KR102196503B1 - Apparatus and method for detecting the spine - Google Patents

Abstract

The present invention provides a detection device for automatically detecting a spine in a medical image. According to the present invention, the detection device may comprise: an acquisition unit which acquires a medical image; and a detection unit for detecting a spine in the medical image using image analysis or machine learning.

Description

Spinal detection device and detection method {APPARATUS AND METHOD FOR DETECTING THE SPINE}

The present invention relates to an apparatus and method for detecting a spine in a medical image.

In order to automatically analyze various lesions related to the spine, the process of detecting the spine in medical images such as X-ray and CT (Computed Comography) should be preceded.

Although it is relatively easy for a medical staff to directly classify a spine in a medical image, the task of automatically classifying and detecting the spine mechanically involves various difficulties.

Korean Patent Publication No. 2015-0104889 discloses a CT device capable of diagnosing cervical canal stenosis by photographing the spinal canal of the cervical spine.

Korean Patent Publication No. 2015-0104889

The present invention is to provide a detection apparatus and detection method for automatically detecting a spine in a medical image.

The detection apparatus of the present invention comprises: an acquisition unit for obtaining a medical image; It may include a; detection unit for detecting the spine in the medical image using image analysis or machine learning.

The detection method of the present invention includes an acquisition step of obtaining a medical image; A division step of distinguishing at least one of a bone region representing a bone, a first region other than the bone region, and a second region representing a bone other than the spine in the medical image; And detecting a spine region representing a spine in the medical image using at least one of the bone region, the first region, and the second region.

The detection apparatus and detection method of the present invention may automatically detect a spine in a medical image through image analysis or machine learning.

According to the present invention, the detection accuracy of the spine can be improved by using the geographic characteristics of the spine with respect to the area around the spine.

The detection apparatus and detection method of the present invention can virtually generate a boundary line required for detection of the spine, and detect the spine by using geometric characteristics between the virtually generated boundary lines.

1 is a block diagram showing an analysis device of the present invention.
2 is a photograph showing a medical image acquired by an acquisition unit.
3 is a schematic diagram showing the operation of the detection unit.
4 is a schematic diagram showing a spine candidate group selected by the detection unit.
5 is a schematic diagram showing a spine region detected by a detection unit.
6 is a schematic diagram showing the operation of the classification unit.
7 is a schematic diagram showing the operation of the auxiliary unit.
8 is a schematic diagram showing borders and highlights introduced to analyze medical images.
9 is a schematic diagram showing the operation of the classification unit.
10 is a flow chart showing the analysis method of the present invention.
11 is a flowchart showing the detection method of the present invention.
12 is a flowchart showing a classification method of the present invention.
13 is a flowchart showing a diagnosis method of the present invention.
14 is a diagram illustrating a computing device according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement the embodiments of the present invention. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are assigned to similar parts throughout the specification.

In this specification, redundant descriptions of the same components are omitted.

In addition, in the present specification, when a component is referred to as being'connected' or'connected' to another component, it may be directly connected or connected to the other component, but other components in the middle It should be understood that may exist. On the other hand, in the present specification, when it is mentioned that a certain element is'directly connected' or'directly connected' to another element, it should be understood that no other element exists in the middle.

In addition, terms used in the present specification are only used to describe specific embodiments, and are not intended to limit the present invention.

In addition, in the present specification, a singular expression may include a plurality of expressions unless the context clearly indicates otherwise.

In addition, in the present specification, terms such as'include' or'have' are only intended to designate the existence of features, numbers, steps, actions, components, parts, or a combination thereof described in the specification, and one or more It is to be understood that the presence or addition of other features, numbers, steps, actions, components, parts, or combinations thereof, does not preclude in advance the possibility of being excluded.

In addition, in this specification, the term'and/or' includes a combination of a plurality of listed items or any of a plurality of listed items. In the present specification,'A or B'may include'A','B', or'both A and B'.

In addition, in this specification, detailed descriptions of known functions and configurations that may obscure the subject matter of the present invention will be omitted.

1 is a block diagram showing an analysis device of the present invention.

The analysis device illustrated in FIG. 1 may analyze a medical image to detect a bone region or a spine region. Thus, the analysis device of the present invention may be referred to as a detection device. In addition, the analysis method of the present invention may be referred to as a detection method.

The analysis device illustrated in FIG. 1 may classify or classify bone regions by body parts. Therefore, the analysis device of the present invention may be referred to as a classification device. In addition, the analysis method of the present invention may be referred to as a classification method.

The analysis device illustrated in FIG. 1 may diagnose musculoskeletal diseases by analyzing the bone region itself, or diagnose musculoskeletal diseases using a result of classifying the bone regions for each body part. Therefore, the analysis device of the present invention may be referred to as a diagnostic device. In addition, the analysis method of the present invention may be referred to as a diagnostic method.

The analysis device illustrated in FIG. 1 may include an acquisition unit 110, a detection unit 130, an auxiliary unit 150, a classification unit 170, a diagnosis unit 190, and an output unit 180.

The acquisition unit 110 may acquire a medical image.

Medical images may be generated through computed tomography (CT), magnetic resonance imaging (MRI), X-ray, positron emission tomography (PET), or the like. Medical images are photographed targeting various organs, skeletons (bones), blood vessels, etc. existing inside a patient's body, and may include images that visually represent the inside of the body. By analyzing medical images, various singularities different from those of a normal person can be identified. Diagnosis of various lesions can be made through identification of the singularity. The acquisition unit 110 may receive a medical image from imaging equipment such as CT, MRI, X-ray, and PET.

In order to diagnose musculoskeletal diseases related to various bones such as the spine, the acquisition unit 110 may acquire a medical image including a bone and a spine.

Various medical images can be generated according to DICOM (Digital Imaging and Communication in Medicine) standards. If necessary, the acquisition unit 110 may resize the medical image to a size required for automatic diagnosis. For example, the acquisition unit 110 may change a medical image having a size of 512×512 into a resizing image having a size of 256×256.

The detector 130 may detect a bone region representing a bone in a medical image. Alternatively, the detection unit 130 may detect the spine in the medical image using image analysis or machine learning. In order to classify various body organs included in the medical image, the detection unit 130 may use an image processing algorithm such as an Online Teaching Support Unit (OTSU).

The classification unit 170 may classify at least one of a cervical spine, a thoracic spine, a lumbar spine, a sacral spine, and a tail spine from a spine included in the medical image using image analysis or machine learning. For example, the classification unit 170 may classify at least one of a cervical vertebra, a thoracic vertebra, a lumbar vertebra, a sacral vertebra, and a tail vertebra from the spine detected by the detection unit 130.

The auxiliary unit 150 may improve detection accuracy and detection speed of a bone region or a spine region performed by the detection unit 130. The auxiliary unit 150 may improve classification accuracy and classification speed of the spine performed by the classification unit 170. To this end, the auxiliary unit 150 may detect a region other than the bone region in the medical image. The auxiliary unit 150 may be integrally formed with the detection unit 130.

The diagnosis unit 190 may diagnose musculoskeletal disorders from a medical image using image analysis or machine learning. The diagnosis unit 190 analyzes the bone region detected in the medical image by the detection unit 130, the non-bone region detected by the auxiliary unit 150, and the classification result of the classification unit 170 to analyze scoliosis, lordosis, It can diagnose musculoskeletal disorders related to the spine, such as compression fractures.

The output unit 180 may generate report information including a diagnosis result of a musculoskeletal disorder, and display the report information through a visual output means such as a display or an audio output means such as a speaker.

2 is a photograph showing a medical image acquired by the acquisition unit 110.

The first medical image i1 illustrated in FIG. 2 may include a medical image photographed in front of the user. The second medical image i2 illustrated in FIG. 2 may include a medical image photographing a side of the user.

The first medical image i1 and the second medical image i2 of FIG. 2 are X-ray images, showing the user's bones and some other organs well.

The acquisition unit 110 may acquire a full-body image of a user as a medical image.

The medical image may include the user's whole body regardless of the type or range of the target of interest. For example, even if the target of interest of the detection unit 130 is a spine, a full-body image including leg bones, shoulders, buttocks, etc. may be provided to the detection unit 130.

The detection unit 130 may more accurately detect a target of interest such as a spine by using elements other than the spine. In addition, other elements included in the medical image besides the spine may be helpful in improving the classification accuracy of the classification unit 170 and the diagnosis accuracy of the diagnosis unit 190.

3 is a schematic diagram showing an operation of the detection unit 130.

The detection unit 130 may detect the bone region b in the medical image. For example, the detector 130 may detect all bones included in the medical image and detect a plurality of bone regions b that are separated from each other through boundary line analysis of each bone.

When a plurality of medical images i1 and i2 photographed for the same user are acquired, the detection unit 130 may detect the bone region b for each medical image i1 and i2.

The detection unit 130 may select one or more bone regions estimated to be the spine from among the plurality of detected bone regions b as spine candidate groups.

The detection unit 130 may detect a spine region representing a spine from among the spine candidate group through image analysis or machine learning.

4 is a schematic diagram showing a spine candidate group selected by the detection unit 130.

The detection unit 130 may detect a bone region b from each of the plurality of medical images i1 and i2 acquired by the acquisition unit 110 and select a spine candidate group estimated to be a spine from the bone region b.

The spine candidate group may include spine regions s1 and s2. The spine candidate group can be labeled and machine learned.

5 is a schematic diagram showing a spine region detected by the detection unit 130.

The detection unit 130 may select a spine candidate group from the bone region b, and may detect spine regions s1 and s2 representing the spine from among the selected spine candidate groups. For example, the detection unit 130 may extract or cut out a spine candidate group corresponding to spine regions s1 and s2 from medical images i1 and i2.

The plurality of medical images i1 and i2 may include images photographed from different angles for the same user. By analyzing each of the spine regions s1 and s2 detected from a plurality of medical images i1 and i2 taken from different angles at the same time point, various spine-related diseases can be diagnosed.

The first spine region s1 may include a spine region detected in a medical image photographed in front of the user. In other words, the first spine region s1 may correspond to a view of the user's spine from the front.

The second spine region s2 may include a spine region detected from a medical image photographing a side of the user. In other words, the second spine region s2 may correspond to a side view of the user's spine.

For example, the detection unit 130 may detect a specific region extending upward and downward by stacking a plurality of unit bones having a closed curve shape among the plurality of bone regions b as spinal regions s1 and s2. A specific area in which unit bones having a rectangular closed curve are stacked up and down to extend may represent a spine in a form in which vertebrae corresponding to the unit bones are stacked vertically.

The spine regions s1 and s2 detected by the detection unit 130 may be provided to the diagnosis unit 190 or may be provided to the classification unit 170.

6 is a schematic diagram showing the operation of the classification unit 170.

The classification unit 170 may subdivide the vertebrae into cervical vertebrae a1, thoracic vertebrae a2, lumbar vertebrae a3, sacral vertebrae a4, etc. for the vertebral regions s1 and s2 extracted by the detection unit 130.

The classification unit 170 may divide the spine by labeling the spine regions s1 and s2. Various information acquired by the assistant 150 may be used in a machine learning process for subdividing the spine.

7 is a schematic diagram showing the operation of the auxiliary unit 150.

The auxiliary unit 150 may detect at least one of a first region other than a bone region and a second region representing a bone other than the spine in the medical image.

The detection unit 130 may detect spine regions s1 and s2 representing the spine, or subdivide the detected spine regions s1 and s2 based on at least one of the first region and the second region.

For example, the auxiliary unit 150 may extract the shoulder h and the hip meat r from the medical image.

In the case of an X-ray image, the shoulder h may include the upper part of the shoulder blade or the upper part of the shoulder covering the shoulder blade. The hip flesh r may include an outline of the portion of the flesh covering the hip bone.

The detection unit 130 may detect a specific region extending from the height of the shoulder h to the height of the upper end r1 of the hip meat r among the plurality of bone regions b as spinal regions s1 and s2.

When the detection unit 130 detects a specific bone region extending from the shoulder h to the upper portion of the hip r1 with the help of the auxiliary unit 150, the detection unit 130 adds a region extending over the shoulder to the specific bone region and converts the spine regions s1 and s2. Can be detected. The bone region continuously extending over the shoulder may correspond to a part of the cervical vertebrae among the vertebrae.

The detection unit 130 may detect the vertebral regions s1 and s2 by adding a region continuously extending below the top of the hip meat to the detected specific bone region with the help of the auxiliary unit 150. The bone region continuously extending below the hip flesh may correspond to a part of the lumbar spine or a part of the sacral spine.

For example, the auxiliary unit 150 may extract the shoulder h and the pelvis from the medical image.

The pelvis is a funnel-shaped part of the waist where the hip bones (iliac bones, ischium), the hip bones (ischium), the pubis (pubis), and the vertebrae that join them together form the waist. It is a skeleton. The pelvis protects the internal organs and is the joint of the leg.

In the middle of the pelvis, there is a cranial bone and a hole y surrounded by the hip bone.

The detection unit 130 may detect a specific region extending from the upper end y1 of the hole y formed in the center of the pelvis to the height of the shoulder h among the plurality of bone regions b as spinal regions s1 and s2.

The detection unit 130 may detect the vertebral regions s1 and s2 by adding a region extending below the upper end y1 of the hole y continuously to the detected specific bone region with the help of the auxiliary unit 150. The bone region continuously extending below the upper end y1 of the hole y may correspond to a part of the sacrum or a part of the tail.

8 is a schematic diagram showing borders and highlights introduced to analyze medical images.

The detector 130 may distinguish between the first bone v1 and the second bone v2 disposed above or below the first bone v1 in the medical images i1 and i2. The first bone v1 and the second bone v2 may represent different bone regions. Alternatively, the first bone v1 and the second bone v2 may represent a bone unit included in one bone region.

In summary, the first bone v1 and the second bone v1 may represent a set of unit bone regions or may represent the unit bone region itself.

For extraction of the spine, the detection unit 130 preferably treats the entire spine as one bone area b. However, the spine may include a plurality of vertebrae stacked up and down. In this case, a situation in which the detection unit 130 detects each vertebral body as a single bone region b differently from the intention may occur.

Even if a plurality of vertebrae are extracted into different bone regions, the detection unit 130 may detect the plurality of vertebrae as a single bone region through post-processing. Also, the detection unit 130 may detect a bone region itself having a characteristic in which a plurality of vertebrae are arranged vertically as a spine region.

In order to detect a set of individual vertebrae as one bone region, the detection unit 130 may use a boundary line concept.

The boundary line concept may also be used in the diagnosis unit 190 other than the detection unit 130, but may be different for purposes of use. The diagnosis unit 190 may diagnose various musculoskeletal disorders using a boundary line concept or a highlight concept.

The detection unit 130 generates a first boundary line b1 along one surface of the first bone 1 facing the second bone v2, and generates a second boundary line b2 along the surface of the second bone v2 facing the first bone v1. I can.

The detection unit 130 may generate a center point o of the first bone v1, and generate two first boundary lines b1 facing each other with the center point o interposed therebetween. The center point o of the first bone v1 may be formed at the inner center of the first bone v1 having a closed curve shape.

The first boundary line b1 on the upper side of the center point o may face the second bone v2 disposed on the upper side.

Based on the center point o, the lower first boundary line b1 may face the second bone v2 disposed at the lower side.

The detector 130 may analyze a first geometric characteristic between the first boundary line b1 on the upper side and the second boundary line b2 of the second bone v2 disposed on the upper side. In addition, the detection unit 130 may analyze a second geometric characteristic between the first boundary line b1 on the lower side and the second boundary line b2 of the second bone v2 disposed at the lower side.

The detector 130 may extract a spine region representing the spine from the medical image using the first geometric feature and the second geometric feature.

For example, the first geometric characteristic or the second geometric characteristic is the length of the first boundary line, the length of the second boundary line, the distance between the first boundary line and the second boundary line, the angle between the first boundary line and the second boundary line, and the same first bone. It may include a length difference or an angle difference between the upper first boundary line and the lower first boundary line.

By using the first geometric characteristic and the second geometric characteristic, a vertebral body can be extracted from various bone regions, a spine candidate group in which a plurality of vertebrae are stacked can be selected, or a vertebral region can be easily detected.

The detection unit 130 may detect a spine region having a morphological feature of a spine in a medical image through learning of at least one of the cervical spine, thoracic spine, lumbar spine, sacral spine, and tail spine. Among various bone regions whose morphological features are included in medical images, the spine region refers to a different shape part compared to other bone regions, and a model that detects the spine region in the medical image is created through machine learning of the morphological feature. Can be. The detection unit 130 may detect the spine region using a corresponding model or an image analysis technique.

9 is a schematic diagram showing an operation of the classification unit 170.

The classification unit 170 may subdivide the spine using morphological features or positional features of the spine. The classification unit 170 is at least one of cervical a1, thoracic a2, lumbar a3, sacral a4, and tail a5 in the vertebral regions s1 and s2 through the learning of positional features of the vertebral regions s1 and s2 detected by the detection unit 130 You can specify the range of. Positional features may include a geographic concept of where the vertebrae forming the spine are located on the body.

The diagnosis unit 190 may integrate information extracted from morphological features and positional features of the spine, and classify normal and abnormal spines, or diagnose or discriminate a curved state of the spine using the integrated information.

The diagnosis unit 190 may integrate lesion diagnosis results for each sub-region a1, a2, a3, a4, and a5 classified by the classification unit 170. The diagnosis unit 190 may integrate lesion diagnosis results analyzed for each detailed area, calculate an abnormality of the spine using the integrated information, and predict an integrated disease of the spine.

The output unit 180 may visually or audibly display report information including at least one of abnormality and integrated disease prediction information.

The classification unit 170 may extract a plurality of vertebrae c included in vertebral regions s1 and s2 representing the spine.

The classification unit 170 may count the number of vertebrae c sequentially stacked downward based on the uppermost vertebral body, or count the number of vertebral bodies c sequentially stacked upward based on the lowest vertebral body. The classification unit 170 may divide the spinal regions s1 and s2 into cervical vertebrae region a1, thoracic vertebrae region a2, and lumbar vertebrae region a3 according to the counting values.

The classification unit 170 may determine whether the lateral projection s exists for the vertebral body c disposed around the height of the shoulder h extracted by the auxiliary unit 150. Each of the vertebrae c has a transverse protrusion w, but morphologically, the transverse protrusion length of the transverse protrusion w may be different. When a plurality of vertebral bodies c are arranged in succession, the classification unit 170 treats that there is no lateral projection if the protrusion length difference d1, d2 of the lateral projections between two adjacent vertebrae is greater than or equal to the set value, and , The other transverse projection can be treated as being present.

If there is a second vertebral body c2 having a lateral projection w immediately under the first vertebral body c1 without the lateral projection w, the classification unit 170 includes the second vertebral body c2 and the upper vertebral body c into the cervical vertebrae region a1 Alternatively, the lower vertebrae c including the vertebrae immediately below the second vertebrae c2 may be classified as a thoracic region a2. This embodiment may be based on the fact that the cervical vertebrae are included in the cervical vertebrae up to the second vertebral body c2 having the lateral protrusions abruptly protruding from the immediately lower end of the cervical vertebrae having almost no transverse protrusions.

The classification unit 170 may determine whether there is a lateral protrusion w of the vertebrae disposed around the height of the upper end r1 of the hip meat r extracted by the auxiliary unit 150. If there is a fourth vertebral body c4 having a lateral projection w immediately below the third vertebral body c3 without the lateral projection w, the classification unit 170 includes the third vertebral body c3 and the upper vertebral body c into the thoracic region a2 Alternatively, the lower vertebral body c including the fourth vertebral body c4 may be divided into a lumbar region a3. In this embodiment, when a fourth vertebral body c4 having a lateral projection sharply protruding from the lower end of the thoracic vertebrae is continuously arranged at the thoracic vertebrae portion, the third vertebral body c3 immediately above the fourth vertebral body c4 It can only be based on the fact that it is included in the thoracic spine.

The classification unit 170 for extracting a plurality of vertebrae from the spine region of the medical image may search for a specific vertebrae (the fifth vertebrae c5) having a sacral blade among vertebrae c existing at the lower ends of the vertebral regions s1 and s2.

The classification unit 170 may divide the upper vertebrae into the lumbar region a3, including the vertebrae that are immediately above the fifth vertebrae c5, or divide the region of the fifth vertebrae c5 into the sacral region a4.

The diagnosis unit 190 may detect a feature point in each of the plurality of bone regions b detected by the detection unit 130.

The diagnosis unit 190 may collect and integrate the feature points of each bone region.

The diagnosis unit 190 may calculate an abnormality by using the integrated feature points. The diagnosis unit 190 may predict an integrated disease in the bone region using the abnormality. The prediction information of the integrated disease predicted by the diagnosis unit 190 may be displayed as report information through the output unit 180 such as a display.

The diagnosis unit 190 may grasp body flexion information according to the photographing posture of the medical image by using the feature points. The diagnosis unit 190 may diagnose musculoskeletal disorders such as compression fracture, scoliosis, and lordosis using the flexion information.

The diagnosis unit 190 may predict or estimate various diseases using the boundary line concept discussed above.

For example, the diagnosis unit 190 may distinguish between a first bone v1 and a second bone v2 facing each other in the medical images i1 and i2. The diagnosis unit 190 may distinguish between the first bone v1 and the second bone v2 disposed above or below the first bone in the medical image.

The diagnostic unit 190 generates a first boundary line b1 along one surface of the first bone v1 facing the second bone v2, and generates a second boundary line b2 along one surface of the second bone v2 facing the first bone v1. can do.

The diagnostic unit 190 may generate a center point o of the first bone v1 and generate two first boundary lines b1 facing each other with the center point o interposed therebetween.

The first boundary line on the upper side based on the center point o may face the second bone disposed on the upper side.

Based on the center point o, the lower first boundary line may face the second bone disposed at the lower side.

The diagnosis unit 190 may determine the center of the first bone v1 and generate a first center point distinguishable from the original medical image at the center of the first bone v1.

The diagnosis unit 190 may determine the center of the second bone v2 and generate a second center point distinguishable from the original medical image at the center of the second bone v2.

The diagnostic unit 190 may analyze a first geometric characteristic between the first boundary line on the upper side and the second boundary line of the second bone disposed on the upper side. The diagnosis unit 190 may analyze a second geometric characteristic between the lower first boundary line and the second boundary line of the second bone disposed at the lower side. According to the present embodiment, when two second bones are disposed to face each other with the first bone therebetween, the diagnosis unit 190 may analyze geometric characteristics of the first bone and each second bone.

The diagnosis unit 190 may diagnose a musculoskeletal disorder by using the first geometric characteristic and the second geometric characteristic.

The diagnosis unit 190 may detect or classify lesions related to the first bone and the second bone using the first boundary line, the second boundary line, the first center point, and the second center point.

The diagnosis unit 190 may generate a closed curve-shaped highlight patch p including a portion of the first bone and a portion of the second bone.

The first boundary line b1 may be included inside the first side t1 of the highlight patch p. The first side t1 may extend along the first boundary line b1.

A second boundary line b2 may be included inside the second side t2 of the highlight patch p. The second side t2 may extend along the second boundary line b2.

The highlight patch may include a closed curve including a first boundary line b1 of the first bone v1 inside one side and a second boundary line b2 of the second bone v2 inside the other side.

The diagnosis unit 190 may detect or classify the lesion using geometric characteristics between the first side t1 and the second side t2 of the highlight patch.

For example, when geometric characteristics such as the shape and size of the highlight patch are determined, the length of the first side t1, the length of the second side t2, the position of the second side t2 with respect to the first side t1, and the first side t1 Various physical characteristics, such as the slope of the second side of and the distance d between the first side t1 and the second side t2, may be recognized. Using the identified physical characteristics, it is possible to identify singularities suspected of lesions such as compression fractures.

Due to the highlight patch p, the boundary between the first bone and the second bone facing each other may be clearly and precisely defined geometrically. When a geometric characteristic is clearly and precisely defined, various values such as length, distance, and angle indicated by the geometric characteristic can be used as a source that can be used for actual diagnosis. Consequently, according to the highlight patch p, the number of input sources usable for lesion diagnosis can be significantly increased. Using this, it is possible to apply various types of diagnostic models using various physical characteristics as an input source.

10 is a flow chart showing the analysis method of the present invention.

The analysis method shown in FIG. 10 may be performed by the analysis device of FIG. 1.

The analysis method of the present invention may include an acquisition step (S510), a detection step (S520), a classification step (S530), a diagnosis step (S540), and an output step (S550).

In the acquisition step S510, a medical image such as computed tomography (CT), magnetic resonance imaging (MRI), X-ray, and positron emission tomography (PET) may be acquired. The acquisition step (S510) may be performed by the acquisition unit 110.

The detection step (S520) may be performed by the detection unit 130. In the detection step (S520), the spine may be detected from the medical image using image analysis or machine learning. The detection of the spine performed in the detection step (S520) may be performed not for a body but for a medical image. Accordingly, detection of the spine performed in the detection step S520 may substantially mean detecting a spine region representing the spine in a medical image. In the detection step (S520 ), the detection unit 130 may also detect a bone region included in the medical image.

In the classification step (S530), at least one of the cervical, thoracic, lumbar, sacral and caudal vertebrae may be classified from the spine included in the medical image using image analysis or machine learning. The classification step (S530) may be performed by the classification unit 170.

In the diagnosis step (S540), a musculoskeletal disorder may be diagnosed from a medical image using image analysis or machine learning. The diagnosis step S540 may be performed by the diagnosis unit 190.

In the output step S 550, report information including a diagnosis result of a musculoskeletal disorder may be generated and displayed. The outputting step S 550 may be performed by the output unit 180 having a visual output means such as a display or an audio output means such as a speaker.

11 is a flowchart showing the detection method of the present invention.

The detection method shown in FIG. 11 may include the detection step (S520) of FIG. 10. The detection method illustrated in FIG. 11 may be performed by a detection device that detects a spine in a medical image.

The detection method includes an acquisition step of obtaining a medical image (S510), a classification step of discriminating at least one of a bone region representing a bone in a medical image, a first region other than a bone region, and a second region representing a bone other than the spine. Can include. The detection method may include a spine detection step (S524) of detecting a spine region representing a spine in a medical image using at least one of a bone region, a first region, and a second region.

In the classification step, the detection unit 130 may first detect the bone region and extract the first region and the second region (S521).

The detection unit 130 may extract morphological features of the spine through learning of at least one of the cervical spine, thoracic spine, lumbar spine, sacral spine, and tail spine (S522).

The detection unit 130 may extract a positional reference, such as a shoulder or a hip, through analysis of the first region and the second region (S523).

The detection unit 130 may detect the spine using morphological features of the spine or detect the spine using a positional reference (S524).

12 is a flowchart showing a classification method of the present invention.

The classification method shown in FIG. 12 may include the classification step (S530) of FIG. 10. The classification method illustrated in FIG. 12 may be performed by a classification device that classifies at least one of a cervical spine, a thoracic spine, a lumbar spine, a sacrum, and a tail spine from a spine included in a medical image.

The classification method may include an acquisition step of obtaining a medical image (S510) and a detection step of detecting a spine region representing a spine in the medical image (S520).

The classification method may include a classification step (S530) of subdividing the spine into the spine regions s1 and s2. The classification step (S530) may be embodied through the operation of the classification unit 170 as follows.

First, the classification unit 170 may extract a plurality of vertebrae c included in the spine region (S531).

The classification unit 170 may extract morphological features of the spine (S532).

The classification unit 170 may extract a positional feature of the spine (S533). In order to extract the positional features of the spine, the classification unit 170 may extract a first region, a second region, and the like, which are not bone regions, such as shoulder and hip meat as a reference.

The classification unit 170 may determine the classification range of the vertebrae according to the morphological characteristics of the spine, for example, using a lateral projection, or a position on the upper side or the lower side based on the shoulder. The classification unit 170 may designate at least one of the cervical spine, thoracic spine, lumbar spine, sacral spine, and tail spine in the spine region by using information extracted from morphological features or positional features of the spine (S 534).

13 is a flowchart showing a diagnosis method of the present invention.

The diagnosis method illustrated in FIG. 13 may include a diagnosis step (S540) of FIG. 10. The diagnosis method shown in FIG. 13 may be performed by a diagnosis apparatus for diagnosing musculoskeletal disorders in a medical image using image analysis or machine learning.

The diagnosis method may include an acquisition step (S510) of acquiring a medical image, and a detection step (S520) of detecting a bone region b included in the medical image. The following process may be embodied by the operation of the diagnosis unit 190.

The diagnostic unit 190 may classify the detected bone region b by body part, such as k1, k2, k3,..., kn (S541).

For example, the diagnosis unit 190 may classify the bone region b as a rib region, a spine region, an arm bone region, an hip bone region, and a shoulder bone region. k1, k2, k3,..., kn may correspond to a rib region, a spine region, an arm bone region, an hip region, a shoulder region, and the like in a 1:1 manner.

The diagnostic unit 190 may extract feature points f1, f2, f3,..., fn of each bone region classified by body part (S542).

The diagnostic unit 190 may collect the feature points f1, f2, f3,..., fn of each bone region (S543).

The diagnosis unit 190 may calculate an abnormality of the musculoskeletal system or predict integrated disease information of the musculoskeletal system using the collected information (S544).

In the step of extracting feature points (S 543 ), a highlight concept or a boundary line concept may be used.

The diagnosis unit 190 performing the diagnosis method may distinguish between the first bone and the second bone that face each other in the detected bone region. In order to distinguish between the first bone and the second bone, the diagnosis unit 190 may receive the help of the detection unit 130 or the classification unit 170.

The diagnosis unit 190 may generate a first boundary line along one surface of the first bone that faces the second bone, and may generate a second boundary line along one surface of the second bone that faces the first bone.

The diagnostic unit 190 may generate a closed curve-shaped highlight patch including a portion of the first bone and a portion of the second bone. A first boundary line may be included inside the first side of the highlight patch, and the first side may extend along the first boundary line. A second boundary line may be included inside the second side of the highlight patch, and the second side may extend along the second boundary line.

The diagnosis unit 190 may detect or classify a lesion using geometric characteristics between the first side and the second side of the highlight patch.

14 is a diagram illustrating a computing device according to an embodiment of the present invention. The computing device TN100 of FIG. 14 may be a device (eg, an analysis device, etc.) described herein.

In the embodiment of FIG. 14, the computing device TN100 may include at least one processor TN110, a transmission/reception device TN120, and a memory TN130. In addition, the computing device TN100 may further include a storage device TN140, an input interface device TN150, an output interface device TN160, and the like. Components included in the computing device TN100 may be connected by a bus TN170 to communicate with each other.

The processor TN110 may execute a program command stored in at least one of the memory TN130 and the storage device TN140. The processor TN110 may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which methods according to an embodiment of the present invention are performed. The processor TN110 may be configured to implement the procedures, functions, and methods described in connection with an embodiment of the present invention. The processor TN110 may control each component of the computing device TN100.

Each of the memory TN130 and the storage device TN140 may store various information related to an operation of the processor TN110. Each of the memory TN130 and the storage device TN140 may be configured with at least one of a volatile storage medium and a nonvolatile storage medium. For example, the memory TN130 may be composed of at least one of a read only memory (ROM) and a random access memory (RAM).

The transmission/reception device TN120 may transmit or receive a wired signal or a wireless signal. The transmission/reception device TN120 may be connected to a network to perform communication.

Meanwhile, the embodiment of the present invention is not implemented only through the apparatus and/or method described so far, but may be implemented through a program that realizes a function corresponding to the configuration of the embodiment of the present invention or a recording medium in which the program is recorded. In addition, such an implementation can be easily implemented by a person skilled in the art from the description of the above-described embodiments.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of the skilled person using the basic concept of the present invention defined in the following claims are also present It belongs to the scope of the invention.

110...acquisition section 130...detection section
150...Subsidiary 170...Classification
180...output section 190...diagnostic section

Claims (9)

An acquisition unit that acquires a medical image;
Including; a detection unit for detecting the spine in the medical image using image analysis or machine learning,
An auxiliary unit for extracting shoulder and hip flesh from the medical image is provided,
The detection unit detects a bone region representing a bone in the medical image,
The detection unit detects a specific region of the bone region extending from the height of the shoulder to the height of the upper end of the hip meat as a spine region.
The method of claim 1,
The detection unit selects a spine candidate group from the bone region, and detects the spine region representing a spine from the spine candidate group.
delete The method of claim 1,
A detection device provided with a classification unit for classifying at least one of a cervical vertebra, a thoracic vertebra, a lumbar vertebra, a sacral vertebra, and a tail vertebra from the spine detected by the detection unit.
delete delete An acquisition unit that acquires a medical image;
Including; a detection unit for detecting the spine in the medical image using image analysis or machine learning,
An auxiliary unit for extracting the shoulder and pelvis from the medical image is provided,
The detection unit detects a bone region representing a bone in the medical image,
The detection unit detects a specific region of the bone region extending from the upper end of the hole formed in the middle of the pelvis to the shoulder as a spine region.
delete In the detection method performed by a detection device for detecting a spine in a medical image,
An acquisition step of obtaining a medical image;
A division step of distinguishing at least one of a bone region representing a bone, a first region other than the bone region, and a second region representing a bone other than the spine in the medical image;
A detecting step of detecting a spine region representing a spine in the medical image using at least one of the bone region, the first region, and the second region; and

Extracting the shoulder and hip flesh from the medical image, detecting a bone region representing a bone in the medical image, and detecting a specific region extending from the height of the shoulder to the upper end of the hip flesh among the bone regions as the spine region, or ,

The shoulder and the pelvis are extracted from the medical image, a bone region representing a bone is detected in the medical image, and a specific region extending from the upper end of the hole formed in the center of the pelvis to the shoulder among the bone regions is referred to as the spine region. Detection method to detect.

Images