JP2000126168A - Bone image processing method and bone strength evaluation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an index better reflecting the strength of a bone rather than the density thereof by repeating for the whole of a bone image a processing in which a scanning is made in the direction of the starting point of a raster scanning line from the bottom side of an isosceles triangle to determine an extracted bone region as cortical bone region and the cortical bone region is finally defined if an element of a bone image exists within the region to separate the bone image into the cortical bone region and a bone beam region. SOLUTION: Firstly, a raster scanning line is extended from a starting point and when the raster scanning line reaches a bone region, an isosceles triangle is set having the reached point as apex and the axis of the scan line as height and the apex is fixed to specify an isosceles triangle that can maximize the area thereof as far as drawn in the bone region. A scanning is made reversely to the starting point of the raster scanning from the bottom side of the specified isosceles triangle to determine the selected bone area as cortical bone area and if any element of a bone image exists in the region, it is imaged. Upon the completion of the imaging by the processing, the imaged bone region can be separated as cortical bone region and the non-imaged bone region as bone beam region.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a bone image processing method and a bone strength evaluation method. More specifically, the present invention inputs an image obtained by using a means for imaging a continuous cross-sectional structure of a test bone to an image processing apparatus, and separates a bone in the image into a cortical bone region and a trabecular region. Later, cortical bone volume,
An object of the present invention is to provide a method for accurately evaluating bone strength by calculating the maximum connected trabecular component volume, the number of trabecular sides, and the number of cavities.

[0002]

2. Description of the Related Art Bone strength is defined by bone mass and bone quality, and it is said that about 80% of bone strength can be evaluated by bone mass in living bone. As a method of non-invasively quantifying bone mass, that is, bone mineral mass, MD (Micr
o Densitometry method, DIP (Digital Image Processi)
Although there is a method using a simple X-ray film such as the ng) method, this method cannot be applied to a trunk bone which greatly affects muscles and organs, and is limited to application to a peripheral bone such as a finger. Next, the DPA (Dual Photon Absorptiometry) method appeared, and the amount of bone mineral in the lumbar spine was measured.
Method A has not been widely used due to problems with nuclide handling and management areas and low accuracy. DXA (Dual-energy X-ray Ab) uses X-rays instead of nuclides.
With the advent of the sorptiometry method, the DXA method is currently used as a general method for bone mineral quantification.

[0003] Bone density (BMD =
Bone evaluation is performed by measuring bone density (Bone Mineral Density), but bone fracture may or may not occur even with similar bone density. It has been.

Therefore, various image analysis techniques for evaluating bone quality have been proposed. However, StarVolume method, Run Leng
Conventional bone quality evaluation methods such as the th method and the Fractal analysis method mainly target two-dimensional bone tomographic images, so that the evaluation results may differ significantly depending on the measurement site. In addition, when a fracture is assumed, little consideration is given to the direction perpendicular to the imaging plane that should be required.

Further, there is no suitable method for automatically separating a bone region of a bone cross-sectional image into a cortical bone region and a trabecular bone region. There is a case where the image is erroneously recognized as a region, and most of the images need to be manually corrected for each image.

[0006]

SUMMARY OF THE INVENTION The present invention observes the internal structure of a bone to be examined non-destructively and microscopically, automatically separates a cortical bone region and a trabecular region by image processing technology, and The purpose is to find an index that reflects bone strength better than density and a method of measuring the index.

[0007]

Means for Solving the Problems The inventors of the present invention have conducted intensive studies in order to achieve the above object, and as a result, the object photographed by a microfocus X-ray tomography apparatus capable of sufficiently confirming the trabecular structure. By using a plurality of continuous cross-sectional images of bone examination as an input image group and performing various image processing three-dimensionally, it is possible to measure each index having a higher correlation with bone strength than conventionally used bone density. Heading reached the present invention.

That is, according to the present invention, a continuous cross-sectional image of a region of interest of a bone to be examined is photographed by a micro-focus X-ray tomographic photographing apparatus or the like, and these images are binarized to extract only a bone portion, and a trabecular bone. Create a binary cross-sectional image of the bone taken at a resolution that allows confirmation of the structure, extend a raster scan line toward the inside of the image starting from the end of the image, and determine the point at which the scan line hits the bone region. Apex, select an isosceles triangle with the scanning line axis as the height direction, identify the largest isosceles triangle whose bottom is drawn in the bone region, and select the raster scanning line from the bottom of the isosceles triangle. The bone region to be scanned and extracted in the direction of the starting point is determined as a cortical bone region, and if there is a bone image element in the region, the process of setting the element portion to the cortical bone region is repeated for the entire image. Bone image is divided into cortical bone region and trabecular region Bone image processing method for there is provided a.

Further, the present invention provides, as an input image group, a plurality of binarized continuous cross-sectional images of a bone taken at a resolution that allows the trabecular structure to be confirmed, and for each image, only the cortical bone region is obtained using the above method. A bone strength evaluation method characterized in that cortical bone volume is obtained by extracting and correlating the cortical bone region three-dimensionally, and in particular, a method for evaluating a breaking force and a maximum load, which are one of the indexes, is provided. Is what you do.

In addition, the present invention provides a plurality of binarized continuous cross-sectional images of a bone taken at a resolution at which a trabecular structure can be confirmed as an input image group, and for each image, only the trabecular region is determined using the above method. Bone strength evaluation characterized by extracting a trabecular component volume by extracting and connecting the trabecular regions three-dimensionally, and determining the largest trabecular component volume (maximum connected trabecular component volume) among them. It is intended to provide a method, in particular, a method of evaluating axial displacement, which is one of the indexes reflecting bone strength.

Further, the present invention provides a plurality of binarized continuous cross-sectional images of a bone taken at a resolution that allows the trabecular structure to be confirmed as an input image group. The trabecular region component obtained by extracting and connecting the trabecular region three-dimensionally is subjected to three-dimensional thinning processing,
It is an object of the present invention to provide a bone strength evaluation method characterized by obtaining the number of sides of a trabecular bone from the thinned image, in particular, a method of evaluating toughness which is one of the indexes reflecting bone strength.

Further, according to the present invention, a plurality of binarized continuous cross-sectional images of a bone taken at a resolution that allows the trabecular structure to be confirmed are used as an input image group, and only the trabecular region is determined for each image using the above method. Extracting, performing a three-dimensional thinning process on a trabecular region component obtained by connecting the trabecular regions three-dimensionally, and determining the number of cavities appearing as a structure like an eggshell in the thinned image. The present invention provides a method for evaluating bone strength, and a method for evaluating the maximum elastic force, which is one of the indexes reflecting bone strength.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION The preferred embodiment of the bone strength evaluation method of the present invention will be described below. Although the purpose of this embodiment is to evaluate the bone strength of the lumbar vertebrae of the experimental rat, it can be easily applied to bones of other animals including humans and other parts.

First, normal rats for experiments (Sham model)
Five osteoporotic rats (OVX model) were prepared, and each of the third lumbar vertebrae (L3) was imaged by a micro-focus X-ray CT imaging apparatus with a total of 300 cross-sections of 30 sections per body under the following conditions. did.・ Resolution: 25.8 μm / pixel ・ Slice thickness: 20.7 μm

The OVX model refers to an animal model in which the ovaries on one or both of the right and left sides have been removed.
This OVX model is often used in experiments as a model for primary postmenopausal osteoporosis. Sha for OVX model
The m model refers to a group in which the same operation as that of the OVX model animal is performed, but the ovaries are not removed, in order to exclude the effects of the ovariectomy.

In order to separate the cortical bone region and the trabecular region in the present invention, binarization of an image is required. This time, the threshold was determined and binarized based on the results of Otsu's discriminant analysis method.
Various other methods for binarizing an image are known.

FIG. 1 shows an example of a binarized fine focus X-ray CT image of a healthy rat lumbar spine, and FIG. 2 shows a flowchart of an algorithm for separating the binarized image into a cortical bone region and a trabecular region.

In the "setting of the starting point of raster scanning" shown in FIG. 2, all pixels belonging to the starting point shown in FIG. 3 are sequentially defined as starting points for starting raster scanning, and the direction of raster scanning is determined.

The "trisecting line" shown in FIGS. 2 and 3 means the leftmost end of the bone (leftmost wavy line in FIG. 4) when the second moment axis of the bone is the horizontal axis. This is a line that divides the distance at the right end (the rightmost wavy line in FIG. 4) into three equal parts.

When the outer contour of the bone has only a convex edge in the image, the cortical bone region and the trabecular region can be separated by scanning from the end point of the image. Since the image of the rat lumbar vertebra also has a concave edge, the scan line does not reach the shadowed area of the protruding bone region only by scanning from the image end point,
Cortical bone area and trabecular area cannot be completely separated. Therefore, by setting and using the above-mentioned trisecting lines to divide and process the bone image, it becomes possible to separate the cortical bone region and the trabecular region even in the image of the rat lumbar vertebra having a concave edge.

The following processing is performed in "raster scanning &filling" shown in FIG. Extend the raster scan line from the starting point. When the raster scanning line hits the bone region, an isosceles triangle having the point as the apex and the scanning line axis as the height is set, and the isosceles triangle having the maximum area that can be drawn in the bone region with the apex angle constant is specified. The isosceles triangle to be set has a vertex angle of 60 °
It is preferable that the angle is 20 to 60 °
The isosceles triangle in the range is preferred from the viewpoint of analysis accuracy. In this embodiment, an isosceles triangle having a base: height = 1: 2 is set. From the bottom side of the specified isosceles triangle, scanning is performed in the reverse direction toward the start point of the raster scan, and the selected bone region is determined as a cortical bone region, and if there is a bone image element in that region, it is filled.

That is, this algorithm is based on the idea that a period from when a raster scan line enters a bone region in an image to when it passes through a background region is regarded as a cortical bone region. When the conventional method without any correction is used, the raster scan line advances from the cortical bone to the trabecular region, and the trabecular region may be erroneously recognized as the cortical bone region.

As a method of avoiding this, in the present invention, as shown in FIG. 5, the scanning width is increased from the position where the raster scanning line enters the bone region, and finally the isosceles triangle having the largest area is formed. I decided to draw and set the filled area. When the filling by this processing is completed, it is possible to separate the filled bone region as a cortical bone region and the unfilled bone region as a trabecular region.

FIG. 6 shows an image of the cortical bone region separated from the image of FIG.

Next, the above-described image processing is performed on the entire continuous cross-sectional image group of the subject bone, which is the input image group, and the three-dimensionally connected cortical bone regions are created by superimposing the images. And the unit volume [mm ³ / voxe] represented by 1 [voxel]
l], the cortical bone volume [mm ³ ] for each test bone
Was derived.

Next, a method for extracting three-dimensionally connected trabecular regions will be described.

From a global perspective, trabeculae are large
Considered as two connected components. However, a finite number of CT
Considering the trabecular bone in a spatially limited area reflected in the image, each consists of several independent connected components. Therefore, a plurality of three-dimensionally connected trabecular components are extracted from the plurality of two-dimensional trabecular images.

In order to extract three-dimensionally connected trabecular components, it is necessary to perform three-dimensional labeling on n consecutive images. The labeling process used in this embodiment is performed as follows, and its flowchart is shown in FIG.

An image group obtained by extracting only the trabecular region from the binarized image is set as an input image group, and labeling processing is performed on all the image groups. The re-labeling process is performed on the k-th (1 ≦ k <n) image and the (k + 1) -th image among the n consecutive label images in accordance with the vertical correspondence of the trabecular area.

If k = 1: k ← k + 1 and return. When k = n−1: Re-labeling is performed on the image processed before the k-th image according to the result of the k-th image.
The process ends. In other cases: For the image processed before the k-th image, re-labeling is performed according to the result of the k-th image.
Return to k ← k + 1.

Note that "checking the up / down correspondence" in FIG. 7 means that two adjacent images are viewed,
Is to check whether there is at least one trabecular region. If there is an overlapping pixel, it is considered that there is a vertical correspondence.

[0032] That is k th image is has a label area of a _n from a _0, k + 1 th image from b ₀ has a label area of b _m, for example, the label area as shown in FIG. 8 Suppose there is a vertical correspondence.

FIG. 9 shows the upper and lower correspondence in FIG. 8 in a matrix which is easy to understand.

A portion indicating a space between label regions having upper and lower correspondences is indicated by a circle. At this time, the following matters hold. If there is a circle in the same row or the same column, it becomes the same area in three dimensions. If there is no circle in the same column, there is no three-dimensional connection in the area after the (k + 1) -th sheet. If there is no circle in the same row, three-dimensional connection starts from the (k + 1) -th area in that area.

If the first label number is set to 1 and re-labeled sequentially according to this condition, a matrix as shown in FIG. 10 can be created. Here there is no ○ in the same column to a ₃ in FIG. 9, the region indicated by a ₃ indicates that there is no three-dimensionally connected to k + 1 th and subsequent sheets, the a ₃ is Give a sufficiently large label value. Also there is no ○ on the same line in b ₃ in FIG. 9, the region indicated by b ₃ shows that the first three-dimensional coupling of k + 1 th and subsequent sheets is started, the b ₃ new Gives a label value indicating the area. If the above labeling process is performed, FIG. 8 can be drawn as shown in FIG.

However, due to the re-labeling of the k-th sheet, there is no label correspondence between the k-th sheet and the (k-1) -th image. Therefore, according to the result of the labeling processing of the k-th sheet, the change in the label value of a is reflected on the labels of the labeling-processed images before the (k-1) -th sheet. Then, after performing the processing, according to the label value of b, k
Labeling processing is sequentially performed on the + 1st and subsequent images. By applying this process to the entire n continuous bone cross-sectional images,
The three-dimensional labeling process is completed.

More specifically, for example, assume that there are five trabecular components in the K-th image. At that time, label numbers [1], [2], [3], [4], [5] are assigned to each trabecular component. Thereafter, the (K + 1) th image is examined. It is assumed that there are six trabecular components in the (K + 1) th image.

Five of the six trabecular components are the trabecular components labeled [1], [2], [3], [4], and [5] in the Kth image. If there is a vertical correspondence, five of the trabecular components in the (K + 1) th image can be numbered [1] to [5], and the vertical correspondence with the Kth image Unrecognized trabecular components are given a new label number [6].

Subsequently, the (K + 2) th image is examined. Assume that the K + 2 image has only four trabecular components. When examining the vertical correspondence of the trabecular components of the K + 1 sheet and the K + 2 sheet, K + 1
The trabecular component labeled [1] in the first image does not correspond to any of the trabecular components of the K + 2nd trabecular component. The trabecular component labeled with] is given a sufficiently large label number (here relabeled as [100]), as it does not appear in the second and subsequent images. Until now, the label number of the trabecular component with the label number [1] is the K + 1st image
Since it is [100], all the label numbers [1] are re-labeled as [100] for the images before the K-th image.

Next, as a result of examining the trabecular components which had been [2] and [3] up to the K + 1st sheet between the K + 2nd sheet, the trabecular components of [2] and [3] were obtained. Assume that there is an up-down correspondence with one trabecular component in the K + 2nd image. In that case, it can be understood that the trabeculae that had been recognized as a separate trabecular component from [2] and [3] were actually connected. Therefore, the trabecular component in the K + 2nd image is given a label number [2], and the label number of the trabecular component which was [3] in the image before the K + 1th image Is relabeled as [2].

When none of the remaining three trabecular components in the K + 2nd image correspond to the trabecular component in the K + 1st image in the up-down direction, [7], Label numbers [8] and [9] are given. As a result of the above processing, the label numbers of the trabecular components of the Kth image are [100], [2],
[2], [4], [5] The label number of the trabecular component of the K + 1st image is [100],
[2], [2], [4], [5], [6] The label numbers of the trabecular components of the K + 2nd image are "2", [7],
[8] and [9]. The above processing is repeated endlessly for the entire image.

FIGS. 12, 13 and 14 show examples of three-dimensional connected trabecular components obtained by three-dimensional labeling by applying the processing method described above. FIG. 12 shows the entire obtained three-dimensional connected trabecular component, and FIG. 13 shows the region A in FIG.
FIG. 14 shows the three-dimensional connected trabecular component indicated by.
It shows a three-dimensional connected trabecular component indicated by region B.

After creating the three-dimensionally connected trabecular area,
As in the case of deriving the cortical bone volume based on the unit volume [mm ³ / voxel] represented by [voxel], the volume for each of the three-dimensionally connected trabecular region components was derived. Then, the three-dimensional connected trabecular component having the maximum volume for each test bone was determined, and the obtained value was defined as the maximum connected trabecular component volume [mm ³ ] for each test bone.

Next, a method for determining the number of sides of the three-dimensionally connected trabecular bone will be described.

First, in order to facilitate the analysis of the trabecular structure,
A three-dimensional thinning process was performed on the three-dimensional connected trabecular component obtained by the above method. The three-dimensional thinning is a process of converting a thick or thick figure into a three-dimensional line figure in a three-dimensional space. In the three-dimensional line figure, special processing such as branching, intersection, and cavity is performed. The thickness is 1 except for the appropriate locations, and no erasable pixels are included except for the end points. The three-dimensional line graphic is called a three-dimensional skeleton of the original graphic.

Various methods have been known for the three-dimensional thinning method. In the present invention, "Yasue Masahiro et al., Comparative evaluation of the ability of thinning method of three-dimensional grayscale image and application to medical images, IEICE Transactions, Volume J79-D-II,
pp. 1664-1674, 1996 ", a three-dimensional thinning process of the three-dimensional connected trabecular component was performed. FIG. 15 shows an example of the image after the three-dimensional thinning processing.

When the three-dimensional skeleton of the trabecular component obtained from the continuous cross-sectional image group of each test bone is considered as a three-dimensional graph, information such as end points, sides, and branch points included in the trabecular structure can be obtained. become. That is, to determine the number of sides,
In the three-dimensional skeleton, it is sufficient to count the sides whose distance from the vertex to the vertex is 1 or more. Therefore, this process was performed for each of the test bones, and the number of sides of the trabecular bone of each test bone was determined.

Next, a method for obtaining the number of cavities appearing in the three-dimensional skeleton of the three-dimensional connected trabecular component will be described.

When a three-dimensional image is generated from a group of cross-sectional images of a plurality of bones and further subjected to three-dimensional thinning processing to obtain a three-dimensional skeleton, a structure such as an egg shell is contained in the three-dimensional skeleton. Appears. The structure does not appear in the two-dimensional image processing, but first appears in the three-dimensional image processing, and is called a cavity.

The processing actually performed to determine the number of cavities will be described below. The three-dimensional skeleton of the three-dimensional connected trabecular component of each test bone is obtained by the above-described method. From the obtained skeleton, a line segment whose edge, that is, the distance from the vertex to the vertex is larger than 1 is deleted. Since only the cavities remain due to the above processing, the number of the cavities is counted and set as the number of cavities.

The above treatment was performed for each of the test bones, and the number of cavities of each test bone was determined.

According to the above-described method, the cortical bone volume, the maximum connected trabecular component volume, the number of trabecular sides,
FIG. 16 shows the result of obtaining four indices of the number of cavities.

In order to compare the four indices obtained above with the bone mineral density (BMD) used as the most general bone quality evaluation index, the third lumbar vertebrae of 10 experimental rats (test bones) were used. In the present example, the bone density was measured for all the same bones as the test bone from which the four indices according to the present invention were derived. The measuring device is QDR-20 manufactured by Holic.
The measurement was carried out by the DXA method using No. 00. The measurement results of the bone density were normalized so that the average of the Sham group was 1, and
The VX group was also derived as a relative bone density by calculating according to the results. FIG. 17 shows the measurement results.

Subsequently, the bone strength of the fourth lumbar vertebra of the 10 experimental rats was measured by a compression test. At the time of the compression test, each lumbar vertebra was embedded and fixed in a dental resin, the most developed portion of the trabecular structure was set as a hatched portion as shown in FIG. 18, and a load was applied at a displacement speed of 6 [mm / min]. As a result, five kinds of indices, ie, breaking force, maximum load, axial displacement, toughness, and maximum elastic force, which are indicators reflecting bone strength, were measured. The breaking force [N] refers to the load applied when the test piece fractured, and the maximum load [N] refers to the maximum load applied before the fracture. Further, the axial displacement [mm] is the amount of deformation of the test piece during the breakage after the load is first applied to the test piece, and the toughness [N · mm] is the amount of the test piece before the breakage. Energy applied to the body. Further, the maximum elastic force [N / mm] indicates rigidity, rigidity, difficulty in bending, and the like, and the description of the above five types of indices is shown in FIG. Also, normalization is performed so that the average of the Sham group becomes 1, and the OV
FIG. 20 shows the results of the compression test derived as relative bone strength by calculating the X group in accordance with the results.

Next, the cortical bone volume, the maximum connected trabecular component volume, the number of trabecular sides and the number of cavities derived by the present invention, the breaking force, the maximum load, the axial displacement, The correlation with toughness and maximum elasticity was investigated. Similarly, bone density (BMD), which has been conventionally used as an index for bone evaluation, and the breaking force, maximum load, axial displacement, toughness,
The correlation with the maximum elastic force was also investigated.

As a result, the cortical bone volume was determined by the breaking force, the maximum load,
The maximum connected trabecular component volume is axial displacement, the number of trabecular sides is toughness,
It was found that the number of cavities had a high correlation with the maximum elastic force. FIG. 21 shows the calculation result of the correlation coefficient. Similarly, the calculation results of the correlation coefficient between the bone density and the breaking force, the maximum load, the axial displacement, the toughness, and the maximum elastic force are also shown in FIG. 21 at the same time. For reference, FIG. 22 shows a breaking force-cortical bone volume correlation diagram, FIG. 23 shows a breaking force-bone density correlation diagram, FIG. 24 shows a maximum load-cortical bone volume correlation diagram, and FIG. 24 shows a maximum load-bone density correlation diagram. 25, a correlation diagram of axial displacement-maximum connected trabecular component volume, FIG. 26, a correlation diagram of axial displacement-bone density, and FIG. 28 a correlation diagram of toughness-trabecular number.
FIG. 29 shows a correlation diagram of toughness-bone density, FIG. 30 shows a correlation diagram of maximum elastic force-number of cavities, and FIG. 31 shows a correlation diagram of maximum elastic force-bone density.
Shown in

FIG. 21 shows that four indices according to the present invention, namely, cortical bone volume, maximum connected trabecular component volume, number of trabecular sides,
The number of cavities has a higher correlation with indexes (breaking force, maximum load, axial displacement, toughness, maximum elastic force) that reflect bone strength than conventionally used bone density (BMD), and a more accurate evaluation has been achieved. It can be said that it can be done.

[Brief description of the drawings]

FIG. 1 is an example of a microfocus X-ray CT image (after binarization) of a healthy rat lumbar spine.

FIG. 2 is a flowchart of an algorithm for separating a cortical bone region and a trabecular bone region.

FIG. 3 is a raster scanning pattern.

FIG. 4 shows the starting point and direction of raster scanning.

FIG. 5 is a method for determining a cortical bone filling region.

FIG. 6 is an example of a separated cortical bone region image.

FIG. 7 is a flowchart of three-dimensional labeling.

FIG. 8 is an upper and lower correspondence of each label area.

FIG. 9 is an upper / lower correspondence table of each label area.

FIG. 10 is an upper / lower correspondence table of each label area (after labeling processing).

FIG. 11 shows labeling of three-dimensional connected components.

FIG. 12 shows an example of a three-dimensional connected trabecular component.

FIG. 13 shows a three-dimensional connected trabecular bone component (region A).

FIG. 14 shows a three-dimensional connected trabecular bone component (region B).

FIG. 15 is an example of a skeleton of a three-dimensional connected trabecular component.

FIG. 16 shows a measurement result of an index according to the present invention.

FIG. 17 shows measurement results of relative bone density.

FIG. 18 is a hatched portion during a compression test.

FIG. 19 shows five types of indexes reflecting bone strength.

FIG. 20 shows the results of a compression test.

FIG. 21 shows correlation coefficient calculation results.

FIG. 22 is a correlation diagram of breaking force-cortical bone volume.

FIG. 23 is a correlation diagram of breaking force-bone density.

FIG. 24 is a correlation diagram of maximum load-cortical bone volume.

FIG. 25 is a correlation diagram of maximum load-bone density.

FIG. 26 is a correlation diagram of axial displacement-maximum connected trabecular component volume.

FIG. 27 is a correlation diagram between axial displacement and bone density.

FIG. 28 is a correlation diagram of toughness-trabecular number.

FIG. 29 is a correlation diagram of toughness and bone density.

FIG. 30 is a correlation diagram of the maximum elastic force-the number of cavities.

FIG. 31 is a correlation diagram of maximum elastic force-bone density.

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Osamu Oshiro 5-16, Takayama-cho, Ikoma, Nara Prefecture Inside the Nippon Advanced Institute of Science and Technology (72) Inventor Kunihiro Chihara 5-16, 8816 Takayama-cho, Ikoma, Nara Prefecture Nara F-Term at the National Institute of Science and Technology F-term (reference) 4C093 AA22 CA50 DA10 FF08 FF23

Claims (5)

[Claims] 1. In a binary cross-sectional image of a bone taken at a resolution at which a trabecular structure can be confirmed, a raster scan line is extended from the end of the image toward the inside of the image, and the scan line extends to a bone region. Select the isosceles triangle with the hit point as the vertex and the scanning line axis as the height direction, specify the largest isosceles triangle whose bottom side can be drawn in the bone region, and from the base of the isosceles triangle. A bone region to be scanned and extracted in the direction of the start point of the raster scanning line is determined as a cortical bone region, and if there is a bone image element in the region, a process of setting the element portion as a cortical bone region is performed on the entire image. A bone image processing method for separating a bone image into a cortical bone region and a trabecular bone region by repeating. 2. An input image group comprising a plurality of binarized continuous cross-sectional images of a bone taken at a resolution at which a trabecular structure can be confirmed,
A bone strength evaluation method, wherein only a cortical bone region is extracted from each image by using the method according to claim 1, and the cortical bone volume is obtained by connecting the cortical bone regions three-dimensionally. 3. An input image group comprising a plurality of binarized continuous cross-sectional images of a bone taken at a resolution at which a trabecular structure can be confirmed,
For each image, only the trabecular region is extracted using the method according to claim 1, and the trabecular region volume is obtained by connecting the trabecular regions three-dimensionally. A bone strength evaluation method characterized by obtaining a maximum connected trabecular bone component volume). 4. A plurality of binarized continuous cross-sectional images of a bone taken at a resolution that allows confirmation of a trabecular structure as an input image group.
The method according to claim 1, wherein only the trabecular region is extracted from each image, and the trabecular region component obtained by connecting the trabecular regions three-dimensionally is subjected to three-dimensional thinning processing. A bone strength evaluation method characterized by obtaining the number of sides of a trabecular bone from a digitized image. 5. An input image group comprising a plurality of binary continuous cross-sectional images of a bone taken at a resolution at which a trabecular structure can be confirmed,
Only the trabecular region is extracted from each image using the method according to claim 1, and a three-dimensional thinning process of a trabecular region component obtained by connecting the trabecular regions three-dimensionally is performed. A bone strength evaluation method, wherein the number of cavities is determined.