Large-Scale Multi-Center Dataset

Where it comes from

Highlighted claims

• The study used 1,341 human time-of-flight MRA images from 14 public scanning centers. — AG-TAL: Anatomically-Guided Topology-Aware Loss for Multiclass Segmentation of the Circle of Willis Using Large-Scale Multi-Center Datasets
• The manually annotated training and test pool contained 926 cases. — AG-TAL: Anatomically-Guided Topology-Aware Loss for Multiclass Segmentation of the Circle of Willis Using Large-Scale Multi-Center Datasets
• Independent validation used 415 cases from six datasets. — AG-TAL: Anatomically-Guided Topology-Aware Loss for Multiclass Segmentation of the Circle of Willis Using Large-Scale Multi-Center Datasets
• The dataset covered scanners from multiple vendors and field strengths with variable voxel resolutions. — AG-TAL: Anatomically-Guided Topology-Aware Loss for Multiclass Segmentation of the Circle of Willis Using Large-Scale Multi-Center Datasets