Replicability of a resting-state functional connectivity study in profound early blindness

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Cognitive Neuroscience

Background:

• Preprocessing pipelines in functional magnetic resonance imaging (fMRI) significantly influence resting-state functional connectivity (rsFC) findings.
• Replication studies are crucial for validating results, especially when employing diverse preprocessing methodologies.

Purpose of the Study:

• To reproduce resting-state functional connectivity (rsFC) results from Bauer et al. (2017) using a different software package.
• To investigate the impact of two distinct preprocessing models (RMin and RExt) on whole-brain rsFC results in sighted control (SC) and early blind (EB) subjects.

Main Methods:

• Utilized an existing fMRI dataset from sighted control (SC) and early blind (EB) participants.
• Applied two preprocessing pipelines: RMin (similar to the original) and RExt (more rigorous).
• Analyzed whole-brain resting-state functional connectivity (rsFC) differences between the pipelines.

Main Results:

• The replication study could not fully reproduce the original paper's findings.
• The RExt preprocessing stream significantly altered rsFC value distribution and reduced functional network density compared to RMin and the original pipeline.
• Robust rsFC effects were observed for specific connections, suggesting resilience to variations in preprocessing.

Conclusions:

• Preprocessing pipeline choice critically affects rsFC results, underscoring the need for methodological transparency.
• While some findings are sensitive to preprocessing, the largest effects may represent genuine neural signals resilient to processing variations.
• Verification of neuroimaging results across different computational streams is essential for robust scientific discovery.