SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retraining

TL;DR

SynthSeg is a novel CNN-based brain MRI segmentation method that uses synthetic data with randomized contrast and resolution during training, enabling robust, zero-shot segmentation across diverse domains without retraining.

Contribution

SynthSeg introduces a domain randomization training strategy with synthetic data, allowing a single model to generalize across multiple MRI contrasts, resolutions, and even different imaging modalities.

Findings

SynthSeg outperforms supervised CNNs and domain adaptation methods in diverse datasets.

It achieves accurate segmentation across 6 modalities and 10 resolutions.

SynthSeg generalizes well to cardiac MRI and CT scans.

Abstract

Despite advances in data augmentation and transfer learning, convolutional neural networks (CNNs) difficultly generalise to unseen domains. When segmenting brain scans, CNNs are highly sensitive to changes in resolution and contrast: even within the same MRI modality, performance can decrease across datasets. Here we introduce SynthSeg, the first segmentation CNN robust against changes in contrast and resolution. SynthSeg is trained with synthetic data sampled from a generative model conditioned on segmentations. Crucially, we adopt a domain randomisation strategy where we fully randomise the contrast and resolution of the synthetic training data. Consequently, SynthSeg can segment real scans from a wide range of target domains without retraining or fine-tuning, which enables straightforward analysis of huge amounts of heterogeneous clinical data. Because SynthSeg only requires segmentations to be trained (no images), it can learn from labels obtained by automated methods on diverse populations (e.g., ageing and diseased), thus achieving robustness to a wide range of morphological variability. We demonstrate SynthSeg on 5,000 scans of six modalities (including CT) and ten resolutions, where it exhibits unparalleled generalisation compared with supervised CNNs, state-of-the-art domain adaptation, and Bayesian segmentation. Finally, we demonstrate the generalisability of SynthSeg by applying it to cardiac MRI and CT scans.