Predicting Motor Outcomes Using Atlas-Based Voxel Features of Post-Stroke Neuroimaging: A Scoping Review

Abstract

Background Atlas-based voxel features have the potential to aid motor outcome prognostication after stroke, but are seldom used in clinically feasible prediction models. This could be because neuroimaging feature development is a non-standardized, complex, multistep process. This is a barrier to entry for researchers and poses issues for reproducibility and validation in a field of research where sample sizes are typically small. Objectives The primary aim of this review is to describe the methodologies currently used in motor outcome prediction studies using atlas-based voxel neuroimaging features. Another aim is to identify neuroanatomical regions commonly used for motor outcome prediction. Methods A Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocol was constructed and OVID Medline and Scopus databases were searched for relevant studies. The studies were then screened and details about imaging modality, image acquisition, image normalization, lesion segmentation, region of interest determination, and imaging measures were extracted. Results Seventeen studies were included and examined. Common limitations were a lack of detailed reporting on image acquisition and the specific brain templates used for normalization and a lack of clear reasoning behind the atlas or imaging measure selection. A wide variety of sensorimotor regions relate to motor outcomes and there is no consensus use of one single sensorimotor atlas for motor outcome prediction. Conclusion There is an ongoing need to validate imaging predictors and further improve methodological techniques and reporting standards in neuroimaging feature development for motor outcome prediction post-stroke.

Point of View on Outcome Prediction Models in Post-Stroke Motor Recovery

“…In that study both interhemispheric and intrahemispheric cortical connectivity were studied and intrahemispheric connectivity was only a predictor for the lacunar stroke subanalysis. A recent scoping review investigated the methods currently used in motor outcome prediction studies using atlas-based voxel neuroimaging features 40 in order to better understand existing methodologies. They found a range of factors influencing the reliability and reproducibility of these imaging methods, which complicates model comparison, and should be addressed.…”

Outcome Prediction by Combining Corticospinal Tract Lesion Load with Diffusion-tensor Fractional Anisotropy in Patients after Hemorrhagic Stroke

“…The clinical utility of assessing CST integrity using conventional structural imaging has been investigated in several studies. [5][6][7][8][9] Using magnetic resonance images collected from stroke patients in the chronic phase, Zhu et al 10) first revealed that the volumes of overlapping portions of the lesion and the CST (CST lesion load, CST-LL) correlated with upper extremity functions. To enhance clinical utility during the acute phase, Feng et al 5) collected data from patients treated in acute care and conducted analyses similar to those of Zhu et al 10) These studies demonstrated that CST-LL moderately correlated with impairment levels of extremity functions.…”

“…14,15) In general, patients with a CST-FA lower than 0.40 were associated with severe hemiparesis. 14) Some studies have used CST-LL 9) or CST-FA 3) for the prediction of outcome in stroke patients. Nevertheless, only a few studies have explored the clinical applicability of integrating these two neuroimaging techniques.…”