VEELA: New Benchmark for Liver Vessel Segmentation in CTA
VEELA (Vessel Extraction and Extrication for Liver Analysis) has been launched by researchers as a clinically-constrained standard for segmenting liver vessels in computed tomography angiography (CTA). This benchmark is based on 40 CTA scans from the CHAOS grand-challenge cohort and includes meticulous manual delineation of each slice, achieved through multi-expert consensus and a rigorous visibility-driven annotation approach, which avoids interpolation based on anatomical assumptions. This methodology effectively addresses anatomical variability and uncertainties related to imaging, facilitating reproducible evaluations across different benchmarks and broadening the potential for detailed vessel analysis.
Key facts
- VEELA is a benchmark for liver vessel segmentation in CTA.
- Derived from 40 CTA scans from the CHAOS grand-challenge cohort.
- Vessels manually delineated slice-by-slice under multi-expert consensus.
- Uses a strict visibility-driven annotation policy.
- Avoids anatomically inferred interpolation.
- Captures anatomical variability and imaging-related uncertainty.
- Enables reproducible cross-benchmark evaluation.
- Extends scope to fine-grain vessel analysis.
Entities
—