Cell classified statistic method based on probability distribution

Abstract

The invention relates to a classification recognition technology for detecting different types of cells through human blood cells. The technology comprises the following steps that A, initial partition is conducted on an original scatter diagram, and subgroup types are determined; B, Gaussian distribution characteristic parameters of all subgroups are calculated respectively, and the maximum log likelihood function value is calculated; C, iteration calculation is entered, and the probability distribution of cells located in the subgroups is calculated according to the Gaussian distribution function output from last iteration; D, Gaussian distribution characteristic parameters of all subgroups are calculated again according to the probability distribution of the cells, and the maximum log likelihood function value is calculated; E, whether the maximum log likelihood function value is convergent or not is judged, and if not, the third step and the fourth step are repeated to conduct iteration calculation; if yes, iteration is stopped, the Gaussian distribution characteristic parameters of all subgroups and the subgroup types of all the cells are output. According to the classification recognition technology for detecting the different types of the cells through the human blood cells, the requirement of accuracy on partition of an initial boundary of the scatter diagram is low, the complexity of the algorithm is reduced, the adaptive capacity is strong, and the accuracy and the stability of particle classification are improved.

Description

technical field [0001] The invention relates to a classification and identification technology for detecting different types of cells in human blood cells, in particular to a method for classification and statistics of cells based on probability distribution. Background technique [0002] With the continuous improvement of medical clinical requirements, the identification and division of human blood cells requires more detailed and accurate. At present, the five-differentiation hematology analyzer has been favored by medical units at all levels. The five-differentiation blood cell analyzer divides white blood cells into five subgroups according to the internal and external structural characteristics of the cells, which specifically include lymphocytes (Lym), monocytes (Mon), neutrophils (Neu), eosinophils (Eos), basophils (Baso). At this stage, most instrument manufacturers use laminar flow technology combined with optical detection. High-end five-category models use fluore...

AI Technical Summary

Problems solved by technology

Patent CN103776751 determines the boundary line of each subgroup by finding closed contour lines, which can deal with irregular boundaries and improve the accuracy of classification, but the subgroups that overlap with the boundary are easily identified as the same subgroup, thus Misjudgment

Images