CHROMOSOME IDENTIFYING METHOD BASED ON DEEP LEARNING

摘要

The invention discloses a chromosome identifying method based on deep learning, relating to the technical field of chromosome identification. At present, the methods for analyzing chromosomes generally adopt manual operation; an examination doctor costs a plenty of training time to master the knowledge for identifying each chromosome type, thereby having heavy workload; an experienced doctor generally needs more than two weeks to analyze and identify chromosomes of a patient, the time period of the entire process is long; and artificial identification is strongly subjective, can be easily influenced by external environment, and has low accuracy. The present invention uses a method based on deep learning to accurately and efficiently identify the chromosome types. Compared with the existing identification technology, the method can effectively improve the efficiency of analyzing chromosome karyotypes, shorten the time of identification and sorting, finish the automatic classification and sorting of chromosomes with high accuracy. Besides, the present invention can effectively alleviate the workload of doctors and avoid external interference; the working procedures are concise and reasonable and can be promoted and applied in a large scale.