Static and temporal dynamic changes in brain activity in patients with post-stroke balance dysfunction: a pilot...

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Post-stroke balance dysfunction significantly impacts patient recovery and quality of life.
• Understanding the neural underpinnings of this dysfunction is crucial for developing targeted interventions.

Purpose of the Study:

• To investigate brain activity alterations in patients with post-stroke balance dysfunction.
• To correlate these changes with clinical balance assessments.
• To develop a machine learning model for discriminating stroke patients from healthy controls.

Main Methods:

• Resting-state functional magnetic resonance imaging (rs-fMRI) was used to examine 26 post-stroke patients and 24 healthy controls (HCs).
• Static and dynamic functional brain imaging metrics (sALFF, sfALFF, sReHo, dALFF, dfALFF, dReHo) were calculated and compared.
• Extreme Gradient Boosting (XGBoost) algorithm was employed to build a classification model using imaging features and Berg Balance Scale (BBS) scores.

Main Results:

• Significant functional abnormalities were observed in various brain regions, including the insula, fusiform gyrus, thalamus, and supplementary motor area in patients compared to HCs.
• Pearson correlation analyses explored relationships between imaging metrics and BBS scores.
• The XGBoost model built using static imaging features demonstrated superior classification performance.

Conclusions:

• The study provides evidence of localized functional brain abnormalities contributing to post-stroke balance dysfunction.
• These abnormalities are associated with visual processing, motor control, and cognitive functions, elucidating neuropathological mechanisms.
• Extreme Gradient Boosting (XGBoost) shows promise as a machine learning tool for analyzing these neuroimaging changes.