Computational Methods for Measuring and Visualizing Vasospasm

Introduction: Vasospasm is a dreaded sequela of subarachnoid hemorrhage (SAH) and can be the source of significant associated morbidity and mortality. Angiographic vasospasm following SAHs is thought to approach 50%. Current treatments involve hyperdynamic therapy utilizing hypertension and hypervolemia, angioplasty, and intra-arterial treatment of verapamil reserved for refractory cases. As the goals of therapy are improve brain perfusion, it would be useful to have real time assessment of the efficacy of these treatments. Here, we describe a technique for real-time assessment of changes to perfusion during interventional angiography.

Methods: We present 50 cases of clinical vasospasm and evaluate how novel perfusion analytics and image segmentation can assist neurosurgeons and neurointensivists with real-time care planning. We developed software capable of evaluating changes in contrast flow dynamics from digital subtraction angiography (DSA), namely, performing signal intensity curve deconvolution on a voxel by voxel basis to provide quantitative 2D perfusion metrics that include: time to peak, time to drain, area under the curve, root mean transit time, arrival time, tissue concentration, and cerebral blood flow at each voxel of a study in real-time. Furthermore, we are able to display changes to perfusion across different studies of the same vascular territory and compare perfusion to contralateral hemispheres to assist the provider in determining the optimal therapeutic intervention.

Data: We retrospectively reviewed 50 anonymized cases of vasospasm treated with intra-arterial verapamil (4 vertebral distributions and 46 internal carotids). We present pre- and postinterventional DSA imaging demonstrating the initial burden of disease and the changes in perfusion post intra-arterial intervention.

Results: With DSA perfusion analysis, we are able to segment areas of hypoperfusion representing parenchyma at risk of ischemia, analyze the histograms of tissue for arrival time, time-to-peak, mean transit time, tissue concentration and associated cerebral blood flow and blood volume to quantify changes in perfusion, we are able to quantify exact changes to perfusion for each individual case. Processing time for these metrics is, on average, 400 milliseconds, allowing for results to be returned to interventionalists in real time.

Discussion: This software demonstrates the utility of quantifying perfusion parameters for patients in radiographic vasospasm. This data are useful for interventionalists during reperfusion procedures by demonstrating exact vascular territories with perfusion deficits. Quantitative thresholds and analysis based on DSA perfusion may assist with real-time quantification of intra-arterial dosage requirements and help predict response to treatment; however, future prospective analysis is required for validation.

No conflict of interest has been declared by the author(s).