Artificial Intelligence in Functional Imaging for Individualized Treatment of Head and Neck Squamous Cell Carcinoma Patients

The two curative treatment modalities for patients with HNSCC - primary chemoradiation (CRT) or primary surgery (often combined with postoperative (C)RT) - are both associated with serious side effects for which reason further stratification, optimization and personalization of treatment is urgently needed. As novel quantitative image analyses are a promising tool for further risk stratification, the investigators training a three-dimensional Convolutional Neural Network on 18F-Fluorodesoxyglucose (FDG) positron emission tomography (PET) imaging and clinical / histopathological data of a multicentric, retrospective cohort of 1200 patients treated with primary CRT and 800 patients treated with primary surgery at Charité and cooperation institutes in order to predict individual treatment-specific outcomes and identify patients with excellent outcome after primary CRT or primary surgery or unfavorable outcome for both. The trained algorithm of the artificial intelligence will be validated in a prospective trial to see if predicted loco-regional control and recommended treatment strategies are reliable. In total 250 curative HNSCC patients, treated with CRT or primary surgery, will be enrolled on this prospective validation trial with observational character, while biomarker, clinical and FDG-PET data are collected from these patients and follow-up visits will be concluded.