A Pipeline for ADNI Resting-State Functional MRI Processing and Quality Control

Abstract The Alzheimer’s Disease Neuroimaging Initiative (ADNI) provides a comprehensive multimodal neuroimaging resource for studying aging and Alzheimer’s disease (AD). However, its resting-state functional MRI (rs-fMRI) data remain underutilized, in part due to challenges with data processing driven by considerable variability in acquisition protocols, inconsistent organization, missing or incomplete imaging sessions, and frequency of data quality issues due to participant motion and aging-related factors. To facilitate widespread use of ADNI rs-fMRI data, we developed an open-source pipeline encompassing the download and organization of imaging and clinical data, conversion to Brain Imaging Data Structure (BIDS), preprocessing with fMRIPrep, and automated quality control (QC). Our QC considers acquisition parameters and 16 structural and functional MRIQC image-quality metrics, flagging scans that fall outside acceptable ranges based on expert input and robust outlier detection techniques. Processed imaging sessions are matched to temporally aligned clinical assessments to produce an analysis-ready dataset. Applying the pipeline to more than 2,800 sessions from ADNI GO, ADNI 2, and ADNI 3 yielded 2,086 quality-checked functional and structural MRI sessions from 935 unique participants. Exclusions occurred at every processing stage and were driven primarily by BIDS conversion errors, image quality flags, and preprocessing- or motion-related failures. The open-source pipeline described in this work produces one of the largest pre-processed, quality controlled, and analysis-ready functional and structural MRI datasets yet available for studying aging and AD. It lowers barriers to large-cohort and longitudinal neuroimaging research and supports large-scale neuroimaging biomarker discovery in AD.