Method and device for improving the accuracy of tps calculation of PD-L1 immunohistochemical pathological sections

Abstract

The present application discloses a method for improving the accuracy of TPS calculation of PD-L1 immunohistochemical pathological slices, which effectively segment target regions in PD-L1 immunohistochemical pathological slices ( Effective tissue area), on the basis of regional constraints, the cell key point localization method that effectively mines cell fine-grained features and contextual information is used for cell localization and classification, and negative tumor cell reclassification method is used for secondary screening of negative tumor cells , so as to achieve accurate automatic auxiliary analysis of PD-L1 slices, the results of cell localization and classification using this application are more effective, and the number of negative tumor cells in digital pathological images can be accurately counted, which is important for improving PD-L1 immunohistochemical pathology The precision and accuracy of sliced ​​TPS metrics are more effective.

Description

technical field [0001] The invention relates to the field of pathological image processing, in particular to a method and device for improving the TPS calculation accuracy of PD-L1 immunohistochemical pathological slices. Background technique [0002] Pathological diagnosis is the study of the causes, pathogenesis, morphological structure, functional and metabolic changes of the diseased body and the outcome of the disease during the disease process, so as to provide the necessary theoretical basis and practical basis for the diagnosis, treatment and prevention of the disease. PD-L1 immunohistochemical detection and grading should be carried out in patients with non-small cell lung cancer regardless of adenocarcinoma or squamous cell carcinoma. The so-called grading refers to finding and counting positive tumor cells and negative tumor cells on the whole PD-L1 immunohistochemical pathological section, and accurately calculating the Tumor Proportion Score (TPS). TPS is the r...

AI Technical Summary

Problems solved by technology

The daily workload of doctors has greatly increased

At the same time, a pathological slide usually contains millions of cells, and a pathologist usually needs to analyze multiple digital pathological slides a day, which brings a lot of work intensity to the doctor, and the phenomenon of fatigue reading sometimes occurs

[0005] (2) During slice analysis, there are certain similarities between tumor cells and non-tumor cells, for example, the cell membranes of positive tumor cells and immune-related cells are stained brown, and it is difficult to distinguish them due to the fine-grained features at the cell level , making it extremely difficult for pathologists to interpret cells. Therefore, clinical interpretation of PD-L1 can only be performed after specific training. Even so, since pathologists need to accurately count cells under high-power fields, However, the lack of rich context information in the local field of view leads to false positives and false negatives.

b) positive tumor cell area; c) immune tissue cell area; d) necrosis, mucus area; e) negative tumor cell area; f) interstitial cell area; Compared with non-tumor cells, the practical level is very difficult, and its accuracy is greatly affected when calculating the TPS index

Images