PADDIT: Probabilistic Augmentation of Data using Diffeomorphic Image Transformation

TL;DR

PADDIT is a systematic framework that generates realistic diffeomorphic transformations to augment training data, improving CNN performance in medical image segmentation by enhancing invariance to shape variations.

Contribution

We introduce PADDIT, a novel probabilistic augmentation method using diffeomorphic transformations to improve CNN generalization in medical image segmentation.

Findings

CNNs trained with PADDIT outperform those without augmentation.

PADDIT outperforms generic augmentation methods.

Improved segmentation accuracy on brain MRI scans.

Abstract

For proper generalization performance of convolutional neural networks (CNNs) in medical image segmentation, the learnt features should be invariant under particular non-linear shape variations of the input. To induce invariance in CNNs to such transformations, we propose Probabilistic Augmentation of Data using Diffeomorphic Image Transformation (PADDIT) -- a systematic framework for generating realistic transformations that can be used to augment data for training CNNs. We show that CNNs trained with PADDIT outperforms CNNs trained without augmentation and with generic augmentation in segmenting white matter hyperintensities from T1 and FLAIR brain MRI scans.