A hands-on, application-focused workshop that takes complete beginners from raw MRI data to a reproducible fMRI analysis on an HPC cluster.
The emphasis is on actually running the pipeline — preprocessing, the GLM, quality control, troubleshooting, and post-processing — with just enough principles and experimental design to use them well.
True beginners with no programming or fMRI background. A basic grasp of applied statistics and comfort reading papers and writing academically are expected; no mathematics or theoretical-statistics background is needed.
| # | Session | Focus | Key tools |
|---|---|---|---|
| 1 | Principles | Linux, HPC, dev tools | bash, SLURM, git, conda |
| 2 | Overall Pipeline 1 | scanner, data, software | DICOM/NIfTI, BIDS, AFNI, fMRIPrep |
| 3 | Overall Pipeline 2 | preprocessing | sswarper2, afni_proc.py |
| 4 | Design & GLM | experimental design, modeling | stimulus timing, GLM, contrasts |
| 5 | Quality Control | qualitative + quantitative QC | APQC, MRIQC, gen_ss_review |
| 6 | Troubleshooting | reading logs, failure types | SLURM, containers |
| 7 | Post-processing | secondary analysis | group, connectivity, ICA, MVPA |
| 8 | Projects & Writing | reproducibility, Methods | NA |
- Syllabus — objectives, prerequisites, and the eight sessions (topics, readings, assignments).
- Slides — per-session teaching slides in Marp-compatible Markdown.
The sessions draw on the fMRI Data Processing Manual on HPC, which contains the step-by-step reference material.
Qiuyu Yu, Human Development and Family Science (HDFS), Auburn University (qiuyuyu.psych@gmail.com)