Chromosome polarity recognition method and system based on deep learning

Abstract

The invention provides a chromosome polarity recognition method based on deep learning. The method comprises the steps of (1) collecting a data set, (2) constructing a training set and a test set, (3) performing learning training on a chromosome polarity recognition model based on the training set, and (4) inputting the test set into the chromosome polarity recognition model for testing, and outputting a polarity result of a chromosome to be predicted. The invention further provides a chromosome polarity recognition system based on deep learning. The method and the system provided by the invention can accurately judge the current chromosome polarity category based on a deep learning classification algorithm, and finish chromosome polarity adjustment according to the current chromosome polarity category, so that all chromosomes are kept in a state that short arms face upwards. According to the method and the system, the chromosome polarity recognition accuracy reaches 96.36%, the data source is simple, the automation degree of chromosome analysis is high, the process is more concise, and the method and the system have wide industrial practicability.

Description

technical field [0001] The invention relates to computer vision image processing, chromosome counting and other technical fields, in particular to a chromosome polarity recognition method and system. Background technique [0002] Chromosomal karyotype analysis is an important means of discovering chromosomal diseases. Chromosome number or structural abnormalities can be found through chromosomal karyotype analysis. In order to facilitate clinicians to give corresponding diagnostic results based on the morphological structure of chromosomes, all the separated chromosomes will be arranged in order, and at the same time ensure that they are in a vertical state with the short arm of the chromosome facing up and the long arm facing down, that is, the polarity of the chromosome is adjusted, and finally Form an accurate and clear chromosome karyotype map. [0003] At present, chromosome analysis systems generally rely heavily on manual adjustment of chromosome polarity, and doctor...

AI Technical Summary

Problems solved by technology

The advantage of these methods is that the identification of chromosome type and polarity can be completed at the same time, which saves the computing time and resources required for computing; but the disadvantage is that the identification of chromosome type should not be related to the polarity of chromosomes, that is, the judgment of chromosome type should not be affected by the polarity of chromosomes. Impact

Furthermore, none of the current methods elucidate how to obtain chromosomes in an upright state, which is a prerequisite step for subsequent chromosome polarity flipping

[0004]Currently, manual adjustment of chromosome polarity is a widely used method. The current chromosome analysis system provides a click-and-drag function, which is convenient for clinicians to analyze the observed abnormal chromosome polarity. Adjustment, that is, flipping the abnormally polarized chromosome up and down to ensure that the short arm is facing up and the long arm is facing down, requires manpower and is inefficient. In particular, the long and short arms of some chromosomes are not easy to identify, which makes diagnosis difficult.

[0005]The chromosome recognition method based on deep learning makes full use of the powerful feature extraction ability of deep learning technology, and also supplements the features extracted by computer graphics, and finally uses two independent classification Chromosome type recognition and dyeing polarity recognition respectively, but the main purpose is to complete the chromosome type recognition, so it is first necessary to ensure that the chromosome type recognition has nothing to do with the chromosome polarity, that is, whether the short arm of the chromosome is facing up or not, the model needs to accurately judge the chromosome category, so there is a certain conflict between the two tasks, and the learned features cannot make the two tasks achieve the optimal effect at the same time.

In addition, the default input chromosome of this method is in a vertical state, and it does not consider that the single chromosome segmented from the metaphase image of cell division is not necessarily in a vertical state, which is a major technical deficiency

[0006]Manual adjustment of chromosome polarity is inefficient, and the defect of chromosome identification method based on deep learning is due to the existence of features required for chromosome type identification and chromosome polarity identification Conflict and it is difficult to clarify how to obtain chromosomes in vertical state from the segmented chromosomes in arbitrary angle state, so that the chromosome polarity recognition model can judge and adjust the chromosome polarity

The purpose of the present invention is to solve the problems of low artificial efficiency in the process of chromosome polarity recognition, task conflicts of chromosome recognition algorithms based on deep learning and lack of technical solutions for key steps of chromosome rotation, and propose a chromosome polarity recognition method and system

Images