Elsevier

Medical Image Analysis

Volume 68, February 2021, 101934
Medical Image Analysis

Semi-supervised task-driven data augmentation for medical image segmentation

https://doi.org/10.1016/j.media.2020.101934Get rights and content
Under a Creative Commons license
open access

Highlights

  • We present a novel task-driven and semi-supervised data augmentation scheme to improve medical image segmentation performance in a limited data setting.

  • In the proposed method, we design two conditional generative models to output two sets of transformations, namely deformation fields and additive intensity masks, to model shape and intensity characteristics, respectively.

  • The generated transformations are optimized for segmentation task performance (task-driven nature), and unlabeled data is leveraged in the generative process (semi-supervised nature).

  • We evaluated the proposed method on three datasets, namely cardiac, prostate, and pancreas, and obtained substantial performance gains over compared methods.

Abstract

Supervised learning-based segmentation methods typically require a large number of annotated training data to generalize well at test time. In medical applications, curating such datasets is not a favourable option because acquiring a large number of annotated samples from experts is time-consuming and expensive. Consequently, numerous methods have been proposed in the literature for learning with limited annotated examples. Unfortunately, the proposed approaches in the literature have not yet yielded significant gains over random data augmentation for image segmentation, where random augmentations themselves do not yield high accuracy. In this work, we propose a novel task-driven data augmentation method for learning with limited labeled data where the synthetic data generator, is optimized for the segmentation task. The generator of the proposed method models intensity and shape variations using two sets of transformations, as additive intensity transformations and deformation fields. Both transformations are optimized using labeled as well as unlabeled examples in a semi-supervised framework. Our experiments on three medical datasets, namely cardiac, prostate and pancreas, show that the proposed approach significantly outperforms standard augmentation and semi-supervised approaches for image segmentation in the limited annotation setting.

The code is made publicly available at https://github.com/krishnabits001/task_driven_data_augmentation.

Keywords

Data augmentation
Semi-supervised learning
Machine learning
Deep learning
Medical image segmentation

Cited by (0)