Method for counting number of complex overlapping cells in microscopic image

Abstract

The invention relates to a method for counting the number of complex overlapping cells in a microscopic image. The method comprises the following steps: pre-processing and further binarizing an original colored microscopic image; locating the concave points in the outline of all the cells by using a chain code difference method, connecting the concave points which are paired with each other, and segmenting the slightly overlapping cells; and finally, counting the number of cells by using a region area statistical method. Compared with the prior art, the invention has the following advantages that: the original colored microscopic image is pre-processed and further binarized according to the color features of the colored microscopic image; the outline of the cells are detected and further segmented by using the chain code difference method because the slightly overlapping cells are characterized in that the concave points of the cells are obvious; and finally, the numbers of the heavily overlapping cells and the single cells are calculated by using the region area statistical method, so the method has small error and high accuracy.

Description

technical field [0001] The invention relates to the field of microscopic cell statistics, in particular to a method for counting the number of complicated adhesion cells in microscopic images. Background technique [0002] The microscopic cell image analysis includes the identification and measurement of the microscopic cell image, the statistics of the number, and the measurement of various parameters such as area, perimeter, and shape factor. In actual microscopic cell images, due to the slice and the cells themselves, multiple cells often overlap to form a larger area, such as blood cells, lymphocytes, cancer cells, etc. Counting the overlapping adherent cells that clump together is a difficult point in the analysis of microscopic cell images. At present, many segmentation algorithms have been proposed to deal with cell overlap and adhesion, such as erosion-dilation method based on mathematical morphology, approximate concave point cutting of binary image polygons, geode...

AI Technical Summary

Problems solved by technology

However, this method is relatively simple, and the times of corrosion and expansion must be manually determined empirically, and the irreversibility of corrosion and expansion will cause certain errors

Dimple segmentation is better for a small number of cells and the area of ​​the adhesion cells is less than one-third, but for half or more than half of the adhesion cells, the effect is not very obvious; the watershed segmentation seed point is more difficult to find and it is wrong for the adhesion of the severe seed point high probability

Images