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

## TL;DR

This study examined how different data processing methods affect brain connectivity results in people who are blind from an early age.

## Contribution

The study shows that different preprocessing pipelines significantly alter resting-state functional connectivity results in early blind individuals.

## Key findings

- The RExt pipeline drastically changed rsFC values and reduced network density more than RMin.
- Largest rsFC effects were consistent across pipelines, suggesting robustness of major findings.
- Results highlight the importance of verifying findings using multiple preprocessing methods.

## Abstract

It has been shown that the choice of preprocessing pipelines to remove contamination from functional magnetic resonance images can significantly impact the results, particularly in resting-state functional connectivity (rsFC) studies. This underscores the critical importance of replication studies with different preprocessing methodologies. In this study, we attempted to reproduce the rsFC results presented in an original study by Bauer et al. in 2017 on a group of sighted control (SC) and early blind (EB) subjects. By using the original dataset, we utilized another widely used software package to investigate how applying different implementations of the original pipeline (RMin model) or a more rigorous and extensive preprocessing stream (RExt model) can alter the whole-brain rsFC results. Our replication study was not able to fully reproduce the findings of the original paper. Overall, RExt shifted the distribution of rsFC values and reduced functional network density more drastically compared with RMin and the original pipeline. Remarkably, the largest rsFC effects appeared to primarily belong to certain connection pairs, irrespective of the pipeline used, likely demonstrating immunity of the larger effects and the true results against suboptimal processing. This may highlight the significance of results verification across different computational streams in pursuit of the true findings.