Automated intracranial vessel labeling with learning boosted by vessel connectivity, radii and spatial context
Jannik Sobisch1,2, Žiga Bizjak1,2, Aichi Chien1,2
1Laboratory of Imaging Technologies, Faculty of electrical engineering, University of Ljubljana.
Insights
Automated labeling of cerebral arteries using vessel radius and spatial context significantly improved accuracy in identifying cerebrovascular disease risk factors from MRA scans.
Area of Science:
- Medical Imaging
- Neuroscience
- Computational Anatomy
Background:
- Cerebrovascular diseases are leading causes of mortality globally.
- Angiographic imaging is crucial for screening and diagnosing these conditions.
- Automated anatomical labeling of cerebral arteries aids in risk factor identification.
Purpose of the Study:
- To develop and evaluate an automated method for labeling cerebral arteries.
- To improve cross-sectional quantification and inter-subject comparison of cerebral vessels.
- To identify geometric risk factors for cerebrovascular diseases through enhanced vessel analysis.
Main Methods:
- Utilized 152 cerebral Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) datasets.
- Extracted and labeled vessel centerlines using nnU-net and VesselVio, referencing manual Slicer3D labels.
- Trained PointNet++ models incorporating centerline coordinates, vessel connectivity, radius, and spatial context.
Main Results:
- A model using only centerline coordinates achieved 0.93 accuracy (ACC) and 0.88 average True Positive Rate (TPR).
- Incorporating vessel radius improved performance to 0.95 ACC and 0.91 average TPR.
- Focusing spatial context to the Circle of Willis yielded the best results: 0.96 ACC and 0.93 average TPR.
Conclusions:
- Automated intracranial vessel labeling is significantly enhanced by incorporating vessel radius and spatial context.
- The developed method shows high performance, paving the way for clinical applications in cerebrovascular disease assessment.
- Accurate vessel labeling facilitates the identification of geometric risk factors, aiding in disease prevention and diagnosis.
Abstract:
Cerebrovascular diseases are among the world's top causes of death and their screening and diagnosis rely on angiographic imaging. We focused on automated anatomical labeling of cerebral arteries that enables their cross-sectional quantification and inter-subject comparisons and thereby identification of geometric risk factors correlated to the cerebrovascular diseases. We used 152 cerebral TOF-MRA angiograms from three publicly available datasets and manually created reference labeling using Slicer3D. We extracted centerlines from nnU-net based segmentations using VesselVio and labeled them according to the reference labeling. Vessel centerline coordinates, in combination with additional vessel connectivity, radius and spatial context features were used for training seven distinct PointNet++ models. Model trained solely on the vessel centerline coordinates resulted in ACC of 0.93 and across-labels average TPR was 0.88. Including vessel radius significantly improved ACC to 0.95, and average TPR to 0.91. Finally, focusing spatial context to the Circle of Willis are resulted in best ACC of 0.96 and best average TPR of 0.93. Hence, using vessel radius and spatial context greatly improved vessel labeling, with the attained perfomance opening the avenue for clinical applications of intracranial vessel labeling.
