Background

Cerebral hypoperfusion is recognized as an early pathophysiological event in the Alzheimer’s disease (AD) continuum, preceding measurable cognitive decline and contributing to neurodegeneration. Arterial Spin Labeling (ASL) MRI provides a non-invasive, contrast-free measure of cerebral blood flow (CBF),  yet its application is limited in resource-constrained settings by protocol variability, technical gaps, and non-reproducible processing. To overcome these barriers, we developed a standardized ASL pipeline to improve reproducibility, facilitate knowledge transfer, and strengthen open-science.

Methods 

A low-cost, scalable perfusion analysis pipeline was implemented using Quantiphyse (v0.9.10) to enhance reproducibility in population studies within resource-constrained settings. This was conducted as a training exercise within the 2025 CONNExIN (COmprehensive Neuroimaging aNalysis Experience In resource-constraiNed settings) program, a neuroimaging analysis training initiative for African researchers. Six-trainees applied the processing pipeline to a local dataset of fourteen healthy Nigerian adults collected as part of the Scan With Me (SWiM) program. MRI data were acquired on a 1.5T United uMR 588 scanner including pseudo-continuous ASL acquired using a multi-postlabelling delay with a 3D GRASE readout, along with T1-weighted MPRAGE processed using Synthstrip. Data preprocessing and CBF quantification were performed using Quantifphyse automated pipeline, with the whole processes including step-by-step pipeline, configuration files and visual guide outlined in protocol.io.

Results 

The CBF maps generated from our pipeline (Figure 1) showed excellent inter-rater reliability (ICC=0.998; 95% CI: 0.995–0.999). Mean CBF values remained remarkably stable across independent operators (Table 1). Comparing graphical user interface (GUI) (n = 4) to command line (CL) (n = 6) pipeline implementation revealed high precision, with minor differences (x < 0.02 ml/100g/min; p< 0.05) within clinically insignificant limits. This demonstrates high reproducibility across different user interfaces.

Conclusions 

We implemented an established analysis pipeline with minimal challenges, including memory and software limitations. Strong agreement with trainer analyses and near-perfect ICC scores demonstrate this robust ASL pipeline’s reproducibility for reliable CBF quantification, offering a scalable pipeline for longitudinal Alzheimer’s research in Africa and other resource-constrained settings.