Classifier model training method and device for detecting chromosome structure abnormality

Abstract

The invention provides a classifier model training method and device for detecting chromosome structure abnormality. The training method comprises the steps that real normal chromosomes are obtained, a first difference matrix between real normal homologous chromosome pairs is obtained, and two homologous chromosomes in the real normal homologous chromosome pairs are both real normal chromosomes; artificial defect chromosomes are constructed based on the real normal chromosomes, a second difference matrix between the artificial defect homologous chromosome pairs is obtained, and at least one of two homologous chromosomes in the artificial defect homologous chromosome pairs is an artificial defect chromosome; training at least by taking the first difference matrix and the second difference matrix as samples to obtain a classifier model for detecting the chromosome structure abnormality; and judging whether the to-be-diagnosed user has chromosome abnormality based on the classifier model. According to the method, abundant structure abnormal chromosomes of different types are artificially constructed, so that sufficient and balanced samples are provided for classifier model training.

Description

technical field [0001] The present invention relates to the detection of chromosome structural abnormalities, in particular to a classifier model training method and device for detecting chromosome structural abnormalities. Background technique [0002] Chromosomal abnormalities, including deletions, duplications or irregularities of chromosomal DNA, are the underlying cause of various genetic diseases. About 0.6% of live births have chromosomal abnormalities, which often lead to deformities and / or developmental disabilities. Diseases caused by chromosomal abnormalities can have serious consequences, such as 25% of miscarriages and stillbirths due to chromosomal abnormalities, and 50% to 60% of miscarriages in the first trimester. With the help of chromosomal abnormality detection, clinicians can identify all abnormalities that may lead to birth defects. According to the general understanding of chromosomal abnormalities, they can be roughly divided into two types: quantit...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the problem in the prior art, if it is necessary to automatically screen abnormal chromosome structure detection methods with the help of deep learning algorithms, it is necessary to train a classifier model that can accurately classify normal and abnormal structural chromosomes. The number is not abundant and the high cost and difficulty of obtaining, only using a small number of existing chromosomes with real structural abnormalities, cannot support the construction of complex deep recognition models. The present invention provides a classifier model training method for detecting abnormal chromosome structures and device

Images