Abstract
The Circle of Willis (CoW) is an important network of arteries connectingmajor circulations of the brain. Its vascular architecture is believed toaffect the risk, severity, and clinical outcome of serious neuro-vasculardiseases. However, characterizing the highly variable CoW anatomy is still amanual and time-consuming expert task. The CoW is usually imaged by twoangiographic imaging modalities, magnetic resonance angiography (MRA) andcomputed tomography angiography (CTA), but there exist limited public datasetswith annotations on CoW anatomy, especially for CTA. Therefore we organized theTopCoW Challenge in 2023 with the release of an annotated CoW dataset. TheTopCoW dataset was the first public dataset with voxel-level annotations forthirteen possible CoW vessel components, enabled by virtual-reality (VR)technology. It was also the first large dataset with paired MRA and CTA fromthe same patients. TopCoW challenge formalized the CoW characterization problemas a multiclass anatomical segmentation task with an emphasis on topologicalmetrics. We invited submissions worldwide for the CoW segmentation task, whichattracted over 140 registered participants from four continents. The topperforming teams managed to segment many CoW components to Dice scores around90%, but with lower scores for communicating arteries and rare variants. Therewere also topological mistakes for predictions with high Dice scores.Additional topological analysis revealed further areas for improvement indetecting certain CoW components and matching CoW variant topology accurately.TopCoW represented a first attempt at benchmarking the CoW anatomicalsegmentation task for MRA and CTA, both morphologically and topologically.