Estimations of the weather effects on brain functions using functional MRI: A cautionary note

Area of Science:

• Neuroimaging
• Environmental Science
• Machine Learning

Background:

• Environmental factors' impact on the human brain remains largely unknown.
• Previous neuroimaging studies on seasonal effects were limited by cross-sectional designs and small sample sizes.
• The stability of MRI scanners, potentially influenced by environmental conditions, is a critical but often overlooked factor.

Purpose of the Study:

• To investigate the influence of environmental factors on longitudinal resting-state functional MRI (fMRI) data.
• To determine if weather and environmental parameters can be predicted from resting-state fMRI metrics.
• To assess the impact of environmental factors on MRI scanner baseline signals versus actual brain function.

Main Methods:

• Longitudinal resting-state fMRI data from eight individuals scanned over months to years were analyzed.
• Machine learning regression was employed to predict environmental parameters using resting-state fMRI metrics (ALFF, ReHo, functional connectivity).
• Raw EPI, anatomical images, extra-cranial signals, and phantom scans were used for control and comparison.

Main Results:

• Daylight length and air temperature were reliably predicted using resting-state fMRI parameters.
• Similar prediction accuracies were achieved using single EPI frames and higher accuracies with anatomical images.
• Signals from outside the brain and phantom scans yielded higher prediction accuracies, indicating scanner-related influences.

Conclusions:

• No detectable influence of weather on human brain function was identified beyond effects on MRI scanner baseline signals.
• The predictability observed is likely attributable to environmental influences on MRI scanner stability, not direct brain responses.
• These findings underscore the significant challenges in studying long-term environmental effects on brain function using MRI.