Prospective Validation and Application of an Artificial Intelligence-based Model for Evaluating the Efficacy of Breast Cancer Patients After Neoadjuvant Therapy

Overview

Breast cancer has become the world's number one cancer. While its therapeutic efficacy is increasing, how to achieve non-invasive evaluation of the efficacy of neoadjuvant therapy (NAT) for breast cancer patients and thus avoid surgery has become a bottleneck problem that needs to be broken through in clinical diagnosis and treatment. Existing non-invasive evaluation strategies are limited to single-center, single-modality modeling, and have problems such as low performance and poor versatility. Therefore, in the early stage of this study, multi-modality breast cancer patient data from multiple centers across the country were collected and the establishment of an artificial intelligence (AI) efficacy prediction model was preliminarily completed. On this basis, this project intends to further improve the multi-center prospective validation study of the prediction model. The research results will help solve the scientific problem of non-invasive judgment of NAT efficacy in breast cancer patients and provide a new paradigm for the research of high-performance AI diagnosis and treatment auxiliary systems applicable to multiple centers.

(1) Prospectively collect breast MRI original images (DCE and ADC sequences) and corresponding clinical and surgical pathological data of multi-center breast cancer patients before and after neoadjuvant treatment, store and transport them via mobile hard disks, and input the processed data into the established efficacy determination model stored in a dedicated cloud server; (2) Use artificial intelligence to automatically delineate the ROI area and extract the imaging genomics and deep learning features therein, and combine the clinical pathological characteristics of the patients to further prospectively verify the effectiveness of the established pCR efficacy determination model.

Conditions