Contrast-Enhanced Mammography-Based Radiomics for Predicting Ductal Carcinoma In Situ in Breast Cancer

Area of Science:

• Radiology
• Oncology
• Medical Imaging Analysis

Background:

• Ductal carcinoma in situ (DCIS) is a non-invasive breast cancer precursor.
• Accurate prediction of DCIS is crucial for appropriate patient management and treatment.
• Contrast-enhanced mammography (CEM) offers improved visualization of breast lesions.

Purpose of the Study:

• To develop and validate a radiomics nomogram for predicting DCIS using CEM and clinical factors.
• To assess the predictive performance of the nomogram in distinguishing DCIS from invasive breast cancer.
• To identify key radiomics features and clinical variables contributing to DCIS prediction.

Main Methods:

• Retrospective analysis of 731 breast cancer cases from five centers who underwent CEM.
• Extraction and selection of radiomics features from CEM images using mRMR and LASSO methods.
• Development of a radiomics nomogram integrating radiomics signature (Rad-score) and clinical factors (age, menstrual status, BPE) via logistic regression.

Main Results:

• A radiomics nomogram was developed using 11 radiomics features, age, menstrual status, and background parenchymal enhancement (BPE).
• The nomogram demonstrated strong predictive performance with AUCs of 0.889 (internal) and 0.822 (external test sets).
• Calibration curves showed excellent agreement between predicted and observed probabilities, indicating reliable predictions.

Conclusions:

• The developed radiomics nomogram incorporating CEM features, age, menstrual status, and BPE shows promising performance in predicting DCIS.
• This tool could aid in the non-invasive diagnosis and management of DCIS.
• Further validation in larger, multi-center external cohorts is warranted to confirm these preliminary findings.