Enhancing the Prediction of Axillary Lymph Node Metastasis in Breast Cancer through Habitat-Based Radiomics and...

Area of Science:

• Medical Imaging Analysis
• Machine Learning in Oncology
• Breast Cancer Diagnostics

Background:

• Accurate prediction of axillary lymph node metastasis (ALNM) is crucial for breast cancer staging and treatment planning.
• Conventional imaging methods have limitations in fully characterizing tumor heterogeneity and predicting ALNM.

Purpose of the Study:

• To develop and evaluate a machine learning model integrating habitat-based radiomics from B-mode and contrast-enhanced ultrasound (CEUS) images with voting algorithms.
• To predict axillary lymph node metastasis (ALNM) in breast cancer patients.

Main Methods:

• Retrospective analysis of 246 breast cancer patients' ultrasound images.
• Extraction of radiomics features from whole-tumor and subregional regions using B-mode and CEUS.
• Development of a voting algorithm to integrate habitat-based radiomics, traditional radiomics, and clinical data.

Main Results:

• The combined habitat-based model achieved 87.76% accuracy in predicting ALNM on the testing set, outperforming the traditional model (79.59% accuracy).
• Habitat-based radiomics demonstrated superior ability in capturing tumor heterogeneity compared to conventional approaches.
• The integration of dual-modality ultrasound and voting algorithms significantly improved prediction performance (p < 0.05).

Conclusions:

• Habitat-based radiomics offers a more effective approach to characterizing tumor heterogeneity in breast cancer.
• Dual-modality ultrasound combined with voting algorithms provides a robust method for ALNM prediction.
• This approach lays a reliable foundation for computer-aided preoperative planning in breast cancer management.