Metric comparison of connectome-based lesion-symptom mapping in post-stroke aphasia.

Autor:	Ding J; Department of Psychology, University of Edinburgh, Edinburgh EH8 9JZ, UK., Thye M; Department of Psychology, University of Edinburgh, Edinburgh EH8 9JZ, UK., Edmondson-Stait AJ; Department of Psychology, University of Edinburgh, Edinburgh EH8 9JZ, UK., Szaflarski JP; Department of Neurology, University of Alabama at Birmingham, Birmingham, AL 35294, USA., Mirman D; Department of Psychology, University of Edinburgh, Edinburgh EH8 9JZ, UK.
Jazyk:	angličtina
Zdroj:	Brain communications [Brain Commun] 2024 Sep 12; Vol. 6 (5), pp. fcae313. Date of Electronic Publication: 2024 Sep 12 (Print Publication: 2024).
DOI:	10.1093/braincomms/fcae313
Abstrakt:	Connectome-based lesion-symptom mapping relates behavioural impairments to disruption of structural brain connectivity. Connectome-based lesion-symptom mapping can be based on different approaches (diffusion MRI versus lesion mask), network scales (whole brain versus regions of interest) and measure types (tract-based, parcel-based, or network-based metrics). We evaluated the similarity of different connectome-based lesion-symptom mapping processing choices and identified factors that influence the results using multiverse analysis-the strategy of conducting and displaying the results of all reasonable processing choices. Metrics derived from lesion masks and diffusion-weighted images were tested for association with Boston Naming Test and Token Test performance in a sample of 50 participants with aphasia following left hemispheric stroke. 'Direct' measures were derived from diffusion-weighted images. 'Indirect' measures were derived by overlaying lesion masks on a white matter atlas. Parcel-based connectomes were constructed for the whole brain and regions of interest (14 language-relevant parcels). Numerous tract-based and network-based metrics were calculated. There was a high discrepancy across processing approaches (diffusion-weighted images versus lesion masks), network scales (whole brain versus regions of interest) and metric types. Results indicate weak correlations and different connectome-based lesion-symptom mapping results across the processing choices. Substantial methodological work is needed to validate the various decision points that arise when conducting connectome-based lesion-symptom mapping analyses. Multiverse analysis is a useful strategy for evaluating the similarity across different processing choices in connectome-based lesion-symptom mapping. Competing Interests: The authors report no competing interests. (© The Author(s) 2024. Published by Oxford University Press on behalf of the Guarantors of Brain.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu