Nuclear staining cell counting method based on deep learning of incomplete marker, computer equipment and storage medium

Abstract

The invention relates to a nuclear staining cell counting method based on deep learning of an incomplete label, computer equipment and a storage medium, and the method comprises the following steps: (1) making label data: loading a pathological image to label software, and obtaining sub-mask image data pairs of all sub-images and positive cells; (2) model training: training a convolutional neural network model to respectively obtain a trained positive cell convolutional neural network model and a trained negative cell convolutional neural network model; (3) reasoning stage: inputting pathological images to be detected into the trained convolutional neural network model to obtain real mask images; and (4) post-treatment stage: calculating the number of positive cells and negative cells, and calculating to obtain the proportion p of the positive cells in all the cells. The method does not need extra parameters, is high in universality, greatly reduces manual adjustment, and effectively improves the recognition accuracy and robustness. The method is faster, more accurate and more effective in data marking.

Description

technical field [0001] The invention relates to a nuclear staining cell counting method, computer equipment and storage medium based on deep learning of incomplete markers, and belongs to the technical field of deep learning. Background technique [0002] Cancer is now the leading cause of death in the world. In addition to improving cancer treatment technology, early diagnosis and screening of cancer is also an important aspect of improving the survival rate of cancer patients. The auxiliary diagnosis of pathological sections is to carry out immunohistochemical typing according to the positive status of ER, PR and Ki-67, and to determine different diagnosis and treatment plans according to different types. This depends on calculating the proportion of negative cells and positive cells in the total cells in the pathological section. Usually, when a doctor judges a positive state, he needs to observe different fields of view in the magnified section of the microscope, and o...

AI Technical Summary

Problems solved by technology

[0007] (1) Traditional cell counting based on methods such as threshold segmentation and color channel separation have disadvantages such as low accuracy, inability to be used universally, and manual adjustment of parameters under different types of cells.

[0008] (2) The cell counting method based on deep learning relies on accurate labeling data, but manual pixel-by-pixel labeling by doctors is time-consuming and laborious, and it is difficult to form large-scale effective training data

[0009] (3) The shape of the cells is complex, and it is difficult to complete the intelligent learning and expansion of complete pixel-by-pixel marking. It is easy for the deep learning model to fall into a state where it can only recognize cells of a single shape and a single color

Images