Deep vessel segmentation with joint multi-prior encoding

TL;DR

This work tackles automatic vascular segmentation by embedding high-level anatomical priors into a unified latent space. It introduces Joint Multi-Prior Encoding (JMPE), a multi-task convolutional auto-encoder that shares a single encoder to jointly encode shape and topology priors, producing a compact code $\mathbf{z} = E_{\theta}(\mathbf{y})$ used by decoders for segmentation and topology. The segmentation loss is augmented with a JMPE-based regularizer, enabling anatomically consistent delineation without multiple encoders. On the 3D-IRCADb hepatic vessel dataset, the method yields improved Dice and Jaccard scores and robust connectivity metrics, highlighting the practical impact of anatomy-aware priors for medical image segmentation and suggesting avenues for generalization and graph-based extensions.

Abstract

The precise delineation of blood vessels in medical images is critical for many clinical applications, including pathology detection and surgical planning. However, fully-automated vascular segmentation is challenging because of the variability in shape, size, and topology. Manual segmentation remains the gold standard but is time-consuming, subjective, and impractical for large-scale studies. Hence, there is a need for automatic and reliable segmentation methods that can accurately detect blood vessels from medical images. The integration of shape and topological priors into vessel segmentation models has been shown to improve segmentation accuracy by offering contextual information about the shape of the blood vessels and their spatial relationships within the vascular tree. To further improve anatomical consistency, we propose a new joint prior encoding mechanism which incorporates both shape and topology in a single latent space. The effectiveness of our method is demonstrated on the publicly available 3D-IRCADb dataset. More globally, the proposed approach holds promise in overcoming the challenges associated with automatic vessel delineation and has the potential to advance the field of deep priors encoding.

Figures (3)

• Figure 1: Proposed pipeline overview. Parameters of the segmentation model $\phi$ are estimated by penalizing the segmentation loss $\ell_{\phi}$ with a regularization term $\mathcal{L}_{reg}^{JMPE}$ that deals with the similarity between the projections of the prediction $\hat{\pmb{y}}$ and the ground truth $\pmb{y}$ in a learned multi-priors embedding.
• Figure 2: Multi-task convolutional auto-encoder network $\xi$ architecture for Joint Multi-Prior Encoding (JMPE).
• Figure 3: Liver vessel segmentation results on 3D-IRCADb soler20103d using various priors and 3D ResUNet as backbone.