Managing Radiotherapy-Induced Side Effects with Artificial Intelligence: Early Detection of Muscle Atrophy and Personalized Nutritional Support in Head and Neck Cancer

Abstract

This project is based on a crucial step in medical image processing: segmentation. Using a trained U-Net model, we segmented two key muscles the masseter and the medial pterygoid located in the neck region of head and neck cancer patients following radiotherapy. Thanks to satisfactory segmentation results, we were able to calculate the surface area of these muscles before and after treatment, allowing us to detect potential muscle atrophy in that region. All of this was integrated into a simple, fast, and well-structured interface, designed for objective and efficient use. The main goal of this project is to better manage the side effects of radiotherapy by offering personalized nutritional recommendations, improving patient follow-up and overall quality of life.