A Cell Detection Method Based on Sliding Window and Deep Structure Extraction Features

Abstract

The invention discloses a cell detection method based on a sliding window and a depth structure extraction feature. First, the depth model is used to extract the feature, and then the sliding window technology is used to act on the automatic detection method of the cells in the pathological slice image. The specific steps include: taking blocks from the sliced ​​image, training the feature extraction model stacked sparse self-encoder, training the detector, scanning the large image with a sliding window, and labeling the cell position. The present invention takes a large slice image as the research object, and adopts a new method of detector and sliding window, which can more accurately find the position of cells in the image, and is more rapid and comprehensive, and can also play a significant role in some cells that are not very obvious in the image. Good detection effect. The automatic cell detection method proposed by the present invention can assist clinicians to quantitatively evaluate digital pathology slices, perform clinical diagnosis accurately and quickly, and reduce diagnostic differences between different observers or the same observer at different time periods.

Description

technical field [0001] The invention discloses a cell detection method based on a sliding window and a depth structure extraction feature, and relates to the technical field of image information processing. Background technique [0002] With the generation of digital scanning technology for large slice images and the improvement of scanning efficiency, the digital display and storage of histopathological slides has become realistic and feasible. Higher quality analysis of pathological images is possible with digital technology. Because the characteristics of various cancer cells and tissues can be found from the pathological imaging images of tissue slices, and can be used to assist doctors in diagnosis, but there are still few technical researches on medical image processing. Analysis tools for pathological images are very important. [0003] A large part of the research on histopathological images focuses on the resolution of specific tissue structures, such as lymphocyt...

AI Technical Summary

Problems solved by technology

[0006] The technical problem to be solved by the present invention is to overcome the problem that the existing segmentation-based detection algorithm cannot handle pathological slice cells due to multiple changes in cell shape and different scales, and to provide a cell detection method based on sliding windows and extracting deep features. While using a large amount of unlabeled cell data, it can effectively detect various types of cells in the picture, and achieve higher accuracy results than segmentation-based detection.

Images