A Hierarchical Probabilistic U-Net for Modeling Multi-Scale Ambiguities
This addresses ambiguous segmentation in medical imaging, which is incremental as it builds on existing probabilistic U-Net methods.
The paper tackles the problem of ambiguous medical image segmentation by proposing a Hierarchical Probabilistic U-Net, which uses a hierarchical latent space decomposition to model multi-scale variations, resulting in high-fidelity sampling and reconstruction of segmentations.
Medical imaging only indirectly measures the molecular identity of the tissue within each voxel, which often produces only ambiguous image evidence for target measures of interest, like semantic segmentation. This diversity and the variations of plausible interpretations are often specific to given image regions and may thus manifest on various scales, spanning all the way from the pixel to the image level. In order to learn a flexible distribution that can account for multiple scales of variations, we propose the Hierarchical Probabilistic U-Net, a segmentation network with a conditional variational auto-encoder (cVAE) that uses a hierarchical latent space decomposition. We show that this model formulation enables sampling and reconstruction of segmenations with high fidelity, i.e. with finely resolved detail, while providing the flexibility to learn complex structured distributions across scales. We demonstrate these abilities on the task of segmenting ambiguous medical scans as well as on instance segmentation of neurobiological and natural images. Our model automatically separates independent factors across scales, an inductive bias that we deem beneficial in structured output prediction tasks beyond segmentation.