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 has remarkably improved computer vision. These advancesalso promise improvements in neuroimaging, where training set sizes are oftensmall. However, various difficulties arise in directly applying modelspretrained on natural images to radiologic images, such as MRIs. In particular,a mismatch in the input space (2D images vs. 3D MRIs) restricts the directtransfer of models, often forcing us to consider only a few MRI slices asinput. To this end, we leverage the 2D-Slice-CNN architecture of Gupta et al.(2021), which embeds all the MRI slices with 2D encoders (neural networks thattake 2D image input) and combines them via permutation-invariant layers. Withthe insight that the pretrained model can serve as the 2D encoder, weinitialize the 2D encoder with ImageNet pretrained weights that outperformthose initialized and trained from scratch on two neuroimaging tasks -- brainage prediction on the UK Biobank dataset and Alzheimer's disease detection onthe ADNI dataset. Further, we improve the modeling capabilities of 2D-Slicemodels by incorporating spatial information through position embeddings, whichcan improve the performance in some cases.

Transferring Models Trained on Natural Images to 3D MRI via Position Encoded Slice Models