Functional Diffusion Mapping in Weekly SPLICE-DWI Data for Outcome Prediction in Head and Neck Squamous Cell Carcinomas

Motivation: Head and neck cancer outcomes vary widely with radiation therapy (RT). Changes in ADC from DWI during RT might predict response, but the optimal time points and metrics are unclear. Goal(s): We tested full-tumor ADC and voxel-wise Functional Diffusion Mapping (fDM) metrics as predictors of relapse-free survival (RFS) at 6- and 24-months post-RT. Approach: ADC and fDM metrics were analyzed at multiple RT time points for 60 patients. Logistic regression modeling was used to identify the strongest predictors of RFS. Results: ADC metrics predicted 6-month RFS more effectively than 24-month, with fDM metrics showing potential as indicators of early RT response. Impact: This study is the first to apply Functional Diffusion Mapping (fDM) in head and neck cancer, revealing its potential to improve adaptive radiation therapy by identifying early predictors of response, which may guide personalized treatment adjustments for better outcomes.