Multimodal Spatiotemporal Deep Learning Framework to Predict Response of Breast Cancer to Neoadjuvant Systemic...

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Neoadjuvant systemic therapy (NST) is a key breast cancer treatment.
• Pathologic complete response (pCR) is a critical efficacy measure for NST.
• Early prediction of pCR is vital for treatment optimization and patient survival.

Purpose of the Study:

• To develop an automated framework for early prediction of pCR in breast cancer patients undergoing NST.
• To improve treatment efficacy and reduce toxicity by accurately identifying responders.

Main Methods:

• Proposed an end-to-end multimodal spatiotemporal deep learning framework (deep-NST).
• Integrated imaging, molecular, and demographic data for prediction.
• Validated the framework on the ISPY-1 dataset.

Main Results:

• The deep-NST framework demonstrated superior performance in predicting pCR.
• Achieved high accuracy and Area Under the Curve (AUC).
• Ablation studies confirmed the effectiveness of individual framework components.

Conclusions:

• The deep-NST framework offers a precise and efficient method for early pCR prediction in NST.
• This tool can aid in personalized treatment decisions and improve breast cancer patient outcomes.