Head and Neck Cancer Tumor Segmentation Using Support Vector Machine in Dynamic Contrast-Enhanced MRI

Area of Science:

• Medical Imaging
• Machine Learning
• Oncology

Background:

• Head and neck cancer (HNC) segmentation from medical images is crucial for diagnosis and treatment planning.
• Accurate tumor delineation remains a challenge in clinical practice.

Purpose of the Study:

• To develop and evaluate an automated segmentation method for HNC tumor lesions.
• To leverage Support Vector Machine (SVM) and Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) for improved segmentation accuracy.

Main Methods:

• Collected 120 DCE-MRI samples for head and neck cancer.
• Extracted five curve features and two principal components from normalized time-intensity curves (TIC).
• Trained three SVM classifiers on 80 samples and tested on 40 samples, evaluating performance using Area Overlap Measure (AOM), Contrast Ratio (CR), and Percent Match (PM).

Main Results:

• The proposed SVM-based method achieved an average AOM of 0.76 ± 0.08 on the testing dataset.
• Mean CR and PM were 79 ± 9% and 86 ± 8%, respectively.
• Demonstrated superior segmentation accuracy compared to previous studies in HNC segmentation.

Conclusions:

• The developed automated segmentation method shows significant potential for clinical application in head and neck cancer.
• Improved segmentation performance can aid in more precise diagnosis and treatment of HNC.