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

We propose to apply non-linear representation learning to voxelwise rs-fMRIdata. Learning the non-linear representations is done using a variationalautoencoder (VAE). The VAE is trained on voxelwise rs-fMRI data and performsnon-linear dimensionality reduction that retains meaningful information. Theretention of information in the model's representations is evaluated usingdownstream age regression and sex classification tasks. The results on thesetasks are highly encouraging and a linear regressor trained with therepresentations of our unsupervised model performs almost as well as asupervised neural network, trained specifically for age regression on the samedataset. The model is also evaluated with a schizophrenia diagnosis predictiontask, to assess its feasibility as a dimensionality reduction method forneuropsychiatric datasets. These results highlight the potential forpre-training on a larger set of individuals who do not have mental illness, toimprove the downstream neuropsychiatric task results. The pre-trained model isfine-tuned for a variable number of epochs on a schizophrenia dataset and wefind that fine-tuning for 1 epoch yields the best results. This work thereforenot only opens up non-linear dimensionality reduction for voxelwise rs-fMRIdata but also shows that pre-training a deep learning model on voxelwisers-fMRI datasets greatly increases performance even on smaller datasets. Italso opens up the ability to look at the distribution of rs-fMRI time series inthe latent space of the VAE for heterogeneous neuropsychiatric disorders likeschizophrenia in future work. This can be complemented with the generativeaspect of the model that allows us to reconstruct points from the model'slatent space back into brain space and obtain an improved understanding of therelation that the VAE learns between subjects, timepoints, and a subject'scharacteristics.

Variational voxelwise rs-fMRI representation learning: Evaluation of sex, age, and neuropsychiatric signatures