A slow feature-based cell division recognition method and its recognition device

Abstract

The invention discloses a slow feature-based cell division identification method and its identification device. The method includes: adopting an unsupervised slow feature analysis method, extracting cell data to obtain a slow feature function; calculating the cumulative square offset feature of the slow cell feature , to obtain the arrangement of slow feature change rates from small to large; use the method of model learning to detect the final cumulative square offset feature, and obtain the probability of mitosis in the process of cell data changing over time; if the output category is marked as 1, then The test data contains mitosis, if the output class is marked as 0, the test data does not contain mitosis. The device includes: a first acquisition module, a second acquisition module, a third acquisition module and an output module. The present invention reduces the difficulty of cell feature extraction, improves the accuracy of cell feature extraction, and provides a good solution for the identification and classification of subsequent dividing cells. The conditions are convenient for the identification and tracking of cells.

Description

technical field [0001] The invention relates to the field of image feature and pattern recognition, in particular to a slow feature-based cell division recognition method and a recognition device thereof, and in particular to the application of the slow feature to the field of cells. Background technique [0002] Cell biology is an important subject that studies cell structure, function, and living organisms. Cells promote the development of organisms through processes such as growth, division, aging, and death. Among them, cell division promotes the growth and development of organisms, as well as metabolism, which is of great significance to the process of cell growth. Then, how to quickly and accurately identify the process of cell division is of inestimable value to the study of cell changes and the development of things. [0003] In the research process of cell data, due to the large amount of original image data, it is usually not used as a feature to directly partici...

AI Technical Summary

Problems solved by technology

[0006] 1) The existing shape or structure feature extraction methods require accurate segmentation of cell regions, but microscope images often contain strong noise, making it difficult to accurately segment cells, so methods based on shape features usually have poor performance and generalization ability lower;

[0007] 2) Cell-based color feature extraction, because cells are relatively small, difficult to observe, and are seriously affected by background, staining, etc., so observation and tracking are relatively difficult, and there are problems such as high misjudgment rate and low work efficiency;

[0008] 3) Recognition methods based on phase models often require large-scale timing information to construct complex models. At the same time, the learning of the model has a high computational complexity, which reduces the speed of cell division recognition.

Images