English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Automated segmentation of multiple sclerosis (MS) lesions from MRI scans isimportant to quantify disease progression. In recent years, convolutionalneural networks (CNNs) have shown top performance for this task when a largeamount of labeled data is available. However, the accuracy of CNNs suffers whendealing with few and/or sparsely labeled datasets. A potential solution is toleverage the information available in large public datasets in conjunction witha target dataset which only has limited labeled data. In this paper, we proposea training framework, SSL2 (self-supervised-semi-supervised), formulti-modality MS lesion segmentation with limited supervision. We adoptself-supervised learning to leverage the knowledge from large public 3Tdatasets to tackle the limitations of a small 7T target dataset. To leveragethe information from unlabeled 7T data, we also evaluate state-of-the-artsemi-supervised methods for other limited annotation settings, such as smalllabeled training size and sparse annotations. We use the shifted-window (Swin)transformer1 as our backbone network. The effectiveness of self-supervised andsemi-supervised training strategies is evaluated in our in-house 7T MRIdataset. The results indicate that each strategy improves lesion segmentationfor both limited training data size and for sparse labeling scenarios. Thecombined overall framework further improves the performance substantiallycompared to either of its components alone. Our proposed framework thusprovides a promising solution for future data/label-hungry 7T MS studies.

SSL^2: Self-Supervised Learning meets Semi-Supervised Learning: Multiple Sclerosis Segmentation in 7T-MRI from large-scale 3T-MRI