BolT: Fused window transformers for fMRI time series analysis

Simple item page
buir.contributor.authorBedel, Hasan Atakan
buir.contributor.authorŞıvgın, Irmak
buir.contributor.authorDalmaz, Onat
buir.contributor.authorDar, Salman Ul Hassan
buir.contributor.authorÇukur, Tolga
buir.contributor.orcidÇukur, Tolga|0000-0002-2296-851X
buir.contributor.orcidDalmaz, Onat|0000-0001-7978-5311
buir.contributor.orcidBedel, Hasan Atakan|0000-0001-5363-0610
dc.citation.epage102841-16en_US
dc.citation.spage102841-1
dc.citation.volumeNumber88
dc.contributor.authorBedel, Hasan Atakan
dc.contributor.authorŞıvgın, Irmak
dc.contributor.authorDalmaz, Onat
dc.contributor.authorUl Hassan Dar, Salman
dc.contributor.authorÇukur, Tolga
dc.date.accessioned2024-03-17T08:20:57Z
dc.date.available2024-03-17T08:20:57Z
dc.date.issued2023-05-18
dc.departmentDepartment of Electrical and Electronics Engineering
dc.departmentGraduate Program in Neuroscience - Ph.D. / Sc.D.
dc.departmentNational Magnetic Resonance Research Center (UMRAM)
dc.description.abstractDeep-learning models have enabled performance leaps in analysis of high-dimensional functional MRI (fMRI) data. Yet, many previous methods are suboptimally sensitive for contextual representations across diverse time scales. Here, we present BolT, a blood-oxygen-level-dependent transformer model, for analyzing multi-variate fMRI time series. BolT leverages a cascade of transformer encoders equipped with a novel fused window attention mechanism. Encoding is performed on temporally-overlapped windows within the time series to capture local representations. To integrate information temporally, cross-window attention is computed between base tokens in each window and fringe tokens from neighboring windows. To gradually transition from local to global representations, the extent of window overlap and thereby number of fringe tokens are progressively increased across the cascade. Finally, a novel cross-window regularization is employed to align high-level classification features across the time series. Comprehensive experiments on large-scale public datasets demonstrate the superior performance of BolT against state-of-the-art methods. Furthermore, explanatory analyses to identify landmark time points and regions that contribute most significantly to model decisions corroborate prominent neuroscientific findings in the literature.
dc.identifier.doi10.1016/j.media.2023.102841
dc.identifier.issn1361-8415
dc.identifier.urihttps://hdl.handle.net/11693/114836
dc.language.isoen_US
dc.publisherElsevier B.V.
dc.relation.isversionofhttps://dx.doi.org/10.1016/j.media.2023.102841
dc.rightsCC BY-NC-ND 4.0 DEED (Attribution-NonCommercial-NoDerivs 4.0 International)
dc.rights.urihttps://creativecommons.org/licenses/by-nc-nd/4.0/
dc.source.titleMedical Image Analysis
dc.subjectFunctional MRI
dc.subjectTime series
dc.subjectDeep learning
dc.subjectTransformer
dc.subjectClassification
dc.subjectConnectivity
dc.subjectExplainability
dc.titleBolT: Fused window transformers for fMRI time series analysis
dc.typeArticle

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
BolT_Fused_window_transformers_for_fMRI_time_series_analysis.pdf
Size:
2.74 MB
Format:
Adobe Portable Document Format
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.01 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Department of Electrical and Electronics Engineering
National Magnetic Resonance Research Center (UMRAM)