Detection Of Lymphocytes In H&E Stained Histopathological Images

The research of medical image recognition has made great progress with the development of image processing technology and medical imaging technology.Medical image recognition technology has been able to assist doctors and experts to make diagnosis results for patients,because lymphocytes are diagnosed in cancer and prognosis of patients.The importance of the segmentation and detection of lymphocyte nuclei is an important task in the current field of histopathological image processing and analysis.In this study,a lymphocyte nuclear detection model was constructed by analyzing the morphological characteristics and texture characteristics of lymphocyte nuclei in histopathological images of H&E staining.This model screened the effective features that can distinguish lymphocyte nuclei and other nuclei,and it has certain innovation.The constructed model is tested with the addition of noise and is highly robust.The specific contents include:(1)Dividing H&E stained histopathological images into training set samples and test set samples,training training set samples in training stage,using probability density function graph to screen out effective features,and using the selected features The classifier training establishes a classification model to construct a lymphocyte nuclear classifier;(2)classifies the nuclei using a classifier obtained during the training phase to detect the image output with lymphocyte nuclear markers;(3)according to different characteristics The values are respectively constructed into three classification models,and the optimal classification model is obtained through experiments.(4)Various noises are added to the test set image,and the robustness of the lymphocyte nuclear model is verified by multiple experimental tests.The development of automated pathological image analysis not only saves diagnostic time and cost,but also facilitates the acquisition of quantitative information and provides relevant prognostic information.This subject uses H&E staining histopathology images of lung cancer patients as raw data,which combines the a priori knowledge of hospital experts,and uses digital image processing,machine learning and pattern recognition to realize computer automatic labeling of lymphocyte nuclei.And help with the quantitative analysis of lymphocytes.