Unsupervised Cross-Modality Domain Adaptation for Segmenting Vestibular Schwannoma and Cochlea with Data Augmentation and Model Ensemble

TL;DR

This paper introduces an unsupervised domain adaptation framework that segments vestibular schwannoma and cochlea in MRI scans without requiring manual labels in the target domain, using image translation, model ensembling, and data augmentation.

Contribution

It presents a novel unsupervised learning approach combining image translation, model ensembling, and online data augmentation for cross-modality MRI segmentation.

Findings

Achieved mean Dice scores of 0.7930 for VS and 0.7432 for cochlea.

Effectively handles variability across different MRI sites and scanners.

Produces promising segmentation results without manual labels in target domain.

Abstract

Magnetic resonance images (MRIs) are widely used to quantify vestibular schwannoma and the cochlea. Recently, deep learning methods have shown state-of-the-art performance for segmenting these structures. However, training segmentation models may require manual labels in target domain, which is expensive and time-consuming. To overcome this problem, domain adaptation is an effective way to leverage information from source domain to obtain accurate segmentations without requiring manual labels in target domain. In this paper, we propose an unsupervised learning framework to segment the VS and cochlea. Our framework leverages information from contrast-enhanced T1-weighted (ceT1-w) MRIs and its labels, and produces segmentations for T2-weighted MRIs without any labels in the target domain. We first applied a generator to achieve image-to-image translation. Next, we ensemble outputs from an ensemble of different models to obtain final segmentations. To cope with MRIs from different sites/scanners, we applied various 'online' augmentations during training to better capture the geometric variability and the variability in image appearance and quality. Our method is easy to build and produces promising segmentations, with a mean Dice score of 0.7930 and 0.7432 for VS and cochlea respectively in the validation set.