Skip to main content

ORIGINAL RESEARCH article

Front. Radiol.
Sec. Artificial Intelligence in Radiology
Volume 4 - 2024 | doi: 10.3389/fradi.2024.1495181

Language Task-based fMRI Analysis using Machine Learning and Deep Learning

Provisionally accepted
Elaine Kuan Elaine Kuan 1,2Viktor Vegh Viktor Vegh 1,2*John Phamnguyen John Phamnguyen 1,3Kieran O'brien Kieran O'brien 4Amanda Hammond Amanda Hammond 4David Reutens David Reutens 1,2,3
  • 1 The University of Queensland, Brisbane, Australia
  • 2 ARC Training Centre for Innovation in Biomedical Imaging Technology, Brisbane, Australia
  • 3 Royal Brisbane and Women's Hospital, Brisbane, Queensland, Australia
  • 4 Siemens Healthcare Pty Ltd, Brisbane, Australia

The final, formatted version of the article will be published soon.

    Task-based language fMRI is a non-invasive method of identifying brain regions subserving language that is used to plan neurosurgical resections which potentially encroach on eloquent regions. The use of unstructured fMRI paradigms, such as naturalistic fMRI, to map language is of increasing interest. Their analysis necessitates the use of alternative methods such as machine learning (ML) and deep learning (DL) because task regressors may be difficult to define in these paradigms. Using task-based language fMRI as a starting point, this study investigates the use of different categories of ML and DL algorithms to identify brain regions subserving language. Data comprising of seven task-based language fMRI paradigms were collected from 26 individuals, and ML and DL models were trained to classify voxel-wise fMRI time series. The general machine learning and the interval-based methods were the most promising in identifying language areas using fMRI time series classification. The geneal machine learning method achieved a mean whole-brain Area Under the Receiver Operating Characteristic Curve (AUC) of 0.97±0.03, mean Dice coefficient of 0.6±0.34 and mean Euclidean distance of 2.7±2.4mm between activation peaks across the evaluated regions of interest. The interval-based method achieved a mean whole-brain AUC of 0.96±0.03, mean Dice coefficient of 0.61±0.33 and mean Euclidean distance of 3.3±2.7mm between activation peaks across the evaluated regions of interest. Hence, this study demonstrates the utility of different ML and DL methods in classifying task-based language fMRI time series. A potential application of these methods is the identification of language activation from unstructured paradigms.

    Keywords: Task-based fMRI, Language, time series, brain activation, machine learning, deep learning

    Received: 12 Sep 2024; Accepted: 12 Nov 2024.

    Copyright: © 2024 Kuan, Vegh, Phamnguyen, O'brien, Hammond and Reutens. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    * Correspondence: Viktor Vegh, The University of Queensland, Brisbane, Australia

    Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.