Modeling of Catheter Microsphere Injection for Patient Specific Y-90 Radioembolization*

Treating cancer patients diagnosed with hepatocellular carcinoma with transarterial radioembolization is increasingly used due to its minimally invasive procedure and sparing of adjacent healthy tissues from radiation exposure. The use of complex physics-based modeling techniques with patient-specific clinical data shows much promise to support strategies for improved tumor targeting through high-precision dosimetry. We have developed a modeling framework, CFDose, that incorporates clinical patient cone-beam Computed Tomography (CBCT) images to predict microsphere transport in the patient liver vasculature using computational fluid dynamics (CFD). Radiation dosimetry is then performed from the predicted microsphere transport. In this work, we focus on improving the accuracy of the CFD modeling by parsing out the various sources of uncertainty in intra-patient geometry and microsphere transport model fidelity. We have improved our model by including monodispersed dilute finite-sized microspheres injected at the catheter site and varying the injection parameters (e.g. particle velocity, catheter curvature). These improvements to the model accuracy could help reduce the uncertainty in patient-specific predictions of the microsphere distribution between liver segments.