MRI-based Intratumoral and Peritumoral Habitat Radiomics for Early Prediction of Pathologic Complete Response in...

Area of Science:

• Radiology and Oncology
• Medical Imaging Analysis
• Breast Cancer Research

Background:

• Tumor heterogeneity significantly impacts neoadjuvant therapy (NAT) response in breast cancer.
• Conventional radiomics often fails to capture critical intratumoral heterogeneity.
• Accurate prediction of pathologic complete response (pCR) is vital for treatment optimization.

Purpose of the Study:

• To develop and validate a novel MRI-based habitat radiomics model.
• To noninvasively predict pCR in breast cancer patients undergoing NAT.
• To assess the model's performance against conventional radiomics approaches.

Main Methods:

• Retrospective analysis of 301 breast cancer patients undergoing NAT.
• Generation of habitat subregions using K-means clustering.
• Development and validation of clinical, radiomics, habitat, peritumoral, and combined models using AUC, calibration, and decision curves.

Main Results:

• The habitat radiomics model demonstrated high predictive performance (AUCs ranging from 0.802 to 0.931).
• A combined model integrating peritumoral features and clinical data achieved superior AUCs (up to 0.957).
• SHAP analysis identified intratumoral habitats as key predictors of pCR.

Conclusions:

• MRI-based habitat radiomics effectively captures spatial tumor heterogeneity.
• This noninvasive approach enables accurate prediction of pCR in breast cancer.
• The findings support improved, individualized treatment planning for breast cancer patients.