Assessing the Precision of Automated Detection and Diameter Estimation of Cerebral Arteries in the Circle of Willis
Julia Huck1, Davy Vanderweyen2, Tatjana Rundek4, Mitchell Elkind5, Jose Gutierrez5, Maxime Descoteaux3, Kevin Whittingstall2
1Department of Nuclear Medicine and Radiobiology, 2Department of diagnostic radiology, 3Department of Computer Science, Université de Sherbrooke, 4Evelyn F. McKnight Brain Institute and Department of Neurology, Miller School of Medicine, University of Miami, 5Department of Neurology, Columbia University
Objective:
To compare artery diameter estimations from Express IntraCranial Arteries Breakdown (eICAB) against manual measurements to minimize human error and variability, thereby providing a consistent and reliable tool for clinical and research applications.
Background:
The brain relies on arterial blood from four major extracranial arteries via the Circle of Willis (CW), which can be visualized through Magnetic Resonance Angiography (MRA). Accurate artery identification within the CW is vital for detecting anatomical variations, stenosis, aneurysms, and other pathologies that can significantly impact cerebral circulation. To alleviate time-consuming manual CW assessment, this study evaluates eICAB, a recently developed open-source software method for automatically detecting and quantify CW in non-contrast enhanced MRA.
Design/Methods:
631 MRA images of stroke and intracranial stenosis-free individuals from The Northern Manhattan Study (NOMAS) study were analyzed. eICAB-derived estimates of luminal diameter in the Internal Carotid Artery (ICA), Basilar Artery (BA), Anterior Cerebral Arteries (ACA), Middle Cerebral Arteries (MCA), and Posterior Cerebral Arteries (PCA) were compared to manual measurements made by author JG. T-tests were performed to determine CW arteries with diameter discrepancies within 0.5 and 1mm.
Results:
Across arteries, eICAB diameter estimates of the MCA, ACA and PCA were consistently within 0.5mm of manual measurements (p<0.01, Bonferroni corrected). BA and ICA errors were slightly higher, but remained within 1mm (p<0.01, Bonferroni corrected).
Conclusions:
Our open-source, fully automated method for analyzing the CW from MRA imaging yields highly accurate diameter estimates of the major cerebral arteries. This may increase researchers and clinicians’ capacity to reliably analyze and detect cerebrovascular anomalies (e.g. luminal narrowing, dilation) in large databases.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.