Investigating the structure-function relationship of the corticomotor system early after stroke using machine...

Area of Science:

• Neuroscience
• Medical Imaging
• Biomarkers

Background:

• Predicting motor outcomes post-stroke relies on structural and functional biomarkers of the corticomotor pathway.
• Magnetic resonance imaging (MRI) and transcranial magnetic stimulation (TMS) are common measurement tools.
• Precise structural determinants of corticomotor function post-stroke remain unclear.

Purpose of the Study:

• To identify structure-function links in the corticomotor pathway after stroke.
• To understand mechanisms of post-stroke motor impairment.
• To classify upper limb motor evoked potential status using early MRI metrics via machine learning.

Main Methods:

• Retrospective analysis of 91 patients with moderate to severe upper limb weakness within a week post-stroke.
• Support vector machine (SVM) classifiers trained on T1- and diffusion-weighted MRI metrics.
• Motor evoked potential (MEP) status measured empirically using TMS.

Main Results:

• SVM classification of MEP status achieved 81% accuracy.
• Key predictors included diffusion anisotropy asymmetry in motor tracts and lesion overlap in sensorimotor/premotor tracts.
• Mean diffusivity asymmetry in internal capsule posterior limbs was also significant.

Conclusions:

• Structural MRI metrics are valuable but imperfect predictors of corticomotor function post-stroke.
• Residual motor function depends on macrostructural damage and white matter microstructural integrity.
• Multivariable MRI analysis of the corticomotor pathway offers more insight than univariate approaches.