CN114897843A - Overlapping chromosome segmentation method - Google Patents

Abstract

The invention provides a method for segmenting overlapping chromosomes, which comprises the following steps of S100, carrying out edge extraction on an overlapping chromosome image, obtaining edge outlines of all overlapping chromosomes, and adding all edge outline points into a set of cutting points to be selected; step S200, extracting a skeleton, a skeleton end point and a skeleton intersection point of the overlapped chromosome image; step S300, determining the cutting points of the overlapped chromosomes according to the edge contour of the overlapped chromosomes obtained in the step S100 and the chromosome skeleton, the skeleton end points and the intersection points obtained in the step S200; and S400, automatically segmenting the overlapped chromosomes according to the cutting points. According to the contour and skeleton-based overlapped chromosome segmentation method, all contour points are regarded as cut points and used as candidate sets by combining the relation between the contour and the skeleton of the chromosome by utilizing the overlapped features of the overlapped chromosomes, a series of complex processes of curvature calculation are omitted, and the optimal cut points are accurately positioned by combining Euclidean distance with the skeleton of the chromosome.

Description

A method for segmenting overlapping chromosomes

technical field

The invention relates to the technical field of chromosome karyotype image analysis, in particular to a method for segmenting overlapping chromosomes.

Background technique

Chromosomes are the carriers of the genetic material of life. They exist in the nucleus in the form of filaments or rods. In the middle stage of cell mitosis, the number and morphological characteristics of chromosomes can be observed by staining and microscopic imaging. Chromosomal abnormalities can lead to genetic diseases, For diseases such as neonatal defects, karyotype analysis is to look for abnormal chromosome phenotype number and structure, such as number inconsistency, fragment abnormalities (duplication, deletion, translocation, inversion, etc.), morphological abnormalities (chromosomal length or satellite size and normal Chromosome inconsistency) and other abnormal conditions, doctors make accurate diagnosis and treatment of patients through chromosomal karyotype analysis.

Chromosomes in karyotype images are unpredictable, characterized by easy adhesion, blurred edges, and many overlapping and nesting. The existing karyotype image processing algorithms have poor effects on chromosome segmentation, counting, and redundancy elimination, and the accuracy needs to be improved. At the same time, the whole process of chromosome karyotype analysis requires manual participation in the whole process, the degree of automation is low, the work efficiency of chromosome karyotype analysis is low, labor-intensive, and the training cycle of professionals is long. The use of digital image processing technology solves problems such as chromosome counting, segmentation, and classification. , which can improve the analysis efficiency of staff, reduce labor intensity, and reduce the training cycle of professional talents.

SUMMARY OF THE INVENTION

In order to overcome the deficiencies in the background technology, the present invention discloses a method for segmenting overlapping chromosomes, which effectively improves the accuracy and efficiency of chromosome karyotype detection and analysis.

In order to realize the purpose of the invention, the present invention adopts the following technical solutions:

A method for segmentation of overlapping chromosomes, the segmentation method specifically comprises the following steps:

Step S100, perform edge extraction on the overlapping chromosome images, and obtain the edge contours of all overlapping chromosomes, and add all the edge contour points to the set of cutting points to be selected;

Step S200, extracting its skeleton, skeleton endpoints and skeleton intersections from the overlapping chromosome images;

Step S300, determining the cutting point of the overlapping chromosome according to the edge contour of the overlapping chromosome obtained in step S100 and the chromosome skeleton, skeleton endpoints, and intersection points obtained in step S200;

Step S400, the overlapping chromosomes are automatically segmented by an automatic cutting method according to the cutting point described in step S300.

The step S200 specifically includes the following steps:

Step S201, using a thinning algorithm to refine the overlapping chromosome image into a chromosome skeleton with only one pixel, and extract the overlapping chromosome skeleton;

Step S202, locate skeleton endpoints and intersections according to the number of adjacent pixels;

The step S300 specifically includes the following steps:

S301, according to the chromosome skeleton and the skeleton intersection described in step S200, taking the skeleton intersection as the center, calculate the Euclidean distance from the cutting point to be selected to the center point, and set this distance as, excluding the set of cutting points to be selected that is greater than all contour points at this distance;

S302, according to the calculation result of S301, re-divide the set points to be selected into 3 or 4 subsets according to the clustering algorithm, count the number of sets of cutting points to be selected, and calculate the distance in each subset from the center described in step S301 The Euclidean distance of the point, the candidate cutting point with the smallest distance is selected as the optimal cutting point of overlapping chromosomes, and it is included in the cutting point set.

S303, according to the calculation result of S302, determine whether the number of optimal cutting points is 4, if so, go to S304, otherwise go to S305;

S304, connect the four connection points two by two, there are three connection methods, and sequentially determine whether the connecting lines of the four cutting points cross, if not, cut the overlapping chromosome images respectively through the two connection methods in turn.

S305, calculate the curvatures of the three cutting points respectively, take two points with smaller curvatures as candidate cutting points A and B, and select the cutting point C with larger curvature to connect with the center point O described in S301 to form a line line segment OC;

S306, according to the line segment OC described in step S403, calculate the slope of the line segment OC and draw a parallel line L1 parallel to OC, L1 passes through the candidate cutting point A and intersects the chromosome outline at point D1, and another parallel line L2 is parallel to OC, L2 passes through The candidate cutting point B intersects the chromosome outline at point D2, and D1 and D2 are used as candidate cutting points.

S307, according to the method described in step S306, four candidate cutting points are acquired as A, B, D1, and D2, respectively, and the process returns to step S303.

S400, the overlapping chromosomes may be automatically segmented by an automatic cutting method according to the cutting point described in step S300.

Owing to adopting the above-mentioned technical scheme, the present invention has the following beneficial effects:

A method for dividing overlapping chromosomes according to the present invention includes step S100, performing edge extraction on overlapping chromosome images, obtaining edge contours of all overlapping chromosomes, and adding all the edge contour points to the set of cutting points to be selected; step S200, extract its skeleton, skeleton endpoints and skeleton intersections from the overlapping chromosome images; Step S300, determine the cutting points of the overlapping chromosomes according to the edge contour of the overlapping chromosomes obtained in step S100 and the chromosome skeleton, skeleton endpoints, and intersections obtained in step S200; step S400. Automatically segment the overlapping chromosomes according to the cutting points. The overlapping chromosome segmentation method based on contour and skeleton proposed by the present invention utilizes the overlapping features of overlapping chromosomes to combine the relationship between the contour and the skeleton of the chromosome, and regards all contour points as cutting points and as candidate sets, eliminating the need for a series of curvature calculations. It is a complex process that accurately locates the best cutting point through Euclidean distance combined with chromosome skeleton.

Description of drawings

Fig. 1 is a flowchart of a method for segmenting overlapping chromosome karyotype images based on contour and skeleton proposed by the present invention;

Fig. 2 is the edge detection effect diagram of the overlapping chromosome edge detection method used in the present invention;

3 is a flowchart of an effect diagram of “+”-shaped overlapping chromosome segmentation and a method for automatic chromosome segmentation according to Embodiment 1 of the present invention;

FIG. 4 is an effect diagram of segmentation of “T”-shaped overlapping chromosomes and a flowchart of an automatic chromosome segmentation method according to Embodiment 2 of the present invention.

Detailed ways

The present invention can be explained in detail through the following examples, and the purpose of disclosing the present invention is to protect all technical improvements within the scope of the present invention.

A kind of overlapping chromosome segmentation method described in conjunction with accompanying drawing 1～4, is the overlapping chromosome karyotype image segmentation method based on outline and skeleton, and described segmentation method specifically comprises the following steps:

Step S100, perform edge extraction on the overlapping chromosome images, and obtain the edge contours of all overlapping chromosomes, and add all the edge contour points to the set of cutting points to be selected;

Step S200, extracting its skeleton, skeleton endpoints and skeleton intersections from the overlapping chromosome images;

Step S300, determining the cutting point of the overlapping chromosome according to the edge contour of the overlapping chromosome obtained in step S100 and the chromosome skeleton, skeleton endpoints, and intersection points obtained in step S200;

Step S400, the overlapping chromosomes are automatically segmented by an automatic cutting method according to the cutting point described in step S300.

Regarding the detailed steps of the automatic segmentation method for overlapping chromosomes according to the present invention, the method will be described below through two specific embodiments.

Example 1:

Although the shapes of overlapping chromosomes are different, they basically belong to two types, including "+" shape and "T" shape or the superimposed shape of the two types, so the following steps will show the application of overlapping chromosomes based on "+" shape The automatic segmentation method of the present invention.

The overlapping chromosome segmentation algorithm proposed by the present invention is an image segmentation method of overlapping chromosome karyotypes based on contours and skeletons, and the segmentation method specifically includes the following steps:

Step 1: Before performing the edge contour extraction in step S100, the overlapping chromosome map needs to be preprocessed. First, the image is denoised by smoothing filtering to remove the influence of noise, and then the image is enhanced by histogram equalization to improve the chromosome image. Contrast with the background, then perform a binarization operation, use a threshold to separate the foreground and background of the karyotype image, remove most of the irrelevant regions, and obtain the preprocessed overlapping karyotype map, and then perform step S100. step, using the preprocessed overlapping chromosome karyotype map to perform edge contour extraction, and this step specifically includes the following specific steps:

Step 1.1, first use the Gaussian filtering algorithm to perform a convolution operation on the operator and the original image, and perform smooth filtering on the original image to obtain an output image with Gaussian noise removed.

Step 1.2, then perform a convolution operation on the output image of step 1.1 and the gradient operator, and obtain the gradient magnitude and gradient direction through calculation.

Step 1.3, the next step is to perform non-maximum suppression processing. Taking the pixel in the image as the origin, the bidirectional dividing line passing through the origin is at 45 degrees and 135 degrees, respectively. The pixel value comparison of the pixel, if the pixel value of the origin is larger, it is retained; otherwise, it is returned to zero.

Step 1.4: Finally, double-threshold hysteresis threshold processing is performed, and the high and low (Th, Tl) threshold calculation factors are set, and the points higher than Th are marked as strong edge points, the points less than Tl are set to 0, and the pixels between the high and low thresholds are set to 0. The point that is not connected to the strong edge point is set to 0, and finally the edge contour of the overlapping chromosome is obtained, and all the contour points are recorded in the candidate cutting point set P. Figure 1 shows the effect of the edge detection of the chromosome karyotype map.

In step 2, the following method is used to extract the chromosome skeleton according to the preprocessed overlapping chromosome karyotype image in step 1. The specific steps of this method include: firstly calculate step by step according to the thinning algorithm, and refine the chromosome image into the chromosome skeleton, and then according to the pixel point The connection relationship is located at the skeleton end point of the chromosome, according to the division standard of no pixel value on the three sides of the end point, and recorded in the end point set, and then according to the pixel point discrimination method, the position of the skeleton intersection point is extracted and recorded in the skeleton in the intersection set.

Step 3, combine the edge contour of the chromosome binary image and the skeleton intersection to screen the segmentation points, and obtain accurate overlapping chromosome segmentation points through screening. The specific steps include:

Step 3.1, according to the chromosome skeleton and the skeleton intersection described in Step 1 and Step 2, take the skeleton intersection as the center, calculate the Euclidean distance from the cutting point to be selected to the center point, set this distance, and remove the cutting point to be selected All contour points in the set that are larger than this distance, obtain the selected cutting points after screening, as shown in the fourth picture in Figure 2, which is a schematic diagram of the selected cutting points obtained after dividing according to step 3.1.

Step 3.2, according to the calculation result of step 3.1, re-divide the set points to be selected into 4 subsets according to the clustering algorithm, count the number of sets of cutting points to be selected, and calculate the distance in each subset from the center point described in step S301. The Euclidean distance of , select the candidate cutting point with the smallest distance and determine it as the optimal cutting point of overlapping chromosomes, which is included in the cutting point set, as shown in the fifth picture in Figure 2, which is the optimal cutting point obtained according to step 3.2 Schematic.

Step 3.3, determine whether there are 4 optimal cutting points, if there are 4 optimal cutting points to be selected, go to step 3.4.

Step 3.4, connect the 4 connection points two by two, there are 3 connection methods, and then judge whether the connection lines of the 4 cutting points cross, if not, then cut the overlapping chromosome images respectively through the two connection methods.

Step 4, according to the optimal cutting point described in Step 3.3, the overlapping chromosomes are automatically segmented by an automatic cutting method.

Example 2:

Below by showing the embodiment 2 based on "T" type overlapping chromosomes, in the process of applying the automatic segmentation method for chromosome karyotype images described in the present invention, the specific steps include the following:

In step 1, the overlapping chromosome map needs to be preprocessed before the edge contour extraction in step S100 is performed. First, the image is denoised by smoothing filtering to remove the influence of noise, and then the image is enhanced by histogram equalization to improve the contrast between the chromosome image and the background. Separating from the background, removing most of the irrelevant regions, and obtaining the preprocessed overlapping chromosome karyotype map, and then performing the specific steps of step S100, using the preprocessed overlapping chromosome karyotype map to perform edge contour extraction, and this step specifically includes The following specific steps:

Step 1.1, first use the Gaussian filtering algorithm to perform a convolution operation on the operator and the original image, and perform smooth filtering on the original image to obtain an output image with Gaussian noise removed.

Step 1.2, then perform a convolution operation on the output image of step 1.1 and the gradient operator, and obtain the gradient magnitude and gradient direction through calculation.

Step 1.3, the next step is to perform non-maximum suppression processing. Taking the pixel in the image as the origin, the bidirectional dividing line passing through the origin is at 45 degrees and 135 degrees, respectively. The pixel value comparison of the pixel, if the pixel value of the origin is larger, it is retained; otherwise, it is returned to zero.

Step 1.4: Finally, double-threshold hysteresis threshold processing is performed, and the high and low (Th, Tl) threshold calculation factors are set, and the points higher than Th are marked as strong edge points, the points less than Tl are set to 0, and the pixels between the high and low thresholds are set to 0. The point that is not connected to the strong edge point is set to 0, and finally the edge contour of the overlapping chromosome is obtained, and all the contour points are recorded in the candidate cutting point set P. Figure 1 shows the effect of the edge detection of the chromosome karyotype map.

In step 2, the following method is used to extract the chromosome skeleton according to the preprocessed overlapping chromosome karyotype image in step 1. The specific steps of this method include: firstly calculate step by step according to the thinning algorithm, and refine the chromosome image into the chromosome skeleton, and then according to the pixel point The connection relationship is located at the skeleton end point of the chromosome, according to the division standard of no pixel value on the three sides of the end point, and recorded in the end point set, and then according to the pixel point discrimination method, the position of the skeleton intersection point is extracted and recorded in the skeleton in the intersection set.

Step 3, combine the edge contour of the chromosome binary image and the skeleton intersection to screen the segmentation points, and obtain accurate overlapping chromosome segmentation points through screening. The specific steps include:

Step 3.1, according to the chromosome skeleton and the skeleton intersection described in Step 1 and Step 2, take the skeleton intersection as the center, calculate the Euclidean distance from the cutting point to be selected to the center point, set this distance, and remove the cutting point to be selected All the contour points in the set that are greater than this distance get the selected cutting points after screening, as shown in the fourth picture in Figure 3, which is a schematic diagram of the selected cutting points obtained after screening according to step 3.1.

Step 3.2, according to the calculation result of step 3.1, re-divide the set points to be selected into 3 subsets according to the clustering algorithm, count the number of sets of cutting points to be selected, and calculate the distance in each subset from the center point described in step S301. The Euclidean distance, select the candidate cutting point with the smallest distance to determine the optimal cutting point of overlapping chromosomes, and be included in the cutting point set. Schematic.

Step 3.3, determine whether there are 4 optimal cutting points, if the number of optimal cutting points to be selected is 4, go to step 3.4. According to Embodiment 2, only 3 optimal cutting points can be obtained in step 3.2, so perform step 3.2. 3.5.

Step 3.4, connect the 4 connection points in pairs, there are 3 connection methods, and then judge whether the connection lines of the 4 cutting points intersect, if not, cut the overlapping chromosome images in turn through these 2 connection methods, and execute Step 4.

Step 3.5: Calculate the curvature of the three cutting points respectively, take the two points with smaller curvature as candidate cutting points A and B, and select the cutting point C with larger curvature to connect with the center point O obtained in step 2 as A line segment OC;

Step 3.6, according to the line segment OC obtained in step 3.5, calculate the slope of the line segment OC and make a parallel line L1 parallel to OC, L1 passes through the candidate cutting point A and intersects the chromosome outline at point D1, and another parallel line L2 is parallel to OC, L2 After the candidate cutting point B intersects the chromosome outline at point D2, take D1 and D2 as candidate cutting points, and return to step 3.4.

In step 4, according to the optimal cutting point connection method described in step 3.4, the overlapping chromosomes are automatically segmented by an automatic cutting method.

The part that is not described in detail in the present invention is the prior art. Although the present invention is specifically shown and introduced in conjunction with the preferred embodiments, there are many methods and approaches to realize the technical solution. The above are only the preferred embodiments of the present invention, but the It should be understood by those skilled in the art that various changes can be made to the present invention in form and detail without departing from the spirit and scope of the present invention defined by the appended claims, which are all within the protection scope of the present invention.

Claims (7)

1. A method for segmentation of overlapping chromosome karyotypes, characterized in that automatic segmentation is performed on overlapping chromosomes, said segmentation method comprising the steps of:

step S100, carrying out edge extraction on the overlapped chromosome image, obtaining edge outlines of all overlapped chromosomes, and adding all edge outline points into a set of cutting points to be selected;

step S200, extracting a skeleton, a skeleton end point and a skeleton intersection point of the overlapped chromosome image;

step S300, determining the cutting points of the overlapped chromosomes according to the edge contour of the overlapped chromosomes obtained in the step S100 and the chromosome skeleton, the skeleton end points and the intersection points obtained in the step S200;

and S400, automatically segmenting the overlapped chromosomes through an automatic segmentation method according to the segmentation points in the step S300.

2. The method of overlapping chromosome segmentation as set forth in claim 1, wherein: the method used in the overlapping chromosome edge contour extraction method in step S100 is a contour extraction method based on the Canny operator.

3. The method of overlapping chromosome segmentation as set forth in claim 1, wherein: the above-mentioned

The method comprises the following steps of extracting a skeleton, skeleton endpoints and skeleton intersections of overlapped chromosome images, wherein the step S200 specifically comprises the following steps:

step S201, a thinning algorithm is adopted to thin the overlapped chromosome image into a chromosome skeleton with only one pixel point, and the overlapped chromosome skeleton is extracted;

step S202, positioning the skeleton endpoint, namely the number of pixels in the neighborhood is 1 according to the number of the refined adjacent pixels;

step S203, positioning the framework cross point according to the number of the neighborhood pixel points.

4. The method of overlapping chromosome segmentation as set forth in claim 1, wherein: the automatic segmentation method of the overlapped chromosomes is in a shape of a plus sign and a T or the superposition of the two types.

5. The method of overlapping chromosome segmentation according to claim 4, wherein: the method for determining the overlapped chromosome cutting point by using the overlapped chromosome edge contour method obtained in the step S100 and the chromosome skeleton, the skeleton end points and the intersection points obtained in the step S200 specifically includes the following steps:

s301, calculating the Euclidean distance from the cutting point to be selected to the central point by taking the skeleton intersection point as the center according to the chromosome skeleton and the skeleton intersection point in the step S200, setting the distance, and removing all contour points which are greater than the distance in the cutting point set to be selected;

s302, according to the calculation result of S301, subdividing the to-be-selected set points into 3 or 4 subsets according to a clustering algorithm, counting the set number of the to-be-selected cut points, calculating Euclidean distances between the to-be-selected cut points and the central point in the step S301 in each subset, selecting the to-be-selected cut point with the minimum distance, determining the to-be-selected cut point as the optimal cut point of the overlapped chromosomes, and counting the to-be-selected cut point into a cut point set;

s303, judging whether the number of the optimal cutting points is 4 or not according to the calculation result of the S302, if so, carrying out S304, and if not, carrying out S305;

s304, connecting the 4 connecting points pairwise, wherein 3 connecting modes are provided, sequentially judging whether the connecting lines of the 4 cutting points are crossed, and if not, respectively cutting the overlapped chromosome images sequentially through the 2 connecting modes;

s305, respectively calculating curvatures of the 3 cutting points, taking two points with smaller curvatures as candidate cutting points A, B, and selecting a cutting point C with larger curvature to be connected with the central point O in the S301 to form a line segment OC;

s306, according to the line segment OC in the step S403, calculating the slope of the line segment OC, drawing a parallel line L1 parallel to the OC, making L1 pass through a candidate cut point A to intersect with the chromosome contour at a point D1, drawing a parallel line L2 parallel to the OC, making L2 pass through a candidate cut point B to intersect with the chromosome contour at a point D2, and taking D1 and D2 as candidate cut points;

s307, according to the method in the step S306, acquiring that the 4 candidate cutting points are A, B, D1 and D2 respectively, and returning to the step S303;

s400, automatically segmenting the overlapped chromosomes according to the segmentation points in the step S300 by an automatic segmentation method.

6. The method of overlapping chromosome segmentation as set forth in claim 5, wherein: the automatic segmentation method based on the application of the '+' font overlapped chromosome, namely the overlapped chromosome karyotype image segmentation method based on the contour and the skeleton, specifically comprises the following steps:

step 1, before performing the edge contour extraction in step S100, preprocessing an overlapping chromosome map, first performing noise reduction on an image by smooth filtering to remove noise influence, then performing image enhancement on the image by histogram equalization to improve the contrast between a chromosome image and a background, next performing binarization operation, separating a foreground of a chromosome karyotype image from the background by using a threshold value, removing most of irrelevant areas to obtain a preprocessed overlapping chromosome karyotype map, next performing the specific steps in step S100, and performing the edge contour extraction by using the preprocessed overlapping chromosome karyotype map, wherein the specific steps specifically include the following specific steps:

step 1.1, firstly, carrying out convolution operation on an operator and an original image by adopting a Gaussian filtering algorithm, and carrying out smooth filtering on the original image to obtain an output image with Gaussian noise eliminated;

step 1.2, performing convolution operation on the output image in the step 1.1 and a gradient operator, and obtaining a gradient amplitude and a gradient direction through calculation;

step 1.3, then carrying out non-maximum value suppression treatment, taking pixels in the image as an origin, respectively making bidirectional dividing lines passing through the origin at 45 degrees and 135 degrees, comparing the pixel value of the origin with pixel values of two adjacent pixels along the dividing lines, and if the pixel value of the origin is larger, keeping the original; otherwise, returning to zero;

step 1.4, finally, carrying out double-threshold hysteresis threshold processing, setting high and low (Th and Tl) threshold calculation factors, marking the points higher than Th as strong edge points, setting the points smaller than Tl as 0, setting the points between the high and low thresholds and the points which are not connected with the strong edge points as 0, finally obtaining overlapped chromosomes to obtain edge contours, and recording all contour points into a candidate cut point set P;

step 2, extracting a chromosome skeleton according to the overlapping chromosome karyotype image preprocessed in the step 1.1-1.4 by using the following method, wherein the method specifically comprises the following steps: firstly, gradually calculating according to a thinning algorithm, thinning a chromosome image into a chromosome skeleton, then positioning skeleton endpoints of the chromosome according to the connection relation of pixel points, recording in an endpoint set according to a division standard that no pixel value exists on three sides of the endpoints, then extracting the positions of skeleton intersections according to a pixel point discrimination method, and recording in an intersection set of the skeleton;

and 3, screening the segmentation points by combining the chromosome binary image edge contour and the skeleton intersection point, and obtaining accurate overlapped chromosome segmentation points through screening, wherein the method specifically comprises the following steps:

step 3.1, calculating the Euclidean distance from the cutting point to be selected to the central point by taking the skeleton intersection point as the center according to the chromosome skeleton and the skeleton intersection point in the steps 1.1-1.4 and 2, setting the Euclidean distance for the distance, removing all contour points which are greater than the distance in the cutting point set to be selected, and obtaining the screened cutting point to be selected;

step 3.2, according to the calculation result of the step 3.1, the set points to be selected are divided into 4 subsets again according to a clustering algorithm, the set number of the cutting points to be selected is counted, the Euclidean distance from the central point in the step S301 in each subset is calculated, the cutting point to be selected with the minimum distance is selected to be determined as the optimal cutting point of the overlapped chromosomes, and the optimal cutting point is counted into a cutting point set;

step 3.3, judging whether the optimal cutting points are 4, and if the optimal cutting points to be selected are 4, executing the step 3.4;

step 3.4, connecting the 4 connecting points pairwise, wherein 3 connecting modes are provided, sequentially judging whether the connecting lines of the 4 cutting points are crossed, and if not, respectively cutting the overlapped chromosome images sequentially through the 2 connecting modes;

and 4, automatically segmenting the overlapped chromosomes by an automatic segmentation method according to the optimal segmentation points in the step 3.3.

7. The method of overlapping chromosome segmentation as set forth in claim 5, wherein: the application based on the T-shaped overlapping chromosome in the automatic segmentation method process of the chromosome karyotype image comprises the following specific steps:

step 1, before performing the edge contour extraction in step S100, preprocessing an overlapping chromosome map, first performing noise reduction on an image by smooth filtering to remove noise influence, then performing image enhancement on the image by histogram equalization to improve the contrast between a chromosome image and a background, next performing binarization operation, separating a foreground of a chromosome karyotype image from the background by using a threshold value, removing most of irrelevant areas to obtain a preprocessed overlapping chromosome karyotype map, next performing the specific steps in step S100, and performing the edge contour extraction by using the preprocessed overlapping chromosome karyotype map, wherein the specific steps specifically include the following specific steps:

step 1.1, firstly, carrying out convolution operation on an operator and an original image by adopting a Gaussian filtering algorithm, and carrying out smooth filtering on the original image to obtain an output image with Gaussian noise eliminated;

step 1.2, performing convolution operation on the output image in the step 1.1 and a gradient operator, and obtaining a gradient amplitude and a gradient direction through calculation;

step 1.3, then carrying out non-maximum value suppression treatment, taking pixels in the image as an origin, respectively making bidirectional dividing lines passing through the origin at 45 degrees and 135 degrees, comparing the pixel value of the origin with pixel values of two adjacent pixels along the dividing lines, and if the pixel value of the origin is larger, keeping the original; otherwise, returning to zero;

step 1.4, finally, carrying out double-threshold hysteresis threshold processing, setting high and low (Th and Tl) threshold calculation factors, marking the points higher than Th as strong edge points, setting the points smaller than Tl as 0, setting the points between the high and low thresholds and the points which are not connected with the strong edge points as 0, finally obtaining overlapped chromosomes to obtain edge contours, and recording all contour points into a candidate cut point set P;

step 2, extracting a chromosome skeleton according to the overlapping chromosome karyotype image preprocessed in the step 1.1-1.4 by using the following method, wherein the method specifically comprises the following steps: firstly, gradually calculating according to a thinning algorithm, thinning a chromosome image into a chromosome skeleton, then positioning skeleton endpoints of the chromosome according to the connection relation of pixel points, recording in an endpoint set according to a division standard that no pixel value exists on three sides of the endpoints, then extracting the positions of skeleton intersections according to a pixel point discrimination method, and recording in an intersection set of the skeleton;

and 3, screening the segmentation points by combining the chromosome binary image edge contour and the skeleton intersection point, and obtaining accurate overlapped chromosome segmentation points through screening, wherein the method specifically comprises the following steps:

step 3.1, calculating the Euclidean distance from the cutting point to be selected to the central point by taking the skeleton intersection point as the center according to the chromosome skeleton and the skeleton intersection point in the steps 1.1-1.4 and 2, setting the Euclidean distance for the distance, removing all contour points which are greater than the distance in the cutting point set to be selected, and obtaining the screened cutting point to be selected;

step 3.2, according to the calculation result of the step 3.1, the set points to be selected are divided into 3 subsets again according to a clustering algorithm, the set number of the cutting points to be selected is counted, the Euclidean distance between the set points to be selected and the central point in the step S301 in each subset is calculated, the cutting point to be selected with the minimum distance is selected to be determined as the optimal cutting point of the overlapped chromosomes, and the optimal cutting point is counted into a cutting point set;

step 3.3, judging whether the optimal cutting points are 4, if the optimal cutting points to be selected are 4, executing step 3.4, and according to the embodiment 2, only 3 optimal cutting points can be obtained in the step 3.2, so that the step 3.5 is executed;

step 3.4, connecting the 4 connecting points pairwise, wherein 3 connecting modes are provided, sequentially judging whether the connecting lines of the 4 cutting points are crossed, if not, respectively cutting the overlapped chromosome images sequentially through the 2 connecting modes, and executing the step 4; step 3.5, respectively calculating the curvatures of the 3 cutting points, taking two points with smaller curvatures as candidate cutting points A, B, and selecting a cutting point C with larger curvature to be connected with the central point O obtained in the step 2 to form a line segment OC;

step 3.6, according to the line segment OC obtained in step 3.5, calculating the slope of the line segment OC, drawing a parallel line L1 parallel to OC, wherein L1 crosses the chromosome contour through the candidate cut point a at a point D1, drawing a parallel line L2 parallel to OC, wherein L2 crosses the chromosome contour through the candidate cut point B at a point D2, taking D1 and D2 as candidate cut points, and returning to step 3.4;

and 4, automatically segmenting the overlapped chromosomes by an automatic segmentation method according to the optimal cutting point connection method in the step 3.4.

Images