11 - Image-Based Metrics for Characterization of Arterial Embolization Endpoints Using an In Vitro 3D-Printed Hepatic Model

Currently determination of embolization endpoints for transarterial procedures  relies on subjective visual cues (beats of stasis) to interpret underlying hemodynamic phenomena (partial, sub-statis, and stasis). This assessment can vary between patients, imaging systems and operators. What is lacking are quantitative and standardized methods for assessing the consistency and the endpoints of embolization. In this work we investigate the use of image-based measurements to serve as quantitative metrics to characterize vessel occlusion and as embolization endpoints.


Materials and Methods:

We conducted controlled experiments using a benchtop flow system {1} including a vascular phantom, a hydraulic circuit, pressure/flow sensors and an optical camera. The 3D printed hepatic arterial phantom was designed based on CT scans, with artery lumen values from literature {2}. A peristaltic and a continuous pump reproduced systolic and diastolic blood pressure, monitored by pressure sensors. Needle valves were used to adjust the flow, measured with flow meters, to match normal hepatic flow rates {3}. A solution of dark food dye, injected via microcatheter using an automatic syringe pump, was used to mimic a contrast agent in the optical range.

In the experiments, the lighting, the pumps’ flow rate, the pressure and the injection parameters were kept constant, while the flow rate in the downstream arterial branch was varied to simulate 10 occlusion levels, from 75% to 99.3%. For each level, five injections were video recorded. DSA-like images were generated by background subtraction during video post-processing. To characterize the time intensity curves derived from injections at different occlusion rates, we measured the overall contrast enhancement given by the injected solution over time (using the Area Under the Curve or AUC) and the rate of contrast wash out (α).



Results:

Using an in vitro model simulating the hemodynamic conditions of embolization of hepatic arteries we demonstrated the correlation between image-based metrics and induced occlusion rate. The measured AUC and α data show a replicable trend as a function of the occlusion: the values remain low and stable up to approximately 90% occlusion but increase rapidly as occlusion approaches 100%.



Conclusion:

The results demonstrated the potential of using DSA-derived metrics to quantify embolization endpoints. Future work will validate these metrics in more realistic and clinically relevant models with expanding the study to X-ray imaging.