KR20200082746A - Cell-zone labeling apparatus and cell-zone detecting system including the same apparatus - Google Patents

Abstract

One embodiment of the present invention relates to a cell-zone labeling apparatus for reading a bone marrow image, which automatically labels and detects a cell zone to read a bone marrow image. To this end, according to one embodiment of the present invention, the cell-zone labeling apparatus comprises: an image scaling unit down-scaling a stitching image wherein a plurality of divided images acquired by motion-control photography of a sample bone marrow slide are stitched; a color conversion unit converting RGB color of the scaled stitched image into an image processing color space; a segmentation unit applying multiple thresholding including high thresholding and low thresholding to divide the stitched image converted into the image processing color space; a morphology conversion unit enhancing a cell nuclei image by morphology opening with respect to the stitched image divided by applying high thresholding and enhancing a cell continuity image by morphology dilation with respect to the stitched image divided by applying low thresholding; a labeling unit detecting and labeling a plurality of single cells based on the enhanced cell nuclei image and detecting and labeling particles or abnormal cells based on the enhanced cell continuity image; and an occluded cell filtering unit removing one or more occluded single cells occluded with the particles or abnormal cells labeled from the plurality of labeled single cells.

Description

Cell zone labeling device for bone marrow reading and cell zone detection system including the same {CELL-ZONE LABELING APPARATUS AND CELL-ZONE DETECTING SYSTEM INCLUDING THE SAME APPARATUS}

The present invention relates to a supporting device and system for bone marrow reading, and more particularly, to a supporting device and system for bone marrow reading performed based on a captured image of a bone marrow specimen smeared in a slide.

The bone marrow is a soft tissue located in the inner space of the bone and is a hematopoietic organ that produces blood cells such as red blood cells, white blood cells, and platelets. In addition, the bone marrow test is a function of bone marrow by observing bone marrow cells through a microscope and performing differential calculations after aspirating and biopsiing the bone marrow from large bones and staining them if necessary in the case of problems with blood cell production in the human body. This is a test to check for abnormal lesions. This bone marrow test is used as a diagnosis of normal leukocyte levels or abnormal leukocyte levels, whether there are cells that should not be among abnormal leukocytes, or other leukemia diagnosis data.

This bone marrow examination or bone marrow reading process is a task that takes a long time and requires fatigue of a diagnostic test (or an analyst) because several parts of the slide must be observed at high magnification. Furthermore, the manual operation of counting each white blood cell by counting all the white blood cells present in the bone marrow specimen image and counting the number of white blood cells of each type is not only time-consuming, but also highly unlikely depending on the degree of experience of the diagnostic test.

Meanwhile, in order to increase the efficiency and accuracy of bone marrow reading, the diagnostic examiner establishes a'Cell-Zone' for observation of bone marrow samples. The term'cell zone' refers to a number of single cells that are ideal for observing and reading slide specimens at high magnification. That is, the'cell zone' may be defined as a limited number of single cells with high reliability that can ensure the accuracy of bone marrow reading.

Bone marrow reading utilizes images obtained from a bone marrow specimen slide that has been subjected to smearing and staining. All regions of the slide specimen are reliable because the spreading thickness and uniformity of the specimen according to the smear are different and cannot be stained singly It cannot be. Therefore, if all areas are arbitrarily observed at high magnification, there is a high probability of failing to read, and repeated reading failure is inevitable due to deterioration of work efficiency and long labor. Therefore, the experienced and highly skilled diagnostic examiners are performing observation and reading on the above-mentioned'cell zone', that is, a limited number of reliable single cells.

However, there is still a problem in that the selection of the'cell zone' by conventional manual operation and the resultant bone marrow reading are accompanied by a decrease in work efficiency and long hours of labor for diagnostic examiners with low experience and skill. According to apprenticeship education, a method of raising the level of diagnostic tests with low skill is used, but it takes a lot of time and trial and error.

Moreover, even in the case of a diagnostic test with high experience and skill, the conventional'cell zone' selection method has a disadvantage in that it is difficult to have consistency because subjective judgment of a person is involved, and may lead to different reading results depending on the diagnostic test.

Therefore, there is a need for a study on the automatic detection of cell zones through image processing and analysis of bone marrow specimen slides.

The present invention has been derived based on the above needs, and the first object of the present invention is to label the cell zone with high reliability and ideality for constructing a cell zone through an image processing process on an image obtained from a bone marrow specimen slide. It is to provide a cell zone labeling device for the bone marrow reading that can be selected as a candidate cell.

The second object of the present invention is to detect a border region from an image obtained from a bone marrow specimen slide and align cell candidate cells with respect to a leveled border to generate a level map or a confidence map for the cell zone. It is to provide a cell zone detection system for reading.

A third object of the present invention is to provide a cell zone detection system for bone marrow reading which can improve efficiency of a high-magnification imaging and reading process by giving observation priority according to level or reliability to cell zone candidate cells constituting a cell zone. .

An object of the present invention as described above, in a supporting system for bone marrow reading performed based on an image captured of a bone marrow specimen smeared in a slide, stitching a plurality of divided images obtained by motion control imaging of a bone marrow specimen slide An image scaling unit for down-scaling one stitching image; A color conversion unit that converts the RGB color of the scaled stitched image into a color space for image processing; A segmentation unit for classifying the stitched image converted into the color space for image processing by applying multiple thresholding including high-thresholding and low-thresholding; High-thresholding is applied to enhance the cell nucleus image by morphology opening operation on the divided stitching image, and low-thresholding is applied to morphology dilation on the differentiated stitching image. A morphology transformation unit that enhances the cell continuity image by calculation; A labeling unit to detect and label a plurality of single cells based on the enhanced cell nuclear image, and to detect and label particles or abnormal cells based on the enhanced cell continuity image; And an occluded cell filtering unit that removes one or more occluded single cells occluded with labeled particles or labeled abnormal cells from a plurality of labeled single cells, and can be achieved by providing a cell zone labeling device for bone marrow reading. .

Here, the color space for image processing may be an HSV color space.

The cell zone labeling device for bone marrow reading applies a variance value or standard deviation value based on the cell distance value to a plurality of single cells labeled, or a cell zone that constitutes a cell zone of a single cell having a variance value or a standard deviation value greater than a set value. A cell zone candidate cell selection unit for selecting as a candidate cell may be further included.

In addition, the cell zone labeling device for bone marrow reading sets the mask size corresponding to the low magnification angle of view of the motion-controlled imaging around the cell zone candidate cells, and groups the cell zone candidate cells into the cell zone candidate groups based on the grouping conditions using the set mask size. A grouping part may be further included.

On the other hand, the object of the present invention includes a cell zone labeling device for bone marrow reading as described above, removes noise against a stitched image to which low-thresholding is applied, and searches vertically to detect a horizontal beginning point of a boundary line. A boundary line detection unit; A boundary leveling unit that performs low-pass filtering on the stitched image to which low-thresholding is applied, and normalizes the boundary line by normalizing it to a counting result value of a horizontal reference effective pixel; And a level composition unit that aligns the cell zone candidate cells from a high level to a low level of the leveled boundary line, thereby providing a cell zone detection system for bone marrow reading.

In addition, the cell zone detection system for bone marrow reading may further include a confidence map generation unit that generates a cell-zone confidence map based on predetermined number information of cell zone candidate cells.

In addition, the cell zone detection system for bone marrow reading may further include a high magnification imaging unit configured to image the cell zone candidate cells arranged in a high to low level order at high magnification.

In addition, the cell zone detection system for bone marrow reading may further include an alignment unit for relocating aligned cell zone candidate cells in the down-scaled stitching image based on the original stitching image before down-scaling.

According to an embodiment of the present invention as described above, for an image obtained from a bone marrow specimen slide, an effect of labeling and selecting a cell zone candidate cell for a highly reliable single cell capable of constructing a cell zone through an image processing process is obtained. have.

In addition, a level map or a confidence map for the cell zone may be generated by detecting the boundary region from the image obtained from the bone marrow sample slide and aligning the cell zone candidate cells with respect to the leveled boundary.

In addition, it is possible to improve efficiency of a high magnification imaging and reading process by giving an observation priority according to level or reliability to the cell zone candidate cells constituting the cell zone.

1 is a configuration diagram showing an embodiment of the cell zone labeling device for bone marrow reading according to the present inventors,
FIG. 2 is a view showing an image (bottom) of the original image (top) downscaled by a 1/4 ratio during the configuration of an embodiment of the cell zone labeling device for bone marrow reading according to the present inventors;
3 is a view showing an example in which a color conversion unit converts an RGB color space (top) into an HSV color space (bottom) during configuration of an embodiment of the cell zone labeling apparatus for bone marrow reading according to the present inventors;
FIG. 4 shows the results of the segmentation by applying the high-thresholding (top) and low-thresholding (bottom) of the color conversion unit to the segmentation unit of the cell zone labeling apparatus for bone marrow reading by the present inventors, and the binary images of each of them. Drawing showing an example of,
5 is a view showing an example of a morphology result of a high-thresholding morphology (top) and a low-thresholding morphology (bottom);
6 is a view showing an example of a normal cell labeling state (green mark above) and abnormal cell labeling state (green box below) during configuration of an embodiment of the cell zone labeling device for bone marrow reading according to the present inventors;
Figure 7 is a cell zone labeling device for the bone marrow readout of the present invention in the configuration of the occluded cell filtering unit in the normal cell labeling state (green mark in the image above) of the removed cells (green mark in the image below) Example drawing,
8 is a view showing an example of a candidate cell (blue mark) extracted by the cell zone candidate cell selection unit during the construction of an embodiment of the cell zone labeling device for bone marrow reading by the present inventors;
9 is a view showing an example of a state in which a data grouping unit selects a mask corresponding to an angle of view (blue box on the left) and grouped state (blue box on the right) during an embodiment of the cell zone labeling apparatus for bone marrow reading according to the present inventors;
10 is a block diagram showing an embodiment of the cell zone detection system for bone marrow reading according to the present inventors,
11 is a view showing an example of a state in which a boundary detection unit and a boundary leveling unit remove noise and level a boundary area in an embodiment of the cell zone detection system for bone marrow reading according to the present inventors;
12 is a view showing an example of a state in which the level composition unit aligns cell zone candidate cells according to levels in one embodiment of the cell zone detection system for bone marrow reading according to the present inventors;
13 is a view showing an example of a state in which the alignment unit corrects and relocates a single cell in the configuration of an embodiment of the cell zone detection system for bone marrow reading according to the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and the contents described in the accompanying drawings, but the present invention is not limited or limited by the embodiments.

Various modifications may be made to the embodiments described below. The examples described below are not intended to be limiting with respect to the embodiments, and should be understood to include all modifications, equivalents, or substitutes thereof.

On the other hand, in the description of the present invention, when it is determined that a detailed description of related known functions or configurations may unnecessarily obscure the subject matter of the present invention, the detailed description will be omitted. In addition, terms used in the present specification (terminology) are terms used to properly represent an embodiment of the present invention, which may vary depending on a user, an operator's intention, or a custom in a field to which the present invention pertains. Therefore, definitions of these terms should be made based on the contents throughout the present specification.

For bone marrow reading Cell zone Labeling device

An embodiment of the cell zone labeling device 10 for bone marrow reading includes an image scaling unit 110, a color conversion unit 120, a segmentation unit 130, a morphology conversion unit 140, a labeling unit 150, and occluded cells It may be configured to include a filtering unit 160, as shown in Figure 1, it may be configured to further include a cell zone candidate cell selection unit 170 and the data grouping unit 180.

Hereinafter, the configuration of the present embodiment will be described in detail with reference to FIGS. 1 to 9.

The cell zone labeling device of this embodiment is a bone marrow reading support device that efficiently labels and filters ideal cells to provide a reliable cell zone to a diagnostician through processing and analysis of an image of a bone marrow specimen slide.

The term'Cell-Zone' refers to cells that are the observation targets limited to the number required for bone marrow reading from the image of the slide sample after the smearing and staining process of the bone marrow sample. Since these cell zones have various types of cells, such as normal cells, abnormal cells, and particles, on the slide, there is a favorable state for reading or a difficult state for reading, so selecting a reliable cell zone for bone marrow reading is an important factor.

The image scaling unit 110 serves to downscale a stitching image stitching a plurality of divided images obtained through motion control imaging of a bone marrow specimen slide. Here, the bone marrow sample slide is prepared for imaging in a smeared and stained state for observation of cells, and a plurality of divided images corresponding to a plurality of divided regions of the bone marrow sample slide is acquired through scanning. In addition, a stitching image means an image obtained by connecting a plurality of divided images with adjacent ones.

For the above-described scanning, the imaging means including the low magnification lens and the image sensor, the supporting means for fixing the slide, the lighting means for applying the lighting, and the motion of the imaging means for completely imaging the divided area of the slide are controlled. The position control means and the like can be provided.

Down-scaling of the image scaling unit 110 is necessary to minimize noise or blurring as shown in FIG. 2 and to improve processing speed according to a small size.

The color converter 120 serves to convert the RGB color of the downscaled stitched image into a color space for image processing. There are HSI, HSV, HSB, HSL, and the like for image processing color space, but in this embodiment, HSV color space was used. Here, among H and S, H means Hue, which is a color angle based on red, and S means color density, that is, saturation. I, V, B, and L represent brightness information, and mean intensity, value, brightness, and brightness, respectively. The color converter 120 may perform color conversion from the RGB color space (above) to the HSV color space, as shown in the example shown in FIG. 3.

The segmentation unit 130 serves to classify the stitched image converted into a color space for image processing by applying multiple thresholding, including high thresholding and low thresholding. . This applies to high-thresholding for segmentation processing based on the white blood cell image present on the slide, and low-thresholding is applied to classify the image containing red blood cells. Each threshold value can be set around the optical system. FIG. 4 shows an example of a color image resulting from high-thresholding and its binary image (top), and a color image resulting from low-thresholding and its binary image (bottom).

The morphology transformation unit 140 reinforces the cell nucleus image by performing a morphology opening operation on the divided stitching image by applying high-thresholding, and applies the low-thresholding to the differentiated stitching image. It serves to enhance the image of cell continuity by calculating morphology dilation. The morphology opening operation enhances the cell nucleus image, which is advantageous for single cell detection, and at low-threshold, the morphology delay can enhance the continuity image of cells, thus detecting particles or abnormal cells, and described below. It is advantageous for boundary detection of cells.

The labeling unit 150 serves to detect and label a plurality of single cells based on the enhanced cell nuclear image, and to detect and label particles or abnormal cells based on the enhanced cell continuity image. The labeling information of each single cell and particle or abnormal cell may include location information and size information. In the upper image of FIG. 6, an example of a state (green mark) in which the labeling unit 150 labels a plurality of single cells having a normal cell size, and in the lower image of FIG. 6, the labeling unit 150 is abnormally sized An example of a state (green box) labeled for abnormal cells or particles, which are large cells, is shown.

The occluded cell filtering unit 160 serves to remove one or more occluded single cells that are occluded with labeled particles or labeled abnormal cells from a plurality of labeled single cells. This is to exclude the occluded single cells in the occluded state because they are placed in a region that decreases cell zone reliability. 7 shows a state in which such occluded single cells are removed.

The cell zone candidate cell selection unit 170 applies a variance value or a standard deviation value based on the cell distance value to each of a plurality of labeled single cell cell labeling devices for bone marrow reading, thereby setting a single or variance value or a standard deviation value. It serves to select a single cell larger than the value as a cell zone candidate cell constituting the cell zone.

The more favorable the cell zone is to read, the higher the reliability. Therefore, the greater the regional dispersion or the presence of a single regionally independent cell, the more likely it is to be selected as a cell zone candidate cell. That is, when calculating the distribution by classifying the local size, it is preferable to select a cell zone candidate cell when the distribution is large or exists as a single cell in relation to surrounding cells. Here, each of the labeled single cells is calculated to have a variance value or a standard deviation value. An example of a cell zone candidate cell selected (extracted) considering the variance value or standard deviation value is shown in FIG. 8 (blue mark).

The data grouping unit 180 sets the mask size corresponding to the low magnification angle of view of the motion control imaging centered on the cell zone candidate cell, and the cell zone labeling device for bone marrow readout selects the cell zone candidate cell based on the grouping condition using the set mask size. It serves to group into cell zone candidates. Here, the grouping condition is grouped so that if the set mask size is intersected more than a predetermined number of times or overlaps with surrounding cells, it is absorbed by cells having a large number of intersecnes. As shown in FIG. 9, the overlapping of the mask size is excessive (the blue box on the left) because there is a mask size as many as the cell zone candidate cells before grouping. However, after grouping, you can eliminate duplicates and reduce them to the appropriate number (blue box on the right).

The data grouping unit 180 excludes overlapping imaging of the same cell zone candidate cells when the diagnostic tester to read attempts high-magnification imaging for the cell zone detected in the last step, thereby increasing the reading efficiency and tact time according to repeated imaging. time).

For bone marrow reading Cell zone Detection system

An embodiment of the cell zone detection system for bone marrow reading according to the present inventors may include a cell zone labeling device 10, a border line detection unit 20, a border line leveling unit 30, and a level composition unit 40, As shown in FIG. 10, it may be configured to further include a confidence map generation unit 50, an alignment unit 60, and a high magnification imaging unit 70.

The reliability of the cell zone is higher as it is located at the boundary of the slide than in the case where the above-described regional dispersion degree is large or exists as a single cell. This is because cells located in a thin portion of the staining of the smeared sample slide are advantageous for reading. In addition, the cell zone reliability is higher when there is no cell turnover (cell overlapping).

Since the cell zone labeling device 10 has been described above, description thereof will be omitted, and other configurations of this embodiment will be described below with reference to FIGS. 10 to 13.

The boundary line detection unit 20 serves to detect a horizontal beginning point of a boundary line by removing noise and vertically searching for a stitched image to which low-thresholding is applied. The border image is emphasized because the stitching image with low-thresholding is an enhanced image of cell continuity. Therefore, using this, the noise on the 2D image is removed by using at least one of a median filter, a Gaussian filter, and a bilateral filter, and vertical noise is horizontally removed. The boundary can be detected by repeating the detection of the starting point.

The boundary level leveling unit 30 performs low pass filtering on the stitched image to which the low threshold is applied, and normalizes the result to a horizontal pixel based on the valid value of the valid pixel counting (valid pixel counting). It serves to level. In other words, if the high-frequency signal is removed through low-pass filtering, a boundary area (or boundary area) can be leveled by creating a noise-resistant boundary area around the boundary line and counting and normalizing the effective pixel count on a horizontal basis. As a result, an absolute level map and a relative level map are derived considering the number of effective pixels, and a function is selectively selected. In FIG. 11, the boundary detection unit 20 and the boundary leveling unit 30 show an example of a leveled boundary in which noise is removed and filtered.

The level composition unit 40 serves to align cell zone candidate cells from a high level to a low level of the leveled boundary line. The level composition unit 40 aligns the cell zone candidate cells selected by the cell zone labeling apparatus 10 according to the level of the leveled boundary line. In the end, the level means the priority of the ideal cell, and the alignment according to the level also means the order of detection and cell zone provision for the cell zone required by the diagnostic examiner using one embodiment of the present invention.

When the high magnification imaging unit 70 attempts high magnification imaging based on the detected cell zone, high magnification imaging may be attempted from a single standard cell having a high standard deviation value.

The confidence map generating unit 50 serves to generate a cell-zone confidence map based on the preset number information of the cell zone candidate cells. The preset number of pieces of information may be input information or preset information when working with a diagnostic tester using the present invention. 12 shows an example of a cell-zone confidence map showing cell-zone candidate cells mapped according to levels.

In addition, the cell zone detection system for bone marrow reading may further include a high magnification imaging unit configured to image the cell zone candidate cells arranged in a high to low level order at high magnification.

The alignment unit 60 serves to rearrange the aligned cell zone candidate cells in the downscaled stitching image based on the original stitching image before downscaling. Since the cell has a variable shape, a positional error occurs accordingly, and a refinement is required to correct it.

Specifically, when the original stitching image is segmented on the basis of high-thread hold to emphasize the nucleus and the binary image is used, the peak point shift according to region information and neighborhood searching can be known. The downscaled cell zone candidate cells can be relocated to the correct location corresponding to the original stitching image. 13 shows an example in which the center position before and after the alignment is changed through an alignment process.

The embodiments of the present invention have been described above with reference to the accompanying drawings, but the technical configuration of the present invention described above is another specific form without a person skilled in the art to which the present invention pertains does not change the technical spirit or essential features of the present invention. It will be understood that can be carried out. Therefore, the above-described embodiments are to be understood in all respects as illustrative and not restrictive. In addition, the scope of the invention is indicated by the claims below, rather than the detailed description above. In addition, all modifications or variations derived from the meaning and scope of the claims and equivalent concepts should be construed as being included in the scope of the present invention.

10: Cell zone labeling device
20: boundary line detection unit
30: border leveling unit
40: level composition unit
50: confidence map generator
60: Alliance Division
70: high magnification imaging unit
110: image scaling unit
120: color conversion unit
130: segmentation unit
140: morphology conversion unit
150: labeling unit
160: occlusion cell filtering unit
170: cell zone candidate cell selection unit
180: data grouping unit

Claims (8)

A support system for bone marrow reading performed based on an image captured of a bone marrow specimen smeared in a slide, comprising:
An image scaling unit for down-scaling a stitched image stitching a plurality of divided images obtained by motion control imaging of a bone marrow specimen slide;
A color conversion unit that converts the RGB color of the scaled stitched image into a color space for image processing;
A segmentation unit for segmenting the stitched image converted into the color space for image processing by applying multi-thresholding including high-thresholding and low-thresholding;
The cell nucleus image is strengthened by a morphology opening operation on the stitched image segmented by applying the high-thresholding, and a morphology delay is applied to the segmented image segmented by applying the low-thresholding. (morphology dilation) morphology transformation unit to enhance the cell continuity image by calculation;
A labeling unit for detecting and labeling a plurality of single cells based on the enhanced cell nuclear image, and for detecting and labeling particles or abnormal cells based on the enhanced cell continuity image; And
A cell zone labeling device for bone marrow reading, comprising an occluded cell filtering unit that removes the labeled particles or one or more occluded single cells occluded with the labeled abnormal cells from the labeled single cells.
According to claim 1,
The image processing color space is a HSV color space cell zone labeling device for bone marrow reading.
According to claim 1,
A cell zone that selects a single cell having a variance or standard deviation value based on a cell distance value on the plurality of labeled single cells or a single cell having a larger variance value or a standard deviation value than a set value as a cell zone candidate cell constituting a cell zone. Cell zone labeling device for bone marrow reading further comprising a candidate cell selection unit.
According to claim 3,
Further comprising a data grouping unit for setting a mask size corresponding to the low magnification angle of view of the motion control imaging centering on the cell zone candidate cells, and grouping the cell zone candidate cells into a cell zone candidate group based on the grouping conditions using the set mask size. Cell zone labeling device for bone marrow reading.
A cell zone labeling device for bone marrow reading according to claim 3 or 4,
A boundary detector for removing noise from the low-threshold stitched image and searching vertically to detect a horizontal beginning point of the boundary line;
A boundary level leveling unit performing low-pass filtering on the stitched image to which the low-thresholding is applied, and leveling the boundary line by normalizing to a result value counting of the effective pixels of a horizontal reference; And
A cell zone detection system for bone marrow reading further comprising a level composition unit that aligns the cell zone candidate cells from a high level to a low level of the leveled boundary line.
The method of claim 5,
A cell zone detection system for bone marrow reading, further comprising a confidence map generator configured to generate a cell zone confidence map based on the preset number information of the cell zone candidate cells.
The method of claim 5,
And a high magnification imaging unit configured to image the aligned cell zone candidate cells at high magnification from the high level to the low level order.
The method of claim 5,
And an alignment unit for relocating the aligned cell zone candidate cells in the down-scaled stitching image based on the original stitching image before the down-scaling.

Images