Medullary blood cell identification method and device, storage medium and electronic equipment

Abstract

The invention provides a medullary blood cell identification method and device, a storage medium and electronic equipment. The scheme comprises the following steps: acquiring a first medullary blood cell microscope image; performing data enhancement transformation on the first medullary blood cell microscope image to obtain a second medullary blood cell microscope image; performing data standardization processing on the second medullary blood cell microscope image to obtain a third medullary blood cell microscope image; and carrying out instance segmentation processing on the third medullary blood cell microscope image to obtain a target medullary blood cell microscope image, and determining target identification information. According to the scheme, the data distribution of the medullarysystem blood cells in the medullary blood cell microscope images is enriched through data enhancement transformation and data standardization processing, the generalization and robustness of processing are improved, and over-fitting is prevented; and instance segmentation algorithm processing is performed on the processed image by adopting a feature extraction network, and thus classification, positions and number of myeloid hemocytes on a microscope image can be completely and accurately identified.

Description

technical field [0001] The present invention relates to the field of medical examination, and more specifically, to a method, device, storage medium and electronic equipment for identifying myeloid blood cells. Background technique [0002] At present, the existing automatic cell identification and classification technology usually uses convolutional neural network to extract the features of cell images, and then uses binary classification or multi-classification technology to classify the features. The disadvantages of the existing technology are: it cannot process the cells that are crowded or stacked in the image; it cannot classify the cells accurately; it cannot give the location information and specific shape information of the cells. [0003] For example, the Chinese invention patent with the application number CN201911265104.4 and titled "A Computer-Based Automatic Identification and Classification Method for White Blood Cells" discloses the use of convolutional neur...

AI Technical Summary

Problems solved by technology

The shortcomings of the existing technology are: unable to deal with crowded or stacked cells in the image; unable to classify the cells accurately; unable to give the position information and specific shape information of the cells

However, the invention patent uses convolutional neural network classification technology to identify white blood cells, which cannot give the position information and specific shape information of white blood cells in the image, and the invention only roughly classifies white blood cells into lymphocytes, eosinophils, and basophils, neutrophils

Images