Neuroimaging Data Formats
10/07/2025
Acknowledgement
- Some of the slides in this tutorial are adapted from the course at UCSF (BI260 Image Processing) and ReproNim Module for Data Processing.
To begin…
Objectives
- Identify data formats that are commonly used in neuroimaging.
- Compare the types of information stored in different data formats in neuroimaging.
- Navigate through a case of automate QA report generation for functional MRI data.
Introduction
- Neuroimaging covers a diverse set of data acquisition instruments
- MRI scanners
- Physiological measurements
- Cognitive tests
- Clinical interviews
Different analysis software often have their own data formats
To ensure information is preserved, it is important to:
- Understand different data formats
- Know what information is stored in each format
Reproducible data processing
Participant Data
- All imaging experiments store some information about participants.
- Demographic data, physiological measurements, clinical assessments, …
- Typically stored in CSV, TSV, or JSON formats.
- Some types of measurements may be stored in proprietary file formats
- In many cases data are also stored in Excel spreadsheets or databases.
- Before converting or extracting the information, it is important to consider what information is stored in each file or database.
- Understanding the variables that are represented
- Datatypes (string, integer, float, double)
- Range (e.g., age > 0), units (e.g., years, mm), etc.
MRI Data: DICOM
- DICOM stands for “Digital Imaging and Communications in Medicine”
Most scanners output data in DICOM format.
DICOM support many medical imaging devices.
DICOM stores 2D, 3D, and 4D data and associated metadata.
- Given the diversity of devices, DICOM has to cover an extensive array of metadata.
MRI Data: DICOM (cont.)
DICOM Standard
- Created during 1980s to 1990s by the American College of Radiology (ACR) and National Electrical Manufacturers Association (NEMA)
- Goal: make radiological data more easily accessible
- Two key elements
- Create interoperable systems
- Imaging device manufacturers
- PACS (Picture Archiving and Communication System) manufacturers
- Image processing and visualization software
DICOM Files
DICOM files contain a header and data
Many software tools are available for parsing DICOM files
- MATLAB
- Pydicom
- OsiriX
- Horos
- MicroDICOM (our focus)
Demo: Load MicroDICOM
- Locate the data folder.
- Locate the
DICOMDIRfile. - Drag
DICOMDIRinto the main panel of MicroDICOM. - Explore how the DICOM files are structured.
- Explore a series in DICOM browser.
- Explore some of the DICOM tags for a selected image.
MRI Data: NIfTI
Much of the metadata in DICOM is irrelevant to existing imaging analysis tools
Most brain imagers typically convert data to NIfTI format
- NIfTI stands for “Neuroimaging Informatics Technology Initiative”
Across neuroimaging, there are many different data formats:
NIfTI1/2(Most imaging tools)CIFTI2(Connectome Workbench)MGH,MGZ,ANNOT,CURV,pial(FreeSurfer)BRIK/HEAD(AFNI)MINC2(MNI Tools)
Most of these formats store data in binary form and cannot be introspected like a text file. You will need to use specialised software to introspect those files.
Demo: Converting DICOM to NIfTI
- The following software will be used:
dcm2niix, for converting DICOMFSL, for viewing NIfTI
- Locate the data folder.
- Execute
dcm2niixin command line. - Explore how the NIfTI files are structured.
- Launch
FSLeyesprogramme. - Explore a NIfTI Image file (
.niior.nii.gz) in FSLeyes. - Explore a NIfTI Metadata file (
.json) and examine some of the attributes.
Brain Imaging Data Structure (BIDS)
- The BIDS standard has been widely adopted for organizing neuroimaging data.
- Specific, human-readable directory structure and file naming conventions
- Promoting data organization, shareability, and compatibility with various analysis tools
Brain Imaging Data Structure (BIDS): An Example
DICOM (left) to BIDS (right)
Quality Assurance for Functional MRI
- To monitor and ensure stability of gradient performance and temporal signal-to-noise ratio (tSNR) over time, for clinical and research fMRI acquisitions.
- Requirements:
- Acquisition should be relatively quick (5-10 minutes) and easy to implement
- Fully automated analysis, to enable efficient handling of large datasets
Quality Assurance for Functional MRI: Pilot Data
Summary
| DICOM | NIfTI | |
|---|---|---|
| Primary Use | Clinical use | Research use (especially in neuroimaging) |
| File Extension | .dcm |
.nii, .nii.gz |
| Data Structure | Mostly 2D and 3D volumes as separate files | 3D or 4D volumes in a single file |
| Metadata | Rich information, embedded in each file | Light-weight, in header and .json sidecar |
| Interoperability | Widely supported in clinical systems (e.g. PACS) | Widely supported in neuroimaging tools (e.g. FSL, MRIQC) |
Relevant Readings
Li X, Morgan PS, Ashburner J, et al. The first step for neuroimaging data analysis: DICOM to NIfTI conversion. J Neurosci Methods. 2016; 264:47-56.
Hagen MP, Provins C, MacNicol E, et al. Quality assessment and control of unprocessed anatomical, functional, and diffusion MRI of the human brain using MRIQC. bioRxiv [Preprint]. 2024.