Abstract Details

Title Automatic Mild Intracranial Stenosis Detection in the Circle of Willis from Time-of-Flight MRA

Topic Cerebrovascular Disease and Interventional Neurology

Presentation(s) S2 - Advances in Stroke Imaging and Biomarkers (2:24 PM-2:36 PM)

Poster/Presentation
Number 007

Objective To develop an automatic method for detecting mild intracranial stenosis in the Cirlce of Willis from Time-of-Flight MRA (TOF-MRA)

Background Asymptomatic mild intracranial stenosis (MIS) is defined as 20-50% narrowing of brain arteries, including those in the Circle of Willis (CW). Recent studies show that MIS increases the risk of progression from mild cognitive impairment to Alzheimer’s disease and accelerates brain atrophy. However, MIS is often underdiagnosed in clinical settings, leading to missed opportunities for early intervention that could delay cognitive decline and neurodegeneration. Current software-based detection methods have limited sensitivity in detecting MIS, partly due to challenges in reconstructing tortuous arterial segments (e.g., kissing vessels).

Design/Methods We tested our approach using 1.5T TOF-MRA images from the NOMAS dataset, comprising 1052 participants aged 62 to 80 years, with 1157 MIS cases manually identified in anterior cerebral arteries (ACAs), internal carotid arteries (ICAs), middle cerebral arteries (MCAs), and posterior cerebral arteries (PCAs) by a vascular neurologist. Intracranial arteries were automatically segmented, and kissing vessel artifacts were addressed using in-house algorithms. Arterial diameters were measured along the centerline extracted from a 3D surface method. We calculated stenosis ratios by comparing these measurements to distal (Metric 1), proximal (Metric 2), and average diameters (Metric 3) of each artery. Accuracy was defined as the proportion of MIS cases evaluated by the neurologist that had a stenosis ratio of 20-50% from our method.

Results Metric 3 achieved the highest accuracy (89.6%), significantly outperforming Metric 1 (51%) and Metric 2 (46%). Metric 1 performed best in ACAs (92.5%), followed by ICAs (89%), PCAs (88.5%), and MCAs (85.7%). The method effectively addressed kissing vessels in 180 out of 192 ICAs, enhancing stenosis detection accuracy.

Conclusions Our method effectively detects MIS, particularly in the CW anterior circulation, by incorporating our approach to kissing vessels and average arterial diameters in stenosis ratio calculations.

Authors/Disclosures
Chan Nam Nguyen PRESENTER	Mr. Nguyen has nothing to disclose.
Julia Huck, PhD	Ms. Huck has received research support from Fonds de. recherche du Québec – Nature et technologie (FRQNT).
Jose Gutierrez, MD (Columbia University)	Dr. Gutierrez has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Cardiovascular Research Foundation. Dr. Gutierrez has received personal compensation in the range of $500-$4,999 for serving as an Expert Witness for White and Rusell. Dr. Gutierrez has received research support from NIH. Dr. Gutierrez has received publishing royalties from a publication relating to health care. Dr. Gutierrez has received publishing royalties from a publication relating to health care.
Kevin Whittingstall, PhD	Prof. Whittingstall has nothing to disclose.

��ɫ��

��ɫ��