Machine learning-based optimization of photodynamic therapy for human skin affected by malignant melanoma: a Monte Carlo study and 3D renderings

In this work we developed 3D spatially and volumetrically varying melanoma model an investigate light localization therein for effective optimization of photodynamic therapy applications. Utilizing the MC model developed <i>in house</i>, we simulated photon propagation in turbid media, varying parameters like light source geometry, melanoma size, and melanin and blood concentrations. We utilize Machine Learning for the assessment of the functional properties of melanoma-affected human skin and demonstrate its biophysically-based 3D volumetric renderings, highlighting the impact of varying melanin and blood concentrations on visualizations and PDT signal formation.