Whole-Tumor ADC Texture Analysis Is Able to Predict Breast Cancer Receptor Status

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Breast cancer comprises diverse molecular subtypes impacting prognosis and treatment.
• Estrogen receptor (ER) and progesterone receptor (PR) status are critical for treatment decisions.
• Accurate prediction of ER/PR status is essential for personalized breast cancer therapy.

Purpose of the Study:

• To evaluate the efficacy of apparent diffusion coefficient (ADC)-based radiomics for differentiating ER/PR positive from ER/PR negative breast cancer.
• To assess the performance of a combined radiomics and clinical data model in predicting breast cancer hormonal status.

Main Methods:

• Retrospective analysis of 185 breast cancer patients, augmented with 25 SMOTE patients.
• Whole-volume tumor segmentation for extraction of first-order radiomic features from ADC maps.
• Development and validation of radiomics models, including a combined model with ki67 and histological grade.

Main Results:

• The ADC-based radiomics model achieved an AUC of 0.81 in the training cohort and 0.93 in the validation cohort for differentiating ER/PR status.
• A combined model incorporating radiomics, ki67, and histological grade yielded a higher AUC of 0.93 in both training and validation groups.
• Whole-volume ADC texture analysis demonstrated significant predictive capability for breast cancer hormonal status.

Conclusions:

• Whole-volume ADC texture analysis is a valuable non-invasive tool for predicting ER/PR status in breast cancer.
• Combining radiomics with clinical parameters like ki67 and histological grade further enhances the accuracy of hormonal status prediction.
• These findings support the potential of radiomics in guiding personalized treatment strategies for breast cancer patients.