Activation mapping in multi-center rat sensory-evoked functional MRI datasets using a unified pipeline

Abstract

Functional Magnetic Resonance Imaging (fMRI) in rodents is pivotal for understanding the mechanisms underlying Blood Oxygen Level-Dependent (BOLD) signals and phenotyping animal models of disorders, amongst other applications. Despite its growing use, comparing rodent fMRI results across different research sites remains challenging due to variations in experimental protocols. Here, we aggregated and analyzed 22 sensory-evoked rat fMRI datasets from 12 imaging centers, totaling scans from 220 rats, to assess the consistency of results across diverse protocols. We applied a standardized preprocessing pipeline and evaluated the impact of different hemodynamic response function models on group and individual level activity patterns. Our analysis revealed inter-dataset variability attributed to differences in experimental design, anesthesia protocols, and imaging parameters. We identified robust activation clusters in all (22/22) datasets. The comparison between stock human models implemented in software and rat-specific models showed significant variations in the resulting statistical maps. Our findings emphasize the necessity for standardized protocols and collaborative efforts to improve the reproducibility and reliability of rodent fMRI studies. We provide open access to all datasets and analysis code to foster transparency and further research in the field.