Simulation of superselective catheterization for cerebrovascular lesions using a virtual injection software

This report addresses the feasibility of virtual injection software based on contrast-enhanced cone-beam CTs (CBCTs) in the context of cerebrovascular lesion embolization. Intracranial arteriovenous malformation (AVM), dural arteriovenous fistula (AVF) and mycotic aneurysm embolization cases with CBCTs performed between 2013 and 2020 were retrospectively reviewed. Cerebrovascular lesions were reviewed by 2 neurointerventionalists using a dedicated virtual injection software (EmboASSIST, GE Healthcare; Chicago, IL). Points of Interest (POIs) surrounding the vascular lesions were first identified. The software then automatically displayed POI-associated vascular traces from vessel roots to selected POIs. Vascular segments and reason for POI identification were recorded. Using 2D multiplanar reconstructions from CBCTs, the accuracy of vascular traces was assessed. Clinical utility metrics were recorded on a 3-point Likert scale from 1 (no benefit) to 3 (very beneficial).

Nine cases (7 AVM, 1 AVF, 1 mycotic aneurysm) were reviewed, with 26 POIs selected. Three POIs were in 2nd order segments, 8 POIs in 3rd order segments and 15 POIs in 4th order segments of their respective arteries. The reviewers rated all 26 POI traces - involving a total of 90 vascular segments - as accurate. The average utility score across the 8 questions were 2.7 and 2.8 respectively from each reviewer, acknowledging the software's potential benefit in cerebrovascular embolization procedural planning.

The operators considered CBCT-based virtual injection software clinically useful and accurate in guiding and planning cerebrovascular lesion embolization in this retrospective review. Future prospective studies in larger cohorts are warranted for validation of this modality.