Deep convolutional neural networks outperform vanilla machine learning when predicting language outcomes after stroke

Aug 4, 2026

Deep learning and stroke prognosis are addressed through models that predict patients’ language skills after stroke using brain imaging data. Current clinical practice lacks confident prediction of post-stroke language outcomes, motivating machine learning approaches that typically rely on tabular lesion features derived from high-dimensional structural MRI. The study tests whether deep convolutional neural networks can reduce or eliminate this image post-processing by learning robust representations directly.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Introduction

Methods

Model training and validation

Results

Conclusions

University of Birmingham Deep convolutional neural networks outperform vanilla machine learning when predicting language outcomes after stroke Hope, Thomas M. H. ; Bowman, Howard; Leff, Alex P. ; Price, Cathy J. DOI: 10.1016/j.nicl.2025.103880 License: Creative Commons: Attribution (CC BY) Document Version Publisher's PDF, also known as Version of record Citation for published version (Harvard): Hope, TMH, Bowman, H, Leff, AP & Price, CJ 2025, 'Deep convolutional neural networks outperform vanilla machine learning when predicting language outcomes after stroke', NeuroImage: Clinical, vol. 48, 103880. [https://doi.org/10.1016/j.nicl.2025.103880](https://doi.org/10.1016/j.nicl.2025.103880) Link to publication on Research at Birmingham portal General rights Unless a licence is specified above, all rights (including copyright and moral rights) in this document are retained by the authors and/or the copyright holders. The express permission of the copyright holder must be obtained for any use of this material other than for purposes permitted by law. •Users may freely distribute the URL that is used to identify this publication. •Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of private study or non-commercial research. •User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?) •Users may not further distribute the material nor use it for the purposes of commercial gain. Where a licence is displayed above, please note the terms and conditions of the licence govern your use of this document. When citing, please reference the published version. Take down policy While the University of Birmingham exercises care and attention in making items available there are rare occasions when an item has been uploaded in error or has been deemed to be commercially or otherwise sensitive. If you believe that this is the case for this document, [please contact UBIRA@lists.bham.ac.uk](please contact UBIRA@lists.bham.ac.uk) providing details and we will remove access to the work immediately and investigate. Download date: 03. Aug. 2026 NeuroImage: Clinical 48 (2025) 103880 Contents lists available at ScienceDirect NeuroImage: Clinical journal [homepage: www.elsevier.com/locate/ynicl](homepage: www.elsevier.com/locate/ynicl) | Deep convolutional neural networks outperform vanilla machine learning when predicting language outcomes after stroke
Thomas M.H. Hope a,c,*, Howard Bowman b, Alex P. Leffa, Cathy J. Price a
a Department of Imaging Neuroscience, Institute of Neurology, University College London, 12 Queen Square, London WC1N 3AR, the United Kingdom of Great Britain and Northern Ireland
b School of Psychology University of Birmingham Edgbaston Birmingham B15 2TT the United Kingdom of Great Britain and Northern Ireland c Department of Psychological and Social Sciences, John Cabot University, Via della Lungara 233, 00165, Rome, Italy | | | | --- | --- | --- | | A R T I C L E I N F O | | A B S T R A C T | | Keywords:
Stroke Language Cognition Machine learning Lesions
Deep learning | | Background: Current medicine cannot confidently predict patients’ language skills after stroke. In recent years, researchers have sought to bridge this gap with machine learning. These models appear to benefit from access to features describing where and how much brain damage these patients have suffered. Given the very high dimensionality of structural brain imaging data, those brain lesion features are typically post-processed from the images themselves into tabular features. With the introduction of deep Convolutional Neural Networks (CNN), which appear to be much more robust to high dimensional data, it is natural to hope that much of this image post-processing might be unnecessary. But prior attempts to demonstrate this (in the area of p

FAQs

Why is predicting language outcomes after stroke difficult in current medicine?

Current medicine cannot reliably forecast patients’ language abilities after stroke, leaving clinicians and patients without confident prognostic information.

What baseline approach does the study compare against deep CNNs?

The baseline uses state-of-the-art “vanilla” machine learning models (boosted ensembles) that combine demographic variables with post-processed tabular lesion features from brain imaging.

How do the CNN models perform compared with vanilla machine learning?

CNN models consistently outperform the vanilla machine learning models and reduce the need to convert lesion images into lesion-feature tables for prediction.

Aug 4, 2026

Upload by Evangeline