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

Transfer learning represents a recent paradigm shift in the way we buildartificial intelligence (AI) systems. In contrast to training task-specificmodels, transfer learning involves pre-training deep learning models on a largecorpus of data and minimally fine-tuning them for adaptation to specific tasks.Even so, for 3D medical imaging tasks, we do not know if it is best topre-train models on natural images, medical images, or even syntheticallygenerated MRI scans or video data. To evaluate these alternatives, here webenchmarked vision transformers (ViTs) and convolutional neural networks(CNNs), initialized with varied upstream pre-training approaches. These methodswere then adapted to three unique downstream neuroimaging tasks with a range ofdifficulty: Alzheimer's disease (AD) and Parkinson's disease (PD)classification, "brain age" prediction. Experimental tests led to the followingkey observations: 1. Pre-training improved performance across all tasksincluding a boost of 7.4% for AD classification and 4.6% for PD classificationfor the ViT and 19.1% for PD classification and reduction in brain ageprediction error by 1.26 years for CNNs, 2. Pre-training on large-scale videoor synthetic MRI data boosted performance of ViTs, 3. CNNs were robust inlimited-data settings, and in-domain pretraining enhanced their performances,4. Pre-training improved generalization to out-of-distribution datasets andsites. Overall, we benchmarked different vision architectures, revealing thevalue of pre-training them with emerging datasets for model initialization. Theresulting pre-trained models can be adapted to a range of downstreamneuroimaging tasks, even when training data for the target task is limited.

Video and Synthetic MRI Pre-training of 3D Vision Architectures for Neuroimage Analysis