Interhemispheric Cortical Network Connectivity Reorganization Predicts Vision Impairment in Stroke

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Stroke is a leading cause of disability, often causing vision loss (homonymous hemianopia) when the occipital lobe is affected.
• Graph theory-based functional connectivity network (FCN) analysis is emerging for studying brain mechanisms of vision loss and recovery.
• Predictive power of damaged FCN strength for functional impairment extent remains under-explored.

Purpose of the Study:

• To characterize brain FCNs using deep neural network analysis for multiscale network description and physiological pattern exploration.
• To evaluate the efficiency of Bi-directional Long Short-Term Memory (Bi-LSTM) in learning cortical network patterns compared to traditional algorithms.
• To investigate if FCN reorganization patterns can be linked to stroke-induced vision impairment.

Main Methods:

• Deep neural network analysis, specifically Bi-directional Long Short-Term Memory (Bi-LSTM), was applied to characterize brain FCNs.
• The study included 24 patients and 24 healthy controls.
• Network pattern learning efficiency was compared between Bi-LSTM and other algorithms, focusing on the low alpha band.

Main Results:

• Bi-LSTM demonstrated superior performance with a balanced-overall accuracy of 73% (70% sensitivity, 75% specificity) in the low alpha band.
• Bi-directional learning effectively captured brain network feature representations from both hemispheres.
• Brain damage was associated with reorganized FCN patterns, notably an increase in functional connections of intermediate density within the high alpha band.

Conclusions:

• Deep neural network analysis, particularly Bi-LSTM, is effective in characterizing complex brain FCNs and their alterations post-stroke.
• The findings suggest that brain damage reorganizes FCNs, with specific patterns observed in different frequency bands.
• Further research is warranted to translate these FCN insights into clinical diagnostic tools and rehabilitation strategies for vision impairment after stroke.