Harvard MD Anderson Collaborative to Reduce LyMphatic MOrbidity in Head and Neck Cancer with Artificial Intelligence (HARMONiC-AI) - PROJECT SUMMARY Head and neck cancers (HNCs) afflict over 65,000 Americans annually, with survivors often battling lingering complications from treatments like surgery, radiation, and chemotherapy. Notably, lymphedema and fibrosis emerge as prevalent post-treatment issues. Post-radiation inflammation triggers a cascade—from potentially reversible soft tissue edema and lymphedema to, more worryingly, permanent fibrosis. About 75% of radiated patients manifest lymphedema signs within three months, and between 30-50% evolve into moderate-to-severe neck fibrosis. This progression severely impacts functions like swallowing, speech, and neck movement. The diagnostic landscape currently leans heavily on manual and endoscopic assessments, which are inherently subjective and typically catch complications at advanced stages, often when fibrosis is already entrenched. However, routine head and neck CT scans herald promise for early lymphedema detection. Preliminary research points to CT indicators, such as CTLEFAT, as potential lymphedema markers. Yet, widespread clinical adoption remains elusive, primarily due to measurement time and specialized expertise requirements. Our team, harnessing computational imaging and AI, has pioneered CT-based auto-segmentation of cancer lymphatics and soft tissue structures. We posit that AI can harness pre-treatment data to tailor treatment plans, minimizing post-treatment lymphedema (Aim 1). Moreover, we propose that AI-enhanced CT tools can revolutionize lymphedema diagnosis (Aim 2) and risk assessment (Aim 3), offering precise therapeutic interventions. By anticipating and addressing the inflammation-to-edema-to-fibrosis sequence, this approach seeks to radically improve HNC patients' post-treatment quality of life.
