Deep learning-based cell detection counting method and system

Abstract

The invention provides a deep learning-based cell detection counting method and system. The method comprises the steps of scanning a pathological section, and converting entity section information into a digital pathological image; performing real situation marking on cells or cell nuclei in the digital pathological image, recording mark information, and constructing a data set based on the digital pathological image and the mark information; dividing the data set into a training set and a test set; inputting the training set to a deep convolutional neural network-long-short-term memory network for performing training; and inputting the test set to the trained deep convolutional neural network-long-short-term memory network, and outputting a counting result. According to the method and thesystem, the robustness and accuracy of cell detection counting can be remarkably improved; the digital pathological image is subjected to quantitative analysis with a high-flux processing rate; and reliable data analysis support is provided for pathological image-based medical research.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a cell detection and counting method and system based on deep learning. Background technique [0002] In recent years, digital pathological images and microscopic image analysis have played a very important role in pathological diagnosis. They can provide a large amount of information for computer-aided diagnosis, and then enable digital pathological images to be quantitatively analyzed with high-throughput processing rates. Provides great convenience for pathologists. Today's automated digital pathology image analysis has attracted much attention in research and clinical practice. Compared with labor-intensive, time-consuming, poorly reproducible, and highly subjective manual processes, computer-aided analysis methods can Under the premise of ensuring accuracy, significantly improving the reproducibility, timeliness and objectivity of observations can save basi...

AI Technical Summary

Problems solved by technology

These target detection algorithms based on traditional features are often only suitable for cell detection tasks with simple image background and sparse cell distribution. The interference of factors such as different size, shape, and depth of staining leads to poor robustness and low accuracy of cell detection and counting

Images