Multiscale modeling of microsphere distribution in liver radioembolization.

Liver transarterial radioembolization is an effective method to treat inoperable primary or secondary liver cancers. It consists of infusing radiolabeled microspheres from a microcatheter placed on the hepatic artery. The microspheres are transported through the bloodstream ideally targeting only the tumor tissue embolizing it and releasing locally a tumoricidal radiation dose. A pretreatment is always done with macroaggregated albumin microparticles to observe the microsphere distribution before the treatment, but it is mandatory to replicate the exact conditions to ensure the pretreatment is an appropriate indicator to the treatment outcome. Bioengineering tools like Computational Fluid Dynamics or in vitro experiments are widely used to help with the radioembolization planning. The aim of this thesis is to analyze the microsphere distribution in the macro and micro scale of the hepatic artery using those bioengineering tools. CFD simulations show to be great surrogates of the treatment, and they could be used to find optimal treatment conditions.This requires several simulations under different operating conditions to choose the configuration that provides the desired results. However, each simulation lasts 1-2 days, being excessive to provide useful answers to the medical team. During this thesis the computational model was effectively optimized thanks (i) to the geometry truncation rule developed to reduce de computational domain without losing microsphere distribution prediction accuracy; (ii) to the analysis of the validity of considering blood as a Newtonian fluid, reducing the number of equations to be solved by the CFD model; and (iii) to the study of the convergence criterion based on the residuals absolute value reducing these values from 10-5 to 10-2 . After these optimizations, the simulations are 93 % less time consuming (2 h/simulation), being possible to develop a tool to simulate several working conditions in a single day. Besides, the influence of the microsphere concentration on their macroscale distribution was demonstrated using in silico models. It was observed and quantified the relation between the changes on the treatment vial concentration, and segmental microsphere concentration variation concluding that a direct relation happens between them. The possibility of modifying the microsphere concentration in blood by changing the infusion velocity was also demonstrated. Next, an in vitro microfluidic device and test procedure were developed to study how the blood flow rate and the microsphere concentration changes the microsphere behavior on the tissue deposition in terms of distal penetration and spatial distribution. It was observed that increasing the concentration causes a greater tumor target and more proximal occlusions but with an increase on the distal penetration. An increase in the blood flow rate also involves an increase in the tumor target and distal penetration. By joining both models, the whole hepatic artery tree could be modelized obtaining the treatment dosimetry and infusion conditions that provide the desired intervention’s outcome. Finally, concerning the balloon occluded intervention, a patient-inspired in vitro model was developed for the first time to study the complex phenomenon related to these interventions as the pressure gradient effect or the collateral vessels influence. An in silico model was also developed to replicate the experiments with an error of less than 4 percentage points and using this model the influence of the collateral vessel on the treatment efficacy was analyzed. The pressure gradient effect was successfully generated in both models, but the designed collateral vessel proved to be unsuitable to study their influence on the treatment by barely influencing the experiments.