Lung cancer pathological type diagnosis method based on computer vision and CT images

Abstract

The invention relates to a lung cancer pathological type diagnosis method based on computer vision and CT images, and relates to the fields of image processing, medical big data and computer vision. The method comprises the following steps of 1) collecting lung cancer CT medical images, and constructing a training set; 2) processing the images in the training set to generate required training samples; 3) using the training samples for training a lung cancer pathological type diagnosis model based on the CT images; 4) acquiring new lung cancer CT images, and constructing a verification set; 5)using the verification set to verify the diagnostic model. The method solves the problems that existing lung cancer pathological type diagnosis is invasive, easily influenced by specimen sampling, long in diagnosis time and low in efficiency. According to the method, the CT influence is used for pathological type diagnosis, the whole process is rapid, efficient and invasive, and the pathological diagnosis is efficient and noninvasive.

Description

technical field [0001] The invention relates to the field of intelligent diagnosis of lung cancer, in particular to a method for diagnosing pathological types of lung cancer from CT images based on computer vision technology, which belongs to the field of artificial intelligence. Background technique [0002] Lung cancer accounts for more than a quarter of all cancer deaths and is one of the leading threats to human health for men and women worldwide. The pathological types of lung cancer are divided into small cell carcinoma and non-small cell carcinoma, among which non-small cell carcinoma mainly includes squamous cell carcinoma, large cell carcinoma and adenocarcinoma. The proliferation and expansion of different pathological types of lung cancer are completely different, and the treatment measures are also different. Diagnosis is the premise of treatment. Only timely diagnosis can implement effective treatment plan and prolong the survival of patients. Early identifica...

AI Technical Summary

Problems solved by technology

[0003] At present, histopathology and molecular biology are the gold standard for pathological diagnosis of tumors, but they can usually only be performed on tissue samples such as surgical resection or needle biopsy. Obtaining tissue samples is not only invasive, but also easily affected by sample selection. Does not reflect the overall condition of the tumor tissue

Images