Comsol-Based Reactive Oxygen Species Modeling with Deformable Image Registration for Organ-at-Risk Dosimetry In Pleural Photodynamic Therapy
Abstract
Purpose
This study develops an integrated framework combining three-dimensional deformable image registration with reactive oxygen species ([ROS]rx) modeling for comprehensive dosimetry in pleural photodynamic therapy (PDT). We aim to address clinical measurement gaps by enabling dose distribution mapping across patient-specific pleural cavity geometries and organs at risk (OARs).
Methods
Navigation data from 10 clinical cases were processed to reconstruct patient-specific pleural cavity geometries in standardized coordinates. A dynamic iterative boundary matching algorithm was developed for deformable registration between intraoperative navigation volumes and pre-treatment CT imaging, enabling transformation of OAR contours (lung, heart, esophagus) into the treatment coordinate system. Photofrin concentrations measured at 8 anatomical sites were interpolated across cavity surfaces using inverse distance weighting (IDW). Light fluence distributions from navigation tracking were combined with heterogeneous drug maps as inputs to macroscopic kinetic equations implemented in COMSOL Multiphysics for [ROS]rx calculation.
Results
The deformable registration successfully transformed CT-derived organ contours to match intraoperative cavity geometries. COMSOL simulations across 10 patients yielded volume-averaged [ROS]rx values ranging from 0.16 to 0.93 mM. Comparison with clinical point measurements showed differences of -11.4% to +40.3% versus calculated oxygen data (mean clinical [ROS]rx,calc = 0.48±0.23 mM) and -28.7% to +40.7% versus measured oxygen data (mean clinical [ROS]rx,meas = 0.56±0.26 mM). Point-to-point comparison between COMSOL and clinical results were performed for 3 representative cases, with difference less than 10 %. The integrated workflow enabled visualization of [ROS]rxdistributions on OAR surfaces, with values ranging from 0.596 to 0.655 mM on cardiac and pulmonary structures.
Conclusion
This integrated deformable registration and [ROS]rx modeling framework provides 3D dose distributions where only discrete clinical measurements exist. The ability to map dose onto OARs represents a significant advance toward treatment planning optimization and real-time dose feedback for pleural PDT.