This page under construction, beware...
See this link for a publically-commentable draft of the future page...

5 Steps to More Reproducible Neuroimaging Research

Going whole-hog into completely reproducible neuroimaging is hard. But, there are lots of little things you can do today to increase the reproducibility of your current studies.

Study Design

repronim.org/5Steps/design

• Build data shareability into the consent process - see Open Brain Consent, for example.
• Plan your study with a RoadMap - see, for example, BrainVerse.
• Pre-register your study - at, for example, Open Science Framework.

Data Collection

repronim.org/5Steps/acquisition

• Adopt standards-based data representation from the get go.
• Use ReproIn to automate conversion of your imaging data to BIDS.
• Annotate your metadata (all experimental details) as you collect it - see, for example, BrainVerse.
• Use a version control system for all files (data, code, environments, etc.) - i.e. DataLad.

Data Processing

repronim.org/5Steps/processing

• Document your software and computational environment - for example NeuroDocker and ReproMan.
• Use containerization/virtualization to encapsulate and share analysis workflow (Docker, Singularity, Virtual Machines, etc.).
• Annotate your results - using NeuroImaging Data Model (NIDM), for example.

Statistical Analysis

repronim.org/5Steps/stats

• Most studies are underpowered, combat this by sharing your data and reusing available data to enable creating larger data sets, thereby increasing study power. Understand:
  • Effect size, confidence intervals, power, positive predictive value and significance testing - some training materials can be found here.

Publication

repronim.org/5Steps/publication

• Include the anonymized raw data and workflow used - using resources such as zenodo, etc.
• Include complete processed results
• Make it Findable, Accessible, Interoperable and Reusable (FAIR)