Chromosome karyotype structure anomaly detection method and system based on banding coding

Through band encoding technology, chromosomal karyotype encoding information is generated and matched with the database, which solves the problems of low automation and insufficient fine-grainedness of traditional detection methods, and realizes efficient detection of chromosomal structural abnormalities, which is suitable for the fields of medical genetics and oncology.

CN118968152AActive Publication Date: 2024-11-15SHANGHAI JIAOTONG UNIV

Patent Information

Application Number
CN202411013808.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-15
Estimated Expiration
2044-07-26

AI Technical Summary

Technical Problem

Traditional chromosome structural abnormality detection methods have low degree of automation, limited detection accuracy and efficiency, and existing machine learning-based methods have insufficient detection fine-grainedness and stability, so they cannot effectively identify tiny structural abnormalities of chromosomes.

Method used

Using the detection method of chromosomal karyotype structural abnormality based on band encoding, the chromosomal karyotype encoding information is generated and matched with the database, and the tiny structural abnormalities of chromosomes are automatically identified, improving the detection accuracy and fine-grainedness, which is suitable for the fields of medical genetics and oncology. .

Benefits of technology

It significantly improves the accuracy, sensitivity and fine-grained detection of chromosomal karyotype structural abnormalities, can automatically and efficiently identify tiny structural abnormalities, and is suitable for the research and diagnosis of chromosomal structural abnormalities in fields such as medical genetics and oncology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968152A_ABST
    Figure CN118968152A_ABST
Patent Text Reader

Abstract

The invention provides a chromosome karyotype structure anomaly detection method and system based on banding coding, and the method comprises the steps: generating coding information according to the banding gray scale of a chromosome monomer image collected through microscopic imaging, and enabling the codes of all monomers in a karyotype to form a chromosome karyotype coding subset; extracting a skeleton of the dyeing monomer, inputting skeleton information and the dyeing monomer image into a Transform coding network, and generating coding information of the chromosome monomer; and matching the obtained chromosome karyotype coding set with a chromosome karyotype database to obtain an abnormal type, and if no matching result exists, adding the karyotype information to the database. The method greatly improves the sensitivity, precision and fine granularity of chromosome structure karyotype anomaly detection, and is suitable for chromosome structure anomaly research and diagnosis in the fields of medical genetics, oncology and the like.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of biological genetic chromosome information recognition, and in particular to a method and system for detecting chromosome karyotype structural abnormalities based on banding coding. Background Art

[0002] Chromosome structural abnormality refers to changes in the physical structure of chromosomes, which leads to abnormal gene quantity or arrangement, affecting the development, health and reproductive capacity of individuals. Common abnormalities include deletion, duplication, inversion, translocation, ring chromosome and insertion, etc. Chromosome structural abnormality has important clinical significance in genetic counseling, disease diagnosis and prognosis assessment. Through the application of modern genetic technology, it provides an important basis for understanding and treating related diseases.

[0003] Traditional methods for detecting chromosome structural abnormalities mostly rely on experienced cytogeneticists who manually analyze the morphology and structure of chromosomes under a microscope. This method not only has a low degree of automation, but also has great limitations on detection accuracy and efficiency, as a lot of expertise and experience are required to accurately identify chromosome abnormalities. This manpower-dependent method is time-consuming and susceptible to subjective factors, leading to an increased risk of misjudgment and missed detection.

[0004] With the rapid development of artificial intelligence technology, chromosome structural abnormality detection methods based on machine learning and deep learning have gradually emerged. These methods can automatically learn and identify the complex structural features of chromosomes by training a large amount of chromosome image data, thereby greatly improving the automation and accuracy of detection.

[0005] Song Ning et al. proposed a classifier model training method and device for detecting chromosome structural abnormalities, which is a method and device for training a classifier model by obtaining the difference matrix of real normal and artificial defective chromosomes to detect chromosome structural abnormalities. Although the patent supplements data by constructing artificial defective chromosomes, these artificial defects may not fully simulate the complexity and diversity of real defective chromosomes, thus affecting the generalization ability and detection accuracy of the model.

[0006] Cai Yufeng and others proposed a method, system and computer equipment for assisting in identifying abnormal chromosome structure. By generating a reconstructed chromosome image and calculating its similarity with the real chromosome image, it assists doctors in judging abnormal chromosome structure. However, this method determines whether the chromosome is abnormal by setting a similarity abnormality warning threshold, but the similarity threshold between different samples may vary, resulting in insufficient stability and reliability of the test results.

[0007] The structural abnormalities detected by the above methods all focus on an entire chromosome without reaching the band information, resulting in very low granularity. Summary of the invention

[0008] In view of the defects in the prior art, the object of the present invention is to provide a method and system for detecting chromosome karyotype structural abnormalities based on banding coding.

[0009] The method for detecting abnormal chromosome karyotype structure based on banding coding provided by the present invention comprises:

[0010] The steps of constructing the chromosome coding data set are as follows: for the chromosome monomer image collected by microscopic imaging, the coding information is generated according to the gray level of its banding, all monomers in the karyotype are coded to form a chromosome karyotype coding subset, and the coding subsets of all types of chromosomes of normal chromosome karyotypes and structurally abnormal chromosome karyotypes are formed into the training set of the coding network, and the coding network is trained;

[0011] Automatic generation steps of chromosome coding: obtaining the karyotype image to be analyzed and inputting it into the coding network, and the coding network outputs the coding information of the entire chromosome karyotype;

[0012] Chromosome structural abnormality detection steps: Match the obtained chromosome karyotype coding information with the chromosome karyotype database to obtain the abnormality type; if there is no matching result, add the chromosome karyotype information of the case to the database.

[0013] Preferably, the chromosome encoding data set construction step comprises:

[0014] Acquire a metaphase field data set of chromosome division in normal human cells, the data set includes chromosome images captured by microphotography technology, perform image segmentation processing on the acquired metaphase field data set of chromosomes, acquire chromatid images, and straighten and align the segmented chromosome images to ensure that the p arm of the chromosome is located above the image and the q arm is located below;

[0015] All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, the average pixel and variance of the sampled pixel rows are obtained, and for each sampled row, abnormal data that does not meet the preset requirements are eliminated, and then sampling is performed again;

[0016] The obtained sampling data is divided into grayscale levels, and encoding information is generated according to the grayscale level division.

[0017] Preferably, the step of automatically generating the chromosome code comprises:

[0018] Segment the chromosome karyotype image into chromosome monomers and extract the skeleton of each chromosome monomer;

[0019] For each chromatid, the input image is cut into blocks along its skeleton direction, and each block is sent to the Embedding layer as a small area patch to flatten the input sequence;

[0020] The flattened sequence is input into the standard Transformer network to generate chromatid encoding information;

[0021] The coding information of all chromatids in the chromosome karyotype image after being processed by Transformer is integrated to generate the overall coding of the karyotype group.

[0022] Preferably, the chromosome structural abnormality detection step comprises:

[0023] The obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure for similarity, and a similarity threshold is set. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome;

[0024] Determine whether the abnormal chromosome is two chromosomes abnormal, including chromosome translocation and insertion; or one chromosome abnormal, including chromosome deletion, duplication, inversion, insertion, and query the corresponding normal chromosome number;

[0025] The obtained chromosome abnormality coding information is searched and matched with the abnormality coding information in the database, and the specific location of the abnormality and the corresponding band information are determined based on the similarity of the match.

[0026] Preferably, the obtained structural abnormal karyotype code is matched with the karyotype code in the database for similarity;

[0027] If the matching similarity is lower than the threshold, the doctor will judge the type of structural abnormality for this type of karyotype. If there is indeed a structural abnormality, the type of structural abnormality judged by the doctor will be matched with the image of this type of karyotype;

[0028] The group of structural abnormal karyotype images and the matching chromosome codes and structural abnormality types are stored in a database.

[0029] The chromosome karyotype structural abnormality detection system based on banding coding provided by the present invention comprises:

[0030] Chromosome coding data set construction module: for the chromosome monomer image collected by microscopic imaging, the coding information is generated according to the gray level of its banding, all monomers in the karyotype are coded to form a chromosome karyotype coding subset, and the coding subsets of all types of chromosomes of normal chromosome karyotypes and structurally abnormal chromosome karyotypes are formed into the training set of the coding network, and the coding network is trained;

[0031] Chromosome coding automatic generation module: obtains the karyotype image to be analyzed and inputs it into the coding network, which outputs the coding information of the entire chromosome karyotype;

[0032] Chromosome structural abnormality detection module: Match the obtained chromosome karyotype coding information with the chromosome karyotype database to obtain the abnormality type; if there is no matching result, add the chromosome karyotype information of the case to the database.

[0033] Preferably, the chromosome encoding data set construction module includes:

[0034] Acquire a metaphase field data set of chromosome division in normal human cells, the data set includes chromosome images captured by microphotography technology, perform image segmentation processing on the acquired metaphase field data set of chromosomes, acquire chromatid images, and straighten and align the segmented chromosome images to ensure that the p arm of the chromosome is located above the image and the q arm is located below;

[0035] All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, the average pixel and variance of the sampled pixel rows are obtained, and for each sampled row, abnormal data that does not meet the preset requirements are eliminated, and then sampling is performed again;

[0036] The obtained sampling data is divided into grayscale levels, and encoding information is generated according to the grayscale level division.

[0037] Preferably, the chromosome code automatic generation module includes:

[0038] Segment the chromosome karyotype image into chromosome monomers and extract the skeleton of each chromosome monomer;

[0039] For each chromatid, the input image is cut into blocks along its skeleton direction, and each block is sent to the Embedding layer as a small area patch to flatten the input sequence;

[0040] The flattened sequence is input into the standard Transformer network to generate chromatid encoding information;

[0041] The coding information of all chromatids in the chromosome karyotype image after being processed by Transformer is integrated to generate the overall coding of the karyotype group.

[0042] Preferably, the chromosome structural abnormality detection module comprises:

[0043] The obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure for similarity, and a similarity threshold is set. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome;

[0044] Determine whether the abnormal chromosome is two chromosomes abnormal, including chromosome translocation and insertion; or one chromosome abnormal, including chromosome deletion, duplication, inversion, insertion, and query the corresponding normal chromosome number;

[0045] The obtained chromosome abnormality coding information is searched and matched with the abnormality coding information in the database, and the specific location of the abnormality and the corresponding band information are determined based on the similarity of the match.

[0046] Preferably, the obtained structural abnormal karyotype code is matched with the karyotype code in the database for similarity;

[0047] If the matching similarity is lower than the threshold, the doctor will judge the type of structural abnormality for this type of karyotype. If there is indeed a structural abnormality, the type of structural abnormality judged by the doctor will be matched with the image of this type of karyotype;

[0048] The group of structural abnormal karyotype images and the matching chromosome codes and structural abnormality types are stored in a database.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention proposes a method and system for detecting chromosome karyotype structural abnormalities based on banding coding. The method and system can automatically and efficiently identify tiny structural abnormalities of chromosomes, thereby significantly improving the accuracy, sensitivity and granularity of chromosome karyotype structural abnormality detection; and is suitable for chromosome structural abnormality research and diagnosis in the fields of medical genetics, oncology, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0052] Figure 1 A schematic diagram of a method for detecting abnormalities in chromosome karyotype structure based on banding coding;

[0053] Figure 2 Schematic diagram of the grayscale distribution of chromosome bands;

[0054] Figure 3 Generate network diagrams for encoding;

[0055] Figure 4 The figure is a flow chart for determining the abnormality type based on chromosome karyotype coding. DETAILED DESCRIPTION

[0056] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0057] Example 1

[0058] like Figure 1 As shown, the present embodiment provides a method for detecting abnormal chromosome karyotype structure based on banding coding, comprising the following steps:

[0059] S100, chromosome encoding dataset construction.

[0060] Specifically, the overall network structure is as follows Figure 3 As shown, a chromosome karyotype metaphase image is obtained from a microscope or an image database, and a chromatid data set containing normal and abnormal chromosomes is constructed;

[0061] The dataset includes chromosome karyotype images from multiple sources to ensure the robustness and wide applicability of the model;

[0062] Preprocess the acquired chromosome karyotype images, including noise removal, image enhancement and segmentation. Use Gaussian filtering to remove noise, use histogram equalization to enhance image contrast, and use deep learning segmentation algorithms to segment chromatids from the image;

[0063] The segmented data set is classified into 24 categories for normal chromosome karyotype, corresponding to the 22 pairs of autosomes and sex chromosomes X and Y of humans. For abnormal chromosome karyotype images, the abnormal chromosomes are additionally selected and the corresponding structural abnormality types are marked.

[0064] Straighten and correct the classified chromosome images, and ensure that the p arm of the chromosome is located at the top of the image and the q arm is located at the bottom;

[0065] All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, and the average pixel of the sampled pixel rows is obtained;

[0066] For each sampling row, the mean ε and variance σ of the average pixel of the sampling row are calculated based on all chromosome images in this type of chromosome, and the data with average pixels outside [ε-2σ, ε+2σ] are removed as abnormal data;

[0067] The data set after removing abnormal data is sampled again to obtain the average pixel of the sampled pixel rows;

[0068] The average pixel of each sampling row is plotted into a curve, the ordinate of the lowest point is set to 0, the ordinate of the highest point is set to 1, and the parameter δ corresponding to the average pixel is obtained;

[0069]

[0070] Using δ as the threshold, calculate the area of ​​each part of each curve below and above the δ line, and calculate the mean line of each part. The mean line divides the area of ​​each part into equal parts, such as Figure 2 As shown, the curve shown is a processed grayscale curve;

[0071] Calculate the vertical coordinate h of the mean line of each part i and length l i , then l i h i That is, encode the part, combine the codes of all parts to get the total code of the chromosome of this type, and use this coding information as a sub-dataset;

[0072] The coding subsets of all types of chromosomes with normal chromosome karyotypes and structurally abnormal chromosome karyotypes are used to form the training set of the coding network. By training the network, the network has the ability to encode chromosomes.

[0073] S200, chromosome code is automatically generated.

[0074] Specifically, the deep learning algorithm is applied to the training data set for segmentation, so that the chromosome clusters in each chromosome karyotype metaphase image become chromatids;

[0075] Obtain the skeleton of each chromatid;

[0076] The chromatid image is cut into patches along the skeleton in the direction perpendicular to its tangent, and the width of each patch is the width of the chromatid image;

[0077] Input each patch into Linear Projection of Flattened Patches (Embedding layer) and flatten it into a vector X e ;

[0078] Treat each patch as a text unit and add position information X p , and add the special character text unit class token;

[0079] X e +X p Get X t Input to the Encoder part;

[0080] Xt1 =Embedding-Lookup(X)+Positional-Encoding

[0081] The Encoder part consists of the first layer of Layer Normalization, the Multi-Head Attention layer, the second layer of Layer Normalization, and the MLP fully connected layer. t After entering the Encoder, first enter the Layer Normalization layer;

[0082] The role of the Layer Normalization layer is to normalize the input to a standard normal distribution, X t1 After entering the LayerNormalization layer, first calculate X t1 The mean and variance of the elements in;

[0083]

[0084]

[0085] X t1 Each element X ij Perform standardization operation, where ∈ is added to the denominator to prevent the denominator from being zero. The output of Layer Normalization is X t2 ;

[0086]

[0087] Define three matrices W Q , W K , W V , these three matrices will be used to represent X t2 Perform three linear transformations to obtain three new vectors q t , k t , v t ;

[0088] q t =W Q ×X t2

[0089] k t =W K ×X t2

[0090] v t =W V ×X t2

[0091] All qt The vectors are assembled into a large matrix, denoted as the query matrix Q, and all k t The vectors are combined into a large matrix, denoted as the key matrix K, and all v t The vectors are pieced together into a large matrix, denoted as the value matrix V;

[0092] Q=Linear q (X) = XW Q

[0093] K=Linear k (X) = XW K

[0094] V=Linear v (X) = XW V

[0095] Define multiple groups W Q , W K , W V , and then calculate multiple groups of Q, K, V matrices, each group of matrices is a Head;

[0096] For each head, get the attention weight of the first patch, multiply q1 by the key matrix K, and perform a softmax operation on the obtained value so that their sum is 1;

[0097] Multiply the obtained weights by the corresponding value vector v t ;

[0098] Sum the obtained fully weighted value vector to get the output of the first patch. The above steps are expressed in a matrix as where d k is the dimension of the key vector K;

[0099] Perform the same operations on the parameter matrices Q, K, and V in all Heads to get all the outputs after self-attention.

[0100]

[0101] All Head Add to get X A ;

[0102]

[0103] X A With X t1 The sum is input to the second layer, Layer Normalization;

[0104] X1=X A +X t1

[0105] The output of the previous layer is input to the MLP fully connected layer to obtain the output X2, and X1 is added to X2;

[0106] The above steps are repeated six times (N times) to obtain the final output, and the overall output X3 of the final Encoder is obtained;

[0107] Input X3 as the new multi-head attention mechanism parameter matrix K, V into the Decoder;

[0108] The Decoder module consists of the first layer Layer Normalization layer, the Masked Multi-HeadSelf-Attention module, the second layer Layer Normalization layer, the Multi-Head Self-Attention module, the third layer Layer Normalization layer, and the MLP fully connected layer. The Multi-Head Self-Attention module, the Layer Normalization layer, and the MLP fully connected layer are the same as those in the Encoder module.

[0109] Before the decoder, the unit code at time t is continuously input as ground truth (the reference true value required for training), and this input directly enters the internal Mask Multi-Head Attention module;

[0110] The ground truth passes through the Word Embedding and Positional Encoding modules, and the output is X 4, Input X4 into the Layer Normalization layer to obtain X5;

[0111] Input X5 into the Masked Multi-Head Self-Attention module and perform the steps in the Encoder (i.e. calculate ), and get the output Scaled Scores;

[0112] Mask the Scaled Scores and add them to a matrix whose lower triangle is 0 and upper triangle is all negative infinity;

[0113]

[0114] Perform softmax operation on the obtained Masked Scores, and the obtained matrix is ​​the weight X6 of each code;

[0115] Add the obtained weights X6 and X4 as the Q matrix, and input them into the second layer of Layer Normalization together with the K and V matrices output by the Encoder. The subsequent steps are the same as the Encoder part;

[0116] The above Decoder operation is performed 6 times, and the final output is X7;

[0117] Let X7 enter the fully connected layer and perform softmax operation;

[0118] Fine-tune the network so that it has the ability to encode abnormal chromosomes. Input abnormal chromosomes and their corresponding abnormal codes into the network, freeze the first three layers of the Encoder part, load a new data set for training, obtain the final output code, complete the training, and obtain the weight file output by the training network.

[0119] A set of chromosome karyotype metaphase division images is selected as the test data set, which includes normal chromosome images and abnormal chromosome images;

[0120] According to the weight file output by the training network, the chromosome karyotype encoding corresponding to the test data set is obtained.

[0121] S300, detection of chromosome structural abnormalities.

[0122] Specifically, Figure 4 As shown in the flowchart, the obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure, and the dynamic time warping DTW algorithm is used to calculate the similarity between the two. The similarity threshold is set to 0.8. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome.

[0123] For chromosomes that are determined to be abnormal, first determine whether two chromosomes are abnormal (such as chromosome translocation, insertion, etc.) or one chromosome is abnormal (such as chromosome deletion, duplication, inversion, insertion, etc.). According to the type of abnormality, query the corresponding normal chromosome number for further analysis;

[0124] Compare the obtained abnormal coding information with all abnormal chromosome coding information in the data set, set all abnormal chromosome coding information in the data set as the left fixed point set in the bipartite graph, and set the obtained abnormal coding information to be matched as the right fixed point set in the bipartite graph;

[0125] In the initialization phase, the matching status of all nodes in the bipartite graph is set to unmatched. All possible encoding pairing relationships are represented by constructing a bipartite graph;

[0126] Traverse all unmatched nodes. For each unmatched node, use depth-first search (DFS) to find an augmenting path. An augmenting path is defined as a path whose starting and ending points are both unmatched nodes, and the edges in the path belong alternately to the currently matched and unmatched edges. If an augmenting path is found, add the edges on the path to the match and update the matching status.

[0127] If no augmenting path is found, the search ends. If there is an augmenting path, go back to the previous step and continue searching. If there is no augmenting path, the current match is the maximum match and the algorithm ends.

[0128] Output the maximum matching result, that is, as many matching edges as possible in the bipartite graph;

[0129] Determine the similarity between the maximum matching result and the structural abnormal karyotype, set the similarity threshold to 0.8, and when the similarity is less than the threshold, it is considered a match failure;

[0130] When the similarity is greater than the threshold, the specific abnormal part and its corresponding normal chromosome number are determined by combining the results of the augmented path;

[0131] The karyotype images that cannot be matched with the abnormal chromosome karyotypes in the data set are handed over to doctors to determine whether the karyotype has structural abnormalities. If the doctor determines that the karyotype has structural abnormalities and there are no such abnormalities in the database, the karyotype with structural abnormalities and its matching coding information and the type of structural abnormality are stored in the database;

[0132] According to the matching results, the abnormal type and position of the chromosome are generated, and the test report is output, including the abnormal chromosome number, abnormal type and specific abnormal band position;

[0133] The test results and abnormal chromosome images are displayed, and users can input images and view test results through the user interface.

[0134] Example 2

[0135] The present invention also provides a chromosome karyotype structural abnormality detection system based on banding coding. The chromosome karyotype structural abnormality detection system based on banding coding can be realized by executing the process steps of the chromosome karyotype structural abnormality detection method based on banding coding, that is, those skilled in the art can understand the chromosome karyotype structural abnormality detection method based on banding coding as a preferred implementation of the chromosome karyotype structural abnormality detection system based on banding coding.

[0136] The chromosome karyotype structural abnormality detection system based on banding coding provided by the present invention comprises:

[0137] Chromosome coding data set construction module: for the chromosome monomer image collected by microscopic imaging, the coding information is generated according to the gray level of its banding, all monomers in the karyotype are coded to form a chromosome karyotype coding subset, and the coding subsets of all types of chromosomes of normal chromosome karyotypes and structurally abnormal chromosome karyotypes are formed into the training set of the coding network, and the coding network is trained;

[0138] Chromosome coding automatic generation module: obtains the karyotype image to be analyzed and inputs it into the coding network, which outputs the coding information of the entire chromosome karyotype;

[0139] Chromosome structural abnormality detection module: Match the obtained chromosome karyotype coding information with the chromosome karyotype database to obtain the abnormality type; if there is no matching result, add the chromosome karyotype information of the case to the database.

[0140] The chromosome encoding data set construction module includes:

[0141] Acquire a metaphase field data set of chromosome division in normal human cells, the data set includes chromosome images captured by microphotography technology, perform image segmentation processing on the acquired metaphase field data set of chromosomes, acquire chromatid images, and straighten and align the segmented chromosome images to ensure that the p arm of the chromosome is located above the image and the q arm is located below;

[0142] All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, the average pixel and variance of the sampled pixel rows are obtained, and for each sampled row, abnormal data that does not meet the preset requirements are eliminated, and then sampling is performed again;

[0143] The obtained sampling data is divided into grayscale levels, and encoding information is generated according to the grayscale level division.

[0144] The chromosome code automatic generation module comprises:

[0145] Segment the chromosome karyotype image into chromosome monomers and extract the skeleton of each chromosome monomer;

[0146] For each chromatid, the input image is cut into blocks along its skeleton direction, and each block is sent to the Embedding layer as a small area patch to flatten the input sequence;

[0147] The flattened sequence is input into the standard Transformer network to generate chromatid encoding information;

[0148] The coding information of all chromatids in the chromosome karyotype image after being processed by Transformer is integrated to generate the overall coding of the karyotype group.

[0149] The chromosome structural abnormality detection module comprises:

[0150] The obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure for similarity, and a similarity threshold is set. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome;

[0151] Determine whether the abnormal chromosome is two chromosomes abnormal, including chromosome translocation and insertion; or one chromosome abnormal, including chromosome deletion, duplication, inversion, insertion, and query the corresponding normal chromosome number;

[0152] The obtained chromosome abnormality coding information is searched and matched with the abnormality coding information in the database, and the specific location of the abnormality and the corresponding band information are determined based on the similarity of the match.

[0153] The obtained structural abnormal karyotype code is matched with the karyotype code in the database for similarity;

[0154] If the matching similarity is lower than the threshold, the doctor will judge the type of structural abnormality for this type of karyotype. If there is indeed a structural abnormality, the type of structural abnormality judged by the doctor will be matched with the image of this type of karyotype;

[0155] The group of structural abnormal karyotype images and the matching chromosome codes and structural abnormality types are stored in a database.

[0156] Those skilled in the art know that, in addition to implementing the system, device and its various modules provided by the present invention in a purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, the system, device and its various modules provided by the present invention can be considered as a hardware component, and the modules included therein for implementing various programs can also be considered as structures within the hardware component; the modules for implementing various functions can also be considered as both software programs for implementing the method and structures within the hardware component.

[0157] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for detecting chromosome karyotype structural abnormalities based on banding coding, characterized in that: include: The steps of constructing the chromosome coding data set are as follows: for the chromosome monomer image collected by microscopic imaging, the coding information is generated according to the gray level of its banding, all monomers in the karyotype are coded to form a chromosome karyotype coding subset, and the coding subsets of all types of chromosomes of normal chromosome karyotypes and structurally abnormal chromosome karyotypes are formed into the training set of the coding network, and the coding network is trained; Automatic generation steps of chromosome coding: obtaining the karyotype image to be analyzed and inputting it into the coding network, and the coding network outputs the coding information of the entire chromosome karyotype; Chromosome structural abnormality detection steps: Match the obtained chromosome karyotype coding information with the chromosome karyotype database to obtain the abnormality type; if there is no matching result, add the chromosome karyotype information of the case to the database.

2. The method for detecting chromosome karyotype structural abnormality based on banding coding according to claim 1, characterized in that: The chromosome encoding data set construction step comprises: Acquire a metaphase field data set of chromosome division in normal human cells, the data set includes chromosome images captured by microphotography technology, perform image segmentation processing on the acquired metaphase field data set of chromosomes, acquire chromatid images, and straighten and align the segmented chromosome images to ensure that the p arm of the chromosome is located above the image and the q arm is located below; All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, the average pixel and variance of the sampled pixel rows are obtained, and for each sampled row, abnormal data that does not meet the preset requirements are eliminated, and then sampling is performed again; The obtained sampling data is divided into grayscale levels, and encoding information is generated according to the grayscale level division.

3. The method for detecting chromosome karyotype structural abnormality based on banding coding according to claim 1, characterized in that: The chromosome code automatic generation step comprises: Segment the chromosome karyotype image into chromosome monomers and extract the skeleton of each chromosome monomer; For each chromatid, the input image is cut into blocks along its skeleton direction, and each block is sent to the Embedding layer as a small area patch to flatten the input sequence; The flattened sequence is input into the standard Transformer network to generate chromatid encoding information; The coding information of all chromatids in the chromosome karyotype image after being processed by Transformer is integrated to generate the overall coding of the karyotype group.

4. The method for detecting chromosome karyotype structural abnormality based on banding coding according to claim 1, characterized in that: The chromosome structural abnormality detection step comprises: The obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure for similarity, and a similarity threshold is set. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome; Determine whether the abnormal chromosome is two chromosomes abnormal, including chromosome translocation and insertion; or one chromosome abnormal, including chromosome deletion, duplication, inversion, insertion, and query the corresponding normal chromosome number; The obtained chromosome abnormality coding information is searched and matched with the abnormality coding information in the database, and the specific location of the abnormality and the corresponding band information are determined based on the similarity of the match.

5. The method for detecting chromosome karyotype structural abnormality based on banding coding according to claim 4, characterized in that: The obtained structural abnormal karyotype code is matched with the karyotype code in the database for similarity; If the matching similarity is lower than the threshold, the doctor will judge the type of structural abnormality for this type of karyotype. If there is indeed a structural abnormality, the type of structural abnormality judged by the doctor will be matched with the image of this type of karyotype; The group of structural abnormal karyotype images and the matching chromosome codes and structural abnormality types are stored in a database.

6. A chromosome karyotype structural abnormality detection system based on banding coding, characterized in that: include: Chromosome coding data set construction module: for the chromosome monomer image collected by microscopic imaging, the coding information is generated according to the gray level of its banding, all monomers in the karyotype are coded to form a chromosome karyotype coding subset, and the coding subsets of all types of chromosomes of normal chromosome karyotypes and structurally abnormal chromosome karyotypes are formed into the training set of the coding network, and the coding network is trained; Chromosome coding automatic generation module: obtains the karyotype image to be analyzed and inputs it into the coding network, which outputs the coding information of the entire chromosome karyotype; Chromosome structural abnormality detection module: Match the obtained chromosome karyotype coding information with the chromosome karyotype database to obtain the abnormality type; if there is no matching result, add the chromosome karyotype information of the case to the database.

7. The chromosome karyotype structural abnormality detection system based on banding coding according to claim 6, characterized in that: The chromosome encoding data set construction module includes: Acquire a metaphase field data set of chromosome division in normal human cells, the data set includes chromosome images captured by microphotography technology, perform image segmentation processing on the acquired metaphase field data set of chromosomes, acquire chromatid images, and straighten and align the segmented chromosome images to ensure that the p arm of the chromosome is located above the image and the q arm is located below; All chromosome data in each type of straightened chromosome data set are sampled from top to bottom, the average pixel and variance of the sampled pixel rows are obtained, and for each sampled row, abnormal data that does not meet the preset requirements are eliminated, and then sampling is performed again; The obtained sampling data is divided into grayscale levels, and encoding information is generated according to the grayscale level division.

8. The chromosome karyotype structural abnormality detection system based on banding coding according to claim 6, characterized in that: The chromosome code automatic generation module comprises: Segment the chromosome karyotype image into chromosome monomers and extract the skeleton of each chromosome monomer; For each chromatid, the input image is cut into blocks along its skeleton direction, and each block is sent to the Embedding layer as a small area patch to flatten the input sequence; The flattened sequence is input into the standard Transformer network to generate chromatid encoding information; The coding information of all chromatids in the chromosome karyotype image after being processed by Transformer is integrated to generate the overall coding of the karyotype group.

9. The chromosome karyotype structural abnormality detection system based on banding coding according to claim 6, characterized in that: The chromosome structural abnormality detection module comprises: The obtained chromosome karyotype code is compared with the chromosome karyotype code of the normal structure for similarity, and a similarity threshold is set. When the similarity is less than the threshold, it is regarded as a normal chromosome; when the similarity is greater than the threshold, it is regarded as an abnormal chromosome; Determine whether the abnormal chromosome is two chromosomes abnormal, including chromosome translocation and insertion; or one chromosome abnormal, including chromosome deletion, duplication, inversion, insertion, and query the corresponding normal chromosome number; The obtained chromosome abnormality coding information is searched and matched with the abnormality coding information in the database, and the specific location of the abnormality and the corresponding band information are determined based on the similarity of the match.

10. The chromosome karyotype structural abnormality detection system based on banding coding according to claim 9, characterized in that: The obtained structural abnormal karyotype code is matched with the karyotype code in the database for similarity; If the matching similarity is lower than the threshold, the doctor will judge the type of structural abnormality for this type of karyotype. If there is indeed a structural abnormality, the type of structural abnormality judged by the doctor will be matched with the image of this type of karyotype; The group of structural abnormal karyotype images and the matching chromosome codes and structural abnormality types are stored in a database.

Citation Information

Patent Citations

  • Chromosome sequencing method based on stripe recognition

    CN110533672A

  • Chromosome case level abnormity prompting system based on visual semantic combination

    CN116977338A

  • AU2019200162A1

Cited By

  • Chromosome classification method based on sequence cross correlation

    CN120913200A