# PHiSeg: Capturing Uncertainty in Medical Image Segmentation

## TL;DR

This paper introduces PHiSeg, a hierarchical probabilistic model that captures uncertainty in medical image segmentation, producing diverse and realistic segmentation samples by modeling the distribution of possible annotations.

## Contribution

The work presents a novel hierarchical probabilistic framework using variational autoencoders to model segmentation uncertainty, addressing ambiguity in medical image annotations.

## Key findings

- Generates more realistic and diverse segmentation samples.
- Effective with annotations from single or multiple annotators.
- Outperforms existing methods in modeling segmentation uncertainty.

## Abstract

Segmentation of anatomical structures and pathologies is inherently ambiguous. For instance, structure borders may not be clearly visible or different experts may have different styles of annotating. The majority of current state-of-the-art methods do not account for such ambiguities but rather learn a single mapping from image to segmentation. In this work, we propose a novel method to model the conditional probability distribution of the segmentations given an input image. We derive a hierarchical probabilistic model, in which separate latent variables are responsible for modelling the segmentation at different resolutions. Inference in this model can be efficiently performed using the variational autoencoder framework. We show that our proposed method can be used to generate significantly more realistic and diverse segmentation samples compared to recent related work, both, when trained with annotations from a single or multiple annotators.