A proof-of-concept study of the in-vivo validation of a computational fluid dynamics model of personalized radioembolization

Abstract

Radioembolization (RE) with yttrium-90 (90Y) microspheres, a transcatheter intraarterial therapy for patients with liver cancer, can be modeled computationally. The purpose of this work was to correlate the results obtained with this methodology using in vivo data, so that this computational tool could be used for the optimization of the RE procedure. The hepatic artery three-dimensional (3D) hemodynamics and microsphere distribution during RE were modeled for six 90Y-loaded microsphere infusions in three patients with hepatocellular carcinoma using a commercially available computational fluid dynamics (CFD) software package. The model was built based on in vivo data acquired during the pretreatment stage. The results of the simulations were compared with the in vivo distribution assessed by 90Y PET/CT. Specifically, the microsphere distribution predicted was compared with the actual 90Y activity per liver segment with a commercially available 3D-voxel dosimetry software (PLANET Dose, DOSIsoft). The average difference between the CFD-based and the PET/CT-based activity distribution was 2.36 percentage points for Patient 1, 3.51 percentage points for Patient 2 and 2.02 percentage points for Patient 3. These results suggest that CFD simulations may help to predict 90Y-microsphere distribution after RE and could be used to optimize the RE procedure on a patient-specific basis.

CFD Simulations of Radioembolization: A Proof-of-Concept Study on the Impact of the Hepatic Artery Tree Truncation

“…The first observation is that the imposed blood flow split differs from the predicted or calculated microsphere distribution (see Figures 5,7,9 and 11). This finding is not novel, it is indeed observed in previous studies that used a microcatheter to inject the microspheres [8] or the studies that used the particle release maps [7,9,26] as a research tool-particle release maps correlate each point in the injection cross-sectional plane with the computational domain outlet from which microspheres would exit. Even if a blood flow split-matching microsphere distribution can be achieved [5], this cannot be assumed as a general rule.…”

“…The CFD model (type of geometry plus governing equations plus type of boundary conditions) and the simulation strategy (simulation of one cardiac cycle where micro-spheres are injected plus sufficient additional cardiac cycles to ensure that most of the injected microspheres exit the computational domain) used in this study have been recently validated in vivo in a proof-of-concept study, where the CFD model was defined using pre-RE imaging techniques and the simulated segment-to-segment microsphere distribution was compared with the measured segment-to-segment activity distribution taken from post-RE imaging techniques [26].…”

Microspheres Used in Liver Radioembolization: From Conception to Clinical Effects

“…Investigations of the microscopic depositions of microspheres can be investigated in vivo by imaging (SPECT, PET and MRI) or simulated in vitro by computational or artificial models of the hepatic arterial system. An excellent agreement was recently demonstrated between the microscopic deposition of microspheres simulated with a computational model and the real distribution observed in patients with 90 Y PET/CT [ 37 ], especially with time- of-flight (TOF)-PET systems [ 38 , 39 ].…”

Computational Fluid Dynamics Modeling of Liver Radioembolization: A Review

“…These and other developments could be explored using computational techniques, i.e., using computer simulations. For example, a recent study has confirmed that computer simulations of hepatic artery hemodynamics and radioactive microsphere transport can predict the segment-to-segment microsphere distribution [ 10 ].…”