a medical image in which a part of the annotation 34 is included in the medical image cut by the cutting function 23.

  The storage instruction function 25 causes the medical image storage device 102 to store the plurality of medical images including the annotation 34 acquired by the learning image acquisition function 24 as learning images used in machine learning. The medical image processing apparatus 10 may not have the storage instruction function 25.

  An example of the operation of the processor 15 of the processing circuit 15 will be described with reference to FIG. FIG. 4 shows an example in which the medical image processing apparatus 10 also includes an original data acquisition function 21, a cutting function 23, and a storage instruction function 25 in addition to the area extraction function 22 and the learning image acquisition function 24.

  In step S <b> 1, the original data acquisition function 21 acquires the volume image 30 and the information of the annotation 34 from the medical image storage device 102.

  Next, in step S2, the region extraction function 22 extracts annotation-attached tomogram data 33 from a plurality of pieces of tomogram data 32 constituting the volume image 30 (see FIG. 5).

  Next, in step S3, the cutting function 23 sets, in the volume image 30, the clipping direction z1 used in MPR (Multi-Planner Reconstruction) processing. For example, the cutting function 23 sets, as the clipping direction z1, the direction of the transverse line 40 where the width W of the annotation 34 is the longest among the transverse lines 40 of the annotation 34 incidental to the annotation incidental tomographic image data 33 (FIG. 6 upper stage reference). Assuming that the direction of the transverse line 40 in which the width W of the annotation 34 is the longest is the cutout direction z1, the number of pieces of MPR image data 41 including the annotation 34 is maximized when MPR processing is performed along the cutout direction z1 can do.

  Next, in step S4, the cutting function 23 generates a plurality of MPR image data along the cutout direction z1 by performing an MPR process on the volume image 30 along the cutout direction z1. At this time, the cutting function 23 may convert the coordinates of each pixel of the volume image 30 into three orthogonal axes of x1 y1 z1 from the xyz coordinate system based on the cutting direction z1.

  Note that "to perform (cut) MPR along the cutting direction z1 means to perform MPR parallel to the cross section intersecting with the cutting direction z1, and MPR parallel to the cross section obliquely intersecting with the cutting direction z1. Including the case of processing. In the following description, in order to maximize the number of pieces of MPR image data to be generated, an example in which MPR processing is performed in parallel with a cross section orthogonal to the clipping direction z1 is shown.

  Next, in step S5, the learning image acquisition function 24 acquires MPR image data 41 including a part of the affected part 31a from the plurality of generated MPR image data based on the information of the annotation 34. The cutting function 23 obtains the range zs ≦ z1 ≦ ze in which the annotation 34 exists in the cutout direction z1 corresponding to the width W from the information of the annotation 34 acquired from the medical image storage device 102. The width W is equal to ze minus zs (see bottom of FIG. 6). The learning image acquisition function 24 acquires MPR image data of this range zs ≦ z1 ≦ ze as MPR image data 41 including a part of the affected part 31 a.

  The cutting function 23 determines the range zs ≦ z1 ≦ ze in which the annotation 34 exists in the cutting direction z1 before the MPR process in step S4, and performs the MPR process only in this range to partially process the affected part 31a. And MPR image data 41 may be acquired. In this case, execution of step S5 is unnecessary.

  As shown in FIG. 7, when the volume image 30 is sequentially cut in the Z-axis direction along the cutout direction z1 in the range zs ≦ z1 ≦ ze of the area to which the annotation 34 is added, part of the annotation 34 A plurality of contained medical images 41 are acquired from the volume image 30.

  Next, in step S6, the storage instruction function 25 instructs the medical image storage device 102 to store the plurality of medical images 41 acquired by the learning image acquisition function 24 as learning image data used in machine learning. (See FIG. 7).

  According to the above-described procedure, a large number of medical images including a part of the annotations 34 are easily collected as machine learning medical images from one volume image 30 based on the area to which the annotations 34 of the volume image 30 are added. can do.

  For this reason, it is necessary for machine learning only by using medical image data, examination information, interpretation report information, etc. shared within one existing medical facility or group hospital without new security and infrastructure maintenance. Image data for learning can be generated sufficiently.

  A plurality of clipping directions z1 may be set for one annotation attached tomographic image data 33. If the cutout direction z1 is changed, the generated MPR image data 41 becomes different image data. For this reason, as long as it can be expected to lead to reinforcement of learning, a plurality of cutout directions z1 may be set for one annotation attached tomographic image data 33.

  In this case, the plurality of cutout directions z1 preferably include the direction in which the width W intersecting the annotation 34 is the longest. For example, in the case where the shape of the annotation 34 is a shape drawn by a free curve, a plurality of directions such as four directions from the direction in which the width W intersecting the annotation 34 is longest from the longest direction You should set it.

  On the other hand, if the shape of the annotation 34 is circular, the width W intersecting the annotation 34 is the same in all directions, so for example, two directions of the x direction and the y direction, or a direction along y = x and y A cutting direction different from the z direction, such as four directions obtained by adding a direction along = -x, may be determined in advance. When setting a plurality of cutout directions z1 to one annotation-attached tomogram data 33 as described above, the number of learning image data generated from one volume image 30 obtained in one inspection may be further increased. it can.

Second Embodiment
A case where a plurality of combinations of the volume image 30 and the information of the annotation 34 (hereinafter, referred to as an original data set) are handled will be described. An original data set is generated, for example, one set for each examination.

  An example of the operation by the processor of the processing circuit 15 in the case of using a plurality of original data sets will be described with reference to FIGS. 8 and 9. The same steps as in FIG. 4 will be assigned the same reference numerals and overlapping explanations will be omitted.

  The processing circuit 15 may handle a plurality of combinations (original data sets) of the volume image 30 and the information of the annotation 34. At this time, assuming that the direction corresponding to the transverse line 40 in which the width W intersecting the annotation 34 is the longest is the cutout direction z1, different cutout directions z1 are set for each original data set. In this case, the shapes of bones and organs included in the MPR image data 41 differ for each original data set, and these may become noise and adversely affect learning.

  Therefore, when dealing with a plurality of original data sets, it is preferable to share one cutout direction z1 with all the original data sets (see FIG. 9). Further, in this case, it is preferable that the plurality of original data sets sharing the cutout direction z1 share the information of the imaging condition and the information of the subject (hereinafter, related information).

  Therefore, in step S11 of FIG. 8, the original data acquisition function 21 sends a plurality of original data sets (volume image 30 and a plurality of original data sets) common to related information (information on imaging conditions and object information) from the medical image storage device 102. A combination of information of the annotation 34 is acquired.

  Next, in step S12, the region extraction function 22 extracts the annotation-attached tomogram data 33 from the plurality of pieces of tomogram data 32 constituting the volume image 30 (see the second row in FIG. 9).

  Next, in step S13, the cutting function 23 sets a cutting direction z1 shared by all the original data sets. At this time, when the MPR image data is generated from the volume image 30 along the common clipping direction z1 for each of the plurality of original data sets, the cutting function 23 includes the annotations 34 generated from all the original data sets. The direction in which the total number of the MPR image data 41 is maximum is determined as the common clipping direction z1. The total length of the widths of the annotations 34 of the annotation incidental tomographic image data 33 in the common cutout direction z1 of all the original data sets is the longest as compared with the other directions.

  Next, in step S14, the cutting function 23 causes the medical image storage device 102 to store the information of the determined common clipping direction z1 via the storage instruction function 25.

  Next, in step S15, the cutting function 23 generates a plurality of MPR image data along the cutout direction z1 by performing an MPR process on each volume image 30 of the plurality of original data sets along the common cutout direction z1. Do.

  According to the above procedure, it is possible to collect many learning images while setting the common clipping direction z1 to a plurality of original data sets. All the original data sets have common related information and share the clipping direction z1. Therefore, the learning image data generated by the procedure of FIG. 8 can be tomographic image data observed from the same direction.

(Embodiment 3)
A method of classifying each of the MPR image data 41 including the annotation 34 in accordance with the size of the organ 31 and the size of the annotation 34 included in the MPR image data 41 will be described.

  As shown in FIG. 10, the processing circuit 15 may further implement the classification function 26 by reading and executing the program stored in the storage circuit 13.

  The classification function 26 classifies the medical image based on the size of the annotation included in the medical image acquired by the learning image acquisition function 24. For example, the classification function 26 obtains, for each of a plurality of MPR image data including the annotation 34, an index value indicating the size of the annotation 34 with respect to the size of the organ 31 in each MPR image data. Further, the classification function 26 classifies the plurality of MPR image data into a plurality of groups based on the index value. When a plurality of MPR image data are classified into a plurality of groups by the classification function 26, the storage instruction function 25 stores the learning image data in the medical image storage device 102 for each of these groups.

  An example of the procedure for classifying the state of the affected part 31a according to the position or size of the annotation 34 in the organ 31 will be described with reference to FIGS. 10 and 11. FIG. The same steps as in FIG. 4 will be assigned the same reference numerals and overlapping explanations will be omitted.

  The MPR image data 41 including the annotation 34 is extracted by the procedure of steps S1 to S5 shown in FIG.

  Next, in step S21, the classification function 26 acquires the contour of the organ 31 included in the MPR image data 41 by image processing such as edge detection. Then, the classification function 26 obtains, for each of the plurality of MPR image data 41 including the annotation 34, an index value indicating the size of the annotation 34 with respect to the size of the organ 31 in each MPR image data 41.

  As the index value, for example, a ratio R between the width of the organ 31 in the x1 direction and the width of the annotation 34 in the x1 direction can be used. At this time, as the width of the organ 31 in the x1 direction, the width at the same position as the annotation 34 may be used, or the maximum value of the width of the organ 31 in the x1 direction may be used. The width of the organ 31 can be obtained from the contour of the organ 31 acquired by image processing. The width of the annotation 34 can be easily obtained from the information of the annotation 34 acquired from the medical image storage apparatus 102.

  Next, the classification function 26 uses a threshold value stored in advance in the storage circuit 13 or the medical image storage apparatus 102, etc., based on the index value, and uses a plurality of MPR image data 41 used as learning image data. Classify into multiple groups.

  For example, the classification function 26 classifies the MPR image data 41 whose index value is equal to or less than the noise threshold into a noise learning group (data set) using the noise threshold. In addition, the MPR image data 41 whose index value is larger than the noise threshold is classified into a group in which the symptom progresses as the size of the affected part 31a is larger, using the threshold of the index value according to the progress state of the symptom. It may be further divided into a plurality of groups. As a plurality of groups according to the progress state of symptoms, for example, three groups of an early stage, a middle stage, and a late stage can be considered (see FIG. 11).

  The classification function 26 may use the width itself of the annotation 34 in the x1 direction as the index value. In this case, the threshold for classification into groups may be determined in advance for each type of organ 31.

  Next, in step S23, the storage instruction function 25 stores the plurality of MPR image data 41 including the annotation 34 as learning image data in the medical image storage device 102 for each of these groups.

  According to the above procedure, the MPR image data 41 including the annotation 34 can be classified according to the size of the organ 31 and the size of the annotation 34.

  As described above, according to the medical image processing apparatus 10, learning image data considered to be noise can be easily collected according to the size of the annotation 34. In machine learning, it makes sense to learn noise as noise. By learning image data to be determined as noise, CNN becomes robust to noise. In addition, it is possible to automatically classify learning image data according to the progress of a symptom.

  Further, in order to extract the feature structure with high accuracy, information of the modality 100 which generated the volume image 30, maker information of the modality 100, information of the imaging condition such as the use of the contrast agent and the imaging protocol, height, It is preferable to subdivide the learning image data according to the information of the subject such as the weight. In this regard, the medical image processing apparatus 10 can sufficiently collect learning image data even when the learning image data is subdivided. When the learning image data is subdivided, the classification function 26 may classify the MPR image data 41 including the annotation 34 according to classification conditions such as information on imaging conditions and information on objects. In this case, the storage instruction function 25 instructs the medical image storage apparatus 102 to store the MPR image data 41 for each classification.

  According to at least one embodiment described above, a large number of medical image data including an affected area can be easily collected for machine learning.

  The area extraction function 22, the cutting function 23, the learning image acquisition function 24, the storage instruction function 25 and the classification function 26 of the processing circuit 15 are respectively an area extraction unit, a cutting unit, an acquisition unit, and a storage instruction unit in the claims. And an example of a classification unit. The medical image storage device 102 is an example of the storage device in the claims.

  In the above embodiment, the word “processor” means, for example, a dedicated or general-purpose central processing unit (CPU), a graphics processing unit (GPU), or an application specific integrated circuit (ASIC), By circuits such as programmable logic devices (eg, Simple Programmable Logic Devices (SPLDs), Complex Programmable Logic Devices (CPLDs), and FPGAs) are meant. The processor implements various functions by reading and executing a program stored in a storage medium.

  In the above embodiment, an example in which a single processor of the processing circuit realizes each function is described. However, a plurality of independent processors are combined to form a processing circuit, and each processor realizes each function. It is also good. When a plurality of processors are provided, storage media for storing programs may be individually provided for each processor, or one storage medium may collectively store programs corresponding to the functions of all processors. It is also good.

  While certain embodiments of the present invention have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and the gist of the invention, and are included in the invention described in the claims and the equivalent scope thereof.

10 medical image processing apparatus 15 processing circuit 22 area extraction function 23 cutting function 24 learning image acquisition function 25 storage instruction function 26 classification function 30 volume image 31 organ 31 a diseased part 32 tomographic image data 33 ... annotation incidental tomographic image data 34 ... annotation 41 ... MPR image data 102 ... medical image storage device z 1 ... clipping direction z 1 ... common clipping direction

Claims (12)

An area extraction unit which extracts an area to which finding information is added from a three-dimensional medical image;
An acquisition unit which is cut in a direction intersecting with the area extracted by the area extraction unit and acquires a medical image including a part of the finding information from the three-dimensional medical image;
Medical image processing device. The image processing apparatus further comprises: a cutting unit which sets a direction in which the region is cut out based on the region extracted by the region extraction unit, and cuts the three-dimensional medical image in the intersecting direction along the set cutting direction.
The acquisition unit acquires a medical image including a part of the finding information in the medical image cut by the cutting unit.
The medical image processing apparatus according to claim 1. The three-dimensional medical image is composed of a plurality of two-dimensional medical images continuous in a predetermined direction,
The area extraction unit extracts an area to which the finding information is added from the plurality of two-dimensional medical images.
The cutting unit sets a direction different from the predetermined direction as the cutting-out direction, and cuts the plurality of two-dimensional medical images in the intersecting direction along the set cutting-out direction.
The medical image processing apparatus according to claim 2. The cutting unit sets a direction in which the width of the area extracted by the area extraction unit is the largest as the cutting-out direction.
The medical image processing apparatus according to claim 2. The cutting unit sets, as the cutting-out direction, a common direction obtained based on all the extracted regions when the region extraction unit extracts a plurality of regions to which the finding information is added.
The medical image processing apparatus according to claim 2. The cutting unit sets a direction in which the sum of widths of the regions extracted by the region extracting unit is maximum as the cutting-out direction.
The medical image processing apparatus according to claim 5. And a classification unit that classifies the medical image based on the size of the finding information included in the medical image acquired by the acquisition unit.
The medical image processing device according to any one of claims 1 to 6. The classification unit
Classifying the findings information having a size equal to or less than a threshold into a noise data group
The medical image processing apparatus according to claim 7. The classification unit
The medical image is displayed as a symptom based on the size of the finding information such that the condition is classified as progressing as the size of the finding information included in the medical image obtained by the obtaining unit is larger. Classified into groups according to the progress of
The medical image processing apparatus according to claim 7. The classification unit
Classifying the medical image acquired by the acquisition unit on the basis of related information including at least one of information of imaging conditions of the three-dimensional medical image and information of an object of the three-dimensional medical image;
The medical image processing apparatus according to any one of claims 7 to 9. The storage device further includes a storage instructing unit instructing storage of the medical image including a part of the finding information acquired by the acquisition unit as an image for machine learning in a storage device.
The medical image processing apparatus according to any one of claims 1 to 10. An area extraction step of extracting an area to which the finding information is added from the three-dimensional medical image;
Obtaining in the direction crossing the area extracted in the area extraction step, a medical image including a part of the finding information from the three-dimensional medical image;
Medical image processing method.

Images