Cluster analysis of quantitative ultrafast DCE-MRI for DCIS grading and upgrading without tracing lesion region-of-interest

Motivation: Precise grading of ductal carcinoma in situ (DCIS) breast cancer is crucial for selecting the most effective treatment approach and forecasting patient outcomes. Goal(s): This study assesses whether k-means clustering analysis of kinetic ultrafast dynamic-contrast-enhanced MRI (DCE-MRI) could differentiate DCIS grades in 72 patients. Approach: Using k-means clustering (K=5), DCIS lesions were effectively separated from normal tissue. Results: Key kinetic parameters ($$$\alpha$$$, $$$A\cdot\alpha$$$, $$$AUC30$$$) were significantly higher in patients with DCIS and invasive cancer. $$$AUC30$$$ also correlated with DCIS grade, with higher values in high-grade cases. This method could automatically segment DCIS to identify aggressive DCIS and guide treatment strategies. Impact: K-means clustering analysis of ultrafast DCE-MRI can help identify DCIS, differentiate between low- and high-grade DCIS and identify invasive potential, and facilitate personalized treatment by guiding decisions on aggressive treatment versus surveillance for breast cancer patients.