MRI-Based Radiomics Model for Classifying Axillary Lymph Node Burden and Disease-Free Survival in Patients With...

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Axillary lymph node (ALN) burden is a critical prognostic factor in breast cancer, influencing diagnosis and treatment.
• Noninvasive assessment of ALN burden can enhance patient stratification and personalize treatment strategies.

Purpose of the Study:

• To evaluate the efficacy of MRI-based radiomics for preoperative classification of ALN burden in early-stage breast cancer.
• To assess survival differences between patients categorized with high and low ALN burden.

Main Methods:

• Retrospective study including 343 breast cancer patients, with data split for training, testing, internal, and external validation.
• Utilized 3T dynamic contrast-enhanced (DCE) MRI sequences.
• Developed clinical, radiomics, and combined machine learning models (including Random Forest) for ALN burden assessment, employing statistical tests like logistic regression, AUC, and Kaplan-Meier analysis.

Main Results:

• The Random Forest-based combined model achieved high Area Under the Curve (AUC) values, ranging from 0.826 to 0.912 across different datasets.
• The model demonstrated good accuracy in classifying ALN burden (up to 0.860).
• Kaplan-Meier analysis indicated a significant difference in disease-free survival (DFS) between high- and low-burden groups (p=0.022, HR=2.9).

Conclusions:

• MRI-based radiomics models show significant potential for the noninvasive evaluation of axillary lymph node burden in breast cancer.
• These models can aid in prognostic stratification and prediction of survival outcomes, supporting individualized patient management.