Breast tissue classification and identification method based on transfer learning

Abstract

The invention provides a breast tissue classification and identification method based on transfer learning. The method comprises the following steps: firstly, obtaining a suspected patient breast tissue slice pathological image through a tissue slice, then carrying out format adjustment and a series of image processing to obtain a breast tissue slice preprocessed pathological image, and manually marking the preprocessed image pathological category; constructing a transfer learning breast tissue classification model, sequentially inputting the preprocessed images into the model, predicting a pathological category through the model, constructing a loss function model in combination with an actual pathological category of the images, and performing optimization training to obtain an optimized transfer learning breast tissue classification model; and predicting the pathological category of the suspected patient through the optimized model, and combining the pathological category with a heat map generator to obtain a thermodynamic diagram corresponding to the evidence region. The invention has the advantages that the method is suitable for an application environment with a small data set, the robustness is enhanced in training, and the credibility of a result is improved in prediction.

Description

technical field [0001] The invention belongs to the field of medical image data processing, and in particular relates to a breast tissue classification and recognition method based on migration learning. Background technique [0002] The global incidence of breast cancer has been on the rise since the late 1970s. Although China is not a country with a high incidence of breast cancer, the situation is still not optimistic. In the past decade, breast cancer has gradually occupied the first place in the incidence of female cancer. According to the metastasis of cancer tissue to surrounding tissues, breast cancer is mainly divided into carcinoma in situ (CIS) where cancer cells have not spread and invasive carcinoma (CIS) where cancer cells have spread. Therefore, if we can accurately classify and identify breast tissue, it will be of great help in judging whether we have breast cancer and the type of breast cancer, and then we can achieve early diagnosis and correct treatment a...

AI Technical Summary

Problems solved by technology

However, these methods have a narrow scope of application, poor transferability, and have a strong dependence on the amount of data, so they cannot play an advantage in the application environment with small data sets.

In the process of medical image processing, obtaining a database of large samples requires extremely high time and economic costs.

In addition, these methods cannot point out the basis of the results obtained, and can only give the predicted value, which is not convincing, so these machine detection methods also have great limitations

Images