🤖 AI Summary
This work addresses the challenge of accurate segmentation of five abdominal organs in large-field-of-view 3D CT scans by proposing a two-stage lightweight framework. In the first stage, an axial 2D U-Net coarsely localizes organ regions; in the second stage, multi-planar (axial, sagittal, and coronal) 2D U-Net predictions are fused within the localized region, augmented with a fuzzy 3D spatial occurrence map that encodes anatomical location priors for refined segmentation. The integration of the spatial occurrence map, multi-planar context, and region-of-interest cropping substantially enhances segmentation accuracy. Evaluated on 80 multi-source public CT cases, the proposed method achieves up to a 4% improvement in Dice coefficient over a baseline model without the spatial occurrence map.
📝 Abstract
This work proposes a lightweight 2D-U-Net-based framework for segmenting five abdominal organs in large field-of-view 3D CT scans. The method combines coarse-to-fine segmentation, predictions from multiple anatomical planes, and additional fuzzy 3D spatial maps that provide anatomical location cues to improve segmentation accuracy. We combine multi-planar 2D-U-Net models augmented by a spatial occurrence map. The approach involves two main stages. First, the abdominal volume of interest region is detected by traversing the whole scan axially with a 2D-U-Net and determining the x-y-z-minimum and -maximum extents of the 5 abdominal organs of interest. Second, we use spatial occurrence maps to enhance our multi-planar 2D-U-net architecture inside the bounds from the former stage. The method is evaluated on 80 CT scans from various public sources. The results show Dice improvements of about 4% at maximum compared to the same model trained without spatial occurrence maps.