Vascular segmentation of kidney CT-scans through 2.5D model

The Vascular Common Coordinate Framework (VCCF) standardizes vascular structure representation, advancing medical diagnoses and research. This study introduces a 2.5D vascular segmentation model by combining U-Net and ResNet. The 2.5D model merges the benefits of 3D models with the efficiency of 2D approaches, achieving a Dice score of 0.805. It leverages information from adjacent slices, enhancing accuracy and detail compared to 2D methods, though it does not match the performance of full 3D models. Its computational efficiency makes it ideal for vascular segmentation, aiding the VCCF development and supporting disease research and clinical applications.