Chromosome metaphase division phase image sorting method

Abstract

The invention discloses a chromosome metaphase division phase image sorting method, which belongs to the field of image processing, and comprises the following steps of: segmenting crossed and adhered chromosomes in chromosome metaphase division phase images; extracting a plurality of features from the segmented chromosome image; setting a reasonable interval of each feature, when each feature of each image is in the corresponding interval, putting the image into a set S1, when a certain feature of each image is not in the interval, putting the image into a set S2, arranging the images in the set S1 according to the number of connected domains from large to small, arranging the images in the set S2 according to the number of chromosomes from large to small, and setting the images in the set S2 according to the number of chromosomes from large to small; the length of the set S1 is added to the serial numbers of all the images in the set S2, then the two sets are combined, overall sequencing is completed, through the method, the images meeting the requirement for karyotype analysis can be automatically screened out, and manual operation is not needed; according to the sorting method, the processing speed is high, the time required for karyotype analysis is effectively shortened, and the efficiency is improved.

Description

Technical Field

[0001] The invention relates to the field of image processing, in particular to a method for sorting chromosome metaphase images. Background Art

[0002] Chromosome karyotype analysis is the main way to diagnose chromosome translocation, deletion, and various genetic diseases. The preparation of chromosome specimens for karyotype analysis involves multiple steps, such as slide washing, cell culture, colchicine treatment, fixation, Giemsa staining, and sealing. The specimen is then placed in a chromosome analyzer to collect dozens to hundreds of images of chromosome metaphase divisions of different cells. Physicians select images with clear bands, moderate length, complete shape, and less cross-adhesion and noise interference for karyotype analysis tasks.

[0003] Since a large number of metaphase chromosome images are collected and their quality varies, manual image selection is time-consuming and laborious, and the efficiency and accuracy depend on the physician's proficiency. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, one of the objects of the present invention is to provide a method for sorting chromosome metaphase images, which first pre-segment the chromosomes, and then sort the collected chromosome images in order of quality from high to low, so that doctors can quickly find chromosome images suitable for karyotype analysis and improve efficiency.

[0005] One of the purposes of the present invention is achieved by the following technical solution:

[0006] A method for sorting chromosome metaphase images, comprising the following steps:

[0007] S1: Segmentation of crossed and adhered chromosomes in the metaphase image;

[0008] S2: extract multiple features from the segmented chromosome image;

[0009] S3: Set a reasonable interval for each feature. When each feature of each image is within the corresponding interval, the image is considered to be a high-quality image and placed in set S1. When a feature of each image is not within the interval, the image is placed in set S2. The images in the two sets are sorted separately, where the images in set S1 are arranged in descending order according to the number of connected domains, and the images in set S2 are arranged in descending order according to the number of chromosomes. The serial numbers of all images in set S2 are added to the length of set S1, and then the two sets are merged to complete the overall sorting.

[0010] Furthermore, step S1 specifically includes the following steps:

[0011] S11: Establish a chromosome database, including the original chromosome metaphase images and manual annotation information;

[0012] S12: Divide the chromosome database into two subsets: training and validation;

[0013] S13: Expand the number of images in the training set;

[0014] S14: Calculate the mean and variance of the images in the training set, and normalize the images in the training set and the test set according to the mean and variance;

[0015] S15: train the instance segmentation algorithm using the training set;

[0016] S16: Use non-maximum suppression algorithm to optimize the segmentation results;

[0017] S17: Save the algorithm parameters and use them for segmentation of chromosome metaphase images.

[0018] Furthermore, in step S13, at least one of image rotation, upside-down flipping, and left-right flipping is used to expand the number of images in the training set.

[0019] Furthermore, in step S15, the segmentation algorithm is Mask-RCNN.

[0020] Furthermore, in step S15, when the training set is used to train the instance segmentation algorithm, the number of training times is 30, the initial learning rate is 0.0001, and the learning rate is reduced by 10 times every 10 training times.

[0021] Furthermore, in step S15, when the instance segmentation algorithm is trained using the training set, a stochastic gradient descent algorithm is used for optimization.

[0022] Furthermore, in step S16, when the non-maximum suppression algorithm is used to optimize the segmentation result, the image region of interest threshold is 0.7 and the image segmentation threshold is 0.5.

[0023] Further, in step S2, the multiple features are at least two of the number of chromosomes, the number of connected domains, the grayscale distribution, the chromosome area at the border, the distance between the chromosome distribution center and the image center, and the maximum area of ​​the chromosome.

[0024] Furthermore, in step S2, the number of chromosomes is specifically calculated by counting the number of chromosomes segmented from each metaphase image in the same case, calculating the mode, subtracting the mode from the number of chromosomes in each image, and calculating the absolute value.

[0025] Furthermore, in step S2, the number of connected domains is specifically as follows: threshold segmentation is performed on the metaphase image to generate a binary image, connected domains with an area smaller than a preset value are regarded as noise and deleted from the binary image, the current number of connected domains is calculated and the number of noises segmented out is subtracted.

[0026] Furthermore, in step S2, the distance between the chromosome distribution center and the image center is specifically as follows: the geometric center of each chromosome in the image is calculated according to the segmentation result, and then the average value of these geometric centers is calculated, which is regarded as the distribution center of the chromosome, and the distance between the distribution center and the image center is calculated.

[0027] Compared with the prior art, the chromosome metaphase image sorting method of the present invention segments the crossed and adhered chromosomes in the chromosome metaphase image; extracts multiple features from the segmented chromosome image; sets a reasonable interval for each feature, and when each feature of each image is within the corresponding interval, the image is regarded as a high-quality image and placed in set S1; when a certain feature of each image is not within the interval, the image is placed in set S2, and the images in the two sets are sorted respectively, wherein the images in set S1 are arranged in order from large to small according to the number of connected domains, and the images in set S2 are arranged in order from large to small according to the number of chromosomes, and the serial numbers of all images in set S2 are added to the length of set S1, and then the two sets are merged to complete the overall sorting. Through the above method, images that meet the requirements of karyotype analysis can be automatically screened out without manual operation and are not easily affected by the subjective factors of doctors; the sorting method of the present application has a fast processing speed, which can effectively reduce the time required for karyotype analysis and improve efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The present invention is a flowchart of the method for sorting chromosome metaphase images. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be noted that when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may be another intermediate component through which it is fixed. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may be another intermediate component at the same time. When a component is considered to be "set on" another component, it may be directly set on the other component or there may be another intermediate component at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0032] Figure 1 The flowchart of the method for sorting chromosome metaphase images of the present invention is as follows:

[0033] S1: Segmentation of crossed and adhered chromosomes in the metaphase image;

[0034] S2: extract multiple features from the segmented chromosome image;

[0035] S3: Set a reasonable interval for each feature. When each feature of each image is within the corresponding interval, the image is considered to be a high-quality image and placed in set S1. When a feature of each image is not within the interval, the image is placed in set S2. The images in the two sets are sorted separately, where the images in set S1 are arranged in descending order according to the number of connected domains, and the images in set S2 are arranged in descending order according to the number of chromosomes. The serial numbers of all images in set S2 are added to the length of set S1, and then the two sets are merged to complete the overall sorting.

[0036] The segmentation method in step S1 is a chromosome instance segmentation method based on the deep learning algorithm Mask-RCNN, which is used to segment crossover and adhesion chromosomes and can output the contours and categories of chromosomes and noise. Specifically, it includes the following steps:

[0037] S11: Establish a chromosome database, including the original chromosome metaphase images and manual annotation information;

[0038] S12: Divide the chromosome database into two subsets: training and validation;

[0039] S13: Expand the number of images in the training set;

[0040] S14: Calculate the mean and variance of the images in the training set, and normalize the images in the training set and the test set according to the mean and variance;

[0041] S15: train the instance segmentation algorithm using the training set;

[0042] S16: Use non-maximum suppression algorithm to optimize the segmentation results;

[0043] S17: Save the algorithm parameters and use them for segmentation of chromosome metaphase images.

[0044] In step S13, at least one of the following operations is used to expand the number of images in the training set: image rotation, upside down flipping, and left-right flipping. In step S15, the segmentation algorithm is Mask-RCNN. When the training set is used to train the instance segmentation algorithm, the number of training times is 30, the initial learning rate is 0.0001, and the learning rate is reduced by 10 times every 10 training times. When the training set is used to train the instance segmentation algorithm, the stochastic gradient descent algorithm is used for optimization. In step S16, when the non-maximum suppression algorithm is used to optimize the segmentation result, the image region of interest threshold is 0.7 and the image segmentation threshold is 0.5.

[0045] Extracting multiple features of the chromosome image in step S2 refers to quantifying the quality of the chromosome image according to clinical specifications for computer processing. Specifically, the multiple features are the number of chromosomes, the number of connected domains, the grayscale distribution, the chromosome area at the border, the distance between the chromosome distribution center and the image center, and the maximum area of ​​the chromosome.

[0046] The specific method for determining the number of chromosomes is as follows: the number of chromosomes segmented from each metaphase image in the same case is counted, the mode is calculated, the number of chromosomes in each image is subtracted from the mode, and the absolute value is calculated. If the absolute value is 0, it means that there is no abnormal number in the image.

[0047] The number of connected domains is as follows: threshold segmentation is performed on the metaphase image to generate a binary image. Connected domains with an area smaller than the preset value are considered noise and deleted from the binary image. The current number of connected domains is calculated and the number of noises segmented is subtracted. The larger this feature is, the less adhesion, crossover, and overlapping staining there is, and the easier it is to process.

[0048] Grayscale distribution is as follows: extract the chromosome pixels at the corresponding position according to the segmentation results, and count the grayscale distribution of the pixel values. If the grayscale distribution of a chromosome is uneven and too concentrated near a certain pixel value, the clarity of the chromosome is crossed and it is difficult to observe the chromosome banding.

[0049] The specific chromosome area at the border is: determine whether there are segmented chromosomes at the image border and count the area. The width of the border is 10. This feature is used to determine whether the overall distribution of chromosomes is too dispersed, causing some chromosomes to fly out of the image.

[0050] The distance between the chromosome distribution center and the image center is as follows: the geometric center of each chromosome in the image is calculated based on the segmentation results, and then the average of these geometric centers is calculated, which is regarded as the chromosome distribution center, and the distance between the distribution center and the image center is calculated. If the distance is too large, it means that the chromosome distribution is offset, and the image may be missing.

[0051] The maximum area of ​​the chromosome is specifically: calculate the area of ​​each chromosome and select the maximum value as the feature. If the area is small, it means that the length of the chromosome in the image is too short and it is difficult to observe the banding.

[0052] In step S3, due to factors such as the chromosome specimen preparation technique and the accuracy of the collection equipment, the reasonable interval boundary should be adjusted according to the actual situation rather than using a fixed interval.

[0053] The chromosome metaphase image sorting method of the present invention can automatically screen out images that meet the requirements of karyotype analysis without manual operation and is not easily affected by the subjective factors of the physician; the sorting method of the present application has a fast processing speed and can effectively reduce the time required for karyotype analysis and improve efficiency.

[0054] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which are equivalent modifications and improvements made to the above embodiments based on the essential technology of the present invention, and all of them belong to the protection scope of the present invention.

Claims

1. A method for sorting chromosome metaphase images, characterized in that: The following steps are involved: S1: Segmentation of crossed and adhered chromosomes in the metaphase image; S2: extract multiple features from the segmented chromosome image; S3: Set a reasonable interval for each feature. When each feature of each image is within the corresponding interval, the image is considered to be a high-quality image and placed in set S1. When a feature of each image is not within the interval, the image is placed in set S2. The images in the two sets are sorted separately, where the images in set S1 are arranged in descending order according to the number of connected domains, and the images in set S2 are arranged in descending order according to the number of chromosomes. The serial numbers of all images in set S2 are added to the length of set S1, and then the two sets are merged to complete the overall sorting.

2. The method for sorting chromosome metaphase images according to claim 1, characterized in that: Step S1 specifically includes the following steps: S11: Establish a chromosome database, including the original chromosome metaphase images and manual annotation information; S12: Divide the chromosome database into two subsets: training and validation; S13: Expand the number of images in the training set; S14: Calculate the mean and variance of the images in the training set, and normalize the images in the training set and the test set according to the mean and variance; S15: train the instance segmentation algorithm using the training set; S16: Use non-maximum suppression algorithm to optimize the segmentation results; S17: Save the algorithm parameters and use them for segmentation of chromosome metaphase images.

3. The method for sorting chromosome metaphase images according to claim 2, characterized in that: In step S13, at least one of image rotation, upside-down flipping, and left-right flipping is used to expand the number of images in the training set.

4. The method for sorting chromosome metaphase images according to claim 2, characterized in that: In step S15, the segmentation algorithm is Mask-RCNN.

5. The method for sorting chromosome metaphase images according to claim 4, characterized in that: In step S15, when the training set trains the instance segmentation algorithm, the number of training times is 30, the initial learning rate is 0.0001, and the learning rate is reduced by 10 times every 10 training times.

6. The method for sorting chromosome metaphase images according to claim 4, characterized in that: In step S15, when the training set is used to train the instance segmentation algorithm, a stochastic gradient descent algorithm is used for optimization.

7. The method for sorting chromosome metaphase images according to claim 2, characterized in that: In step S16, when the non-maximum suppression algorithm is used to optimize the segmentation result, the image region of interest threshold is 0.7 and the image segmentation threshold is 0.

5.

8. The method for sorting chromosome metaphase images according to claim 1, characterized in that: In step S2, the multiple features are at least two of the number of chromosomes, the number of connected domains, the grayscale distribution, the chromosome area at the border, the distance between the chromosome distribution center and the image center, and the maximum area of ​​the chromosome.

9. The method for sorting chromosome metaphase images according to claim 8, characterized in that: In step S2, the number of chromosomes is specifically calculated by counting the number of chromosomes segmented from each metaphase image in the same case, calculating the mode, subtracting the mode from the number of chromosomes in each image, and calculating the absolute value.

10. The method for sorting chromosome metaphase images according to claim 8, characterized in that: In step S2, the number of connected domains is specifically as follows: threshold segmentation is performed on the metaphase image to generate a binary image, connected domains with an area smaller than a preset value are regarded as noise and deleted from the binary image, the current number of connected domains is calculated and the number of noises segmented is subtracted.

11. The method for sorting chromosome metaphase images according to claim 8, characterized in that: In step S2, the distance between the chromosome distribution center and the image center is specifically as follows: the geometric center of each chromosome in the image is calculated according to the segmentation result, and then the average value of these geometric centers is calculated, which is regarded as the chromosome distribution center, and the distance between the distribution center and the image center is calculated.

Images