Predictive modeling of chronic foot ulcer outcomes using longitudinal photoacoustic imaging.

This study reports the first clinical longitudinal photoacoustic imaging (PAI) of chronic foot ulcers, a major complication in patients with peripheral vascular disorders. Compared to traditional methods such as ABI or near-infrared spectroscopy, the photoacoustic imaging approach provides non-invasive, high-resolution, and quantitative monitoring of vascular dynamics over time. Our system provided dorsal-side imaging of vascular structures with an expanded field of view and incorporated a skin artifact suppression algorithm to improve visualization of subdermal vasculature. From the acquired 2D and 3D images, we extracted a set of 45 quantitative features, representing signal intensity, texture complexity, and morphological changes associated with ulcer progression. Using a LASSO-based feature selection strategy, we identified the top-12 feature subset and validated them through multi-seed cross-validation. Our selection achieved an average classification accuracy of 79.6% and a macro-averaged AUC of 86.6% in distinguishing healing, worsening, and healthy cases. These findings demonstrate the clinical utility of photoacoustic biomarkers for personalized ulcer tracking and risk stratification.