Evaluation of Anatomical Shape Priors in Deep Learning-Based Cardiac Multi-Compartment Segmentation

TL;DR

This study evaluates the effectiveness of explicit anatomical shape priors in deep learning models for cardiac segmentation, finding that standard models already capture much of the anatomical regularity, with handcrafted priors offering limited improvements.

Contribution

The paper systematically assesses the impact of lightweight explicit shape priors on 3D cardiac segmentation performance, revealing their limited benefit over strong baseline models.

Findings

Standard 3D U-Net performs strongly without explicit priors.

Handcrafted priors yield marginal, inconsistent improvements.

Baseline models already encode significant anatomical regularities.

Abstract

Whole-heart multi-compartment CT segmentation is clinically important, but standard CNNs do not explicitly enforce anatomical plausibility. Based on statistics derived from the training data, we evaluate whether lightweight explicit shape priors, implemented as shape-aware losses and spatial label distribution heatmap-guided U-Net variants, improve 3D cardiac segmentation on MM-WHS CT and WHS++. Across all experiments, a standard 3D U-Net surprisingly remained a very strong baseline, with handcrafted priors yielding at best marginal and inconsistent changes and often degrading performance. These results suggest that the baseline already captures substantial implicit anatomical regularities and that future gains will likely require more expressive learned priors rather than simple handcrafted anatomical shape constraints.