Prediction of Neoadjuvant Therapy Efficacy and Prognosis for Breast Cancer Based on Multimodal Data

Prediction of Neoadjuvant Therapy Efficacy and Prognosis for Breast Cancer Based on Multimodal Data: A Multicenter Retrospective and Prospective Validation

This study aims to develop a multimodal deep learning model integrating MRI, ultrasound, digital pathology and clinical information based on multicenter retrospective data. To externally validate the model in an independent prospective cohort, and evaluate its accuracy in predicting pathological complete response (pCR), 3-year and 5-year disease-free survival (DFS). To establish visual tools such as nomograms, assisting clinicians in identifying patients with chemoresistance and facilitating individualized de-escalation or escalation treatment strategies.

Inclusion Criteria:

1. Histopathologically confirmed invasive breast cancer;
2. Planned to receive a full course of neoadjuvant therapy;
3. Complete baseline imaging data (MRI/ultrasound/mammography) and core needle pathology results available.

Exclusion Criteria:

1. Previous history of ipsilateral breast cancer or chest radiotherapy;
2. Distant metastasis (Stage IV);
3. Poor image quality or missing clinical data exceeding 20%.