Hierarchical 3D Feature Learning for Pancreas Segmentation

TL;DR

This paper introduces a hierarchical 3D deep learning model for pancreas segmentation that effectively combines multi-scale features and multiple decoders, achieving state-of-the-art results on CT and promising results on MRI data.

Contribution

The paper presents a novel hierarchical 3D fully convolutional network with multiple decoders for improved pancreas segmentation from MRI and CT scans.

Findings

Outperforms existing methods on CT with 88% Dice score

Achieves 77% Dice score on challenging MRI data

Demonstrates the effectiveness of hierarchical multi-decoder architecture

Abstract

We propose a novel 3D fully convolutional deep network for automated pancreas segmentation from both MRI and CT scans. More specifically, the proposed model consists of a 3D encoder that learns to extract volume features at different scales; features taken at different points of the encoder hierarchy are then sent to multiple 3D decoders that individually predict intermediate segmentation maps. Finally, all segmentation maps are combined to obtain a unique detailed segmentation mask. We test our model on both CT and MRI imaging data: the publicly available NIH Pancreas-CT dataset (consisting of 82 contrast-enhanced CTs) and a private MRI dataset (consisting of 40 MRI scans). Experimental results show that our model outperforms existing methods on CT pancreas segmentation, obtaining an average Dice score of about 88%, and yields promising segmentation performance on a very challenging MRI data set (average Dice score is about 77%). Additional control experiments demonstrate that the achieved performance is due to the combination of our 3D fully-convolutional deep network and the hierarchical representation decoding, thus substantiating our architectural design.