Refined annotations of intrahepatic portal and hepatic veins from thin-slice CT images

Accurate segmentation of intrahepatic vessels in contrast-enhanced CT is limited by spatial resolution, variable contrast timing, and the complexity of peripheral vascular anatomy. Public datasets such as the Medical Segmentation Decathlon Task 08 include annotations of hepatic structures but provide incomplete delineation of small-caliber portal and hepatic veins. This work describes HVA-CT (Hepatic Venous Annotations CT), a dataset derived from the contrast-enhanced thin-slice CT volumes of the Medical Segmentation Decathlon Task 08, in which intrahepatic venous structures were re-annotated with more complete anatomical delineation. The dataset comprises refined three-dimensional segmentations of portal and hepatic venous systems generated through a structured workflow combining baseline model outputs, AI-assisted vessel branch identification, and expert review. For each case, updated venous segmentation masks, liver segmentations, and a standardized tool for intrahepatic vessel extraction are provided. The dataset is intended to support the development and evaluation of automated methods for hepatic vessel segmentation and analysis.