DSI Studio 2025 Workshop Outline

Each workshop will run for 1 hour, divided into four 15-minute sections. During each section, feel free to drop your questions in the Zoom chat at any time. I’ll go through them one by one during the short breaks between sections.

The sessions will be recorded and posted on YouTube afterward. If you prefer, you’re welcome to just watch the recordings instead of attending live—live participation is mainly for real-time interaction.

At the end of each day’s workshop, we’ll have an informal, off-the-record Q&A. You can ask about your own research, bring up ideas, or suggest features you’d like to see in DSI Studio. It’s an open forum to chat and brainstorm together.

• Interface and Fiber Data Hub (WK1)

The new Command History feature records GUI actions and converts them into reusable command-line scripts, making it easy to reproduce and automate workflows. The Fiber Data Hub provides instant access to over 37,000 preprocessed brain fiber datasets, including HCP, ABCD, and OpenNeuro. The hub is hosted independently on GitHub and accessible via API, allowing users to explore brain connectivity without manual preprocessing.

• Minimal Preprocessing Pipeline (WK2)

This session introduces key diffusion MRI acquisition settings, and how they influence reconstruction and analysis. We’ll then walk through a minimal preprocessing pipeline that converts raw data (DICOM or NIfTI) into analysis-ready .sz files using either the GUI batch processor or command-line tools. The pipeline emphasizes simplicity and compatibility, making it suitable for a wide range of datasets, including both human and animal studies.

• Fiber Tracking (WK3)

This session presents updated tracking metrics, including tract-to-voxel and seed-to-voxel ratios, which improve interpretation of tract counts. We’ll introduce the tract-to-region Connectome, which links tracts to specific anatomical regions, offering finer resolution than region-to-region matrices. The session also covers how to use different region types—volume ROIs, surface-based regions, and atlas-defined labels—for fiber tracking and filtering. New features enable tractography on structural images aligned to diffusion space, and updated average templates support improved spatial normalization in group studies.

• Differential and Correlational Tractography (WK4)

This session covers statistical tools for detecting group differences and correlations in white matter tracts. Differential tractography identifies regions with significant changes, while correlational tractography maps associations with variables like age, symptoms, or cognitive scores. These methods support applications in neurology, psychiatry, and developmental neuroscience by linking tract-level findings to clinical or behavioral measures.

Session 1: Introducing DSI Studio

DSI Studio is built for simplicity, transparency, and accessibility—providing researchers and clinicians with an intuitive platform for diffusion MRI tractography and connectome analysis.

1.1 Design Paradigm

• 🔧 Minimal Preprocessing
DSI Studio follows a minimalist preprocessing philosophy, using only essential steps like FSL’s TOPUP and eddy.

• 📈 Diffusion Models
The software supports models including DTI, GQI, and QSDR.

• 🔁 Actively Maintained and User-Driven
Frequent updates reflect direct user feedback through the public forum. Most new features are developed in response to real research needs.

• 💡 Concept-driven Methods
DSI Studio introduces concept-driven methods such as tract-to-region connectome, differential tractography, and correlational tractography.

1.2 Community Support

1.3 Version History

DSI Studio began as a research-focused tool—and it continues to evolve with that mission in mind. The software is actively maintained and updated, typically on a weekly basis.

Keep a working copy of DSI Studio alongside your data, as the software is updated frequently (often daily!).

1.4 Setup & Installation

Session 2: Main Interfaces

Tractography Tabs:

Raw images (DICOM, NIFTI, etc.) 📂 ➝ .sz file (DWI signals + bval/bvec) 📊 ➝ .fz files ( fiber orientations + anisotropy) 🖼️ ➝ .tt.gz ( tractography) 🧠

File Formats:

Fiber Data Hub

Overview:
The Fiber Data Hub is a growing repository of preprocessed diffusion MRI data designed for ease of access and integration with DSI Studio.

Web portal: https://brain.labsolver.org

Key Features:

Console Window:

Command Line Interface:

Example code to run at Google Colab (Ubuntu 2204)

!wget https://github.com/frankyeh/DSI-Studio/releases/download/2024.06.12/dsi_studio_ubuntu2204_cpu.zip && unzip -o -q dsi_studio_ubuntu2204_cpu.zip
!wget https://github.com/data-openneuro/brain/releases/download/ds004299/sub-103_ses-1_dwi.sz
!wget https://github.com/data-openneuro/brain/releases/download/ds004299/sub-104_ses-1_dwi.sz
!dsi-studio/dsi_studio --action=rec --source=*.sz 
!dsi-studio/dsi_studio --action=ana --source=*.fz --region=HCP-MMP:L_V1

Session 3: Tractography Interfaces

Assignments

Assignment 1. Screen capture whole brain tractography from 10 subjects using command history function

  1. ** Download all subjects’s first session gqi.fz files from OpenNeuro ds000244 (Individual Brain Charting)
    • Open [Fiber Data Hub] → Repository: [OpenNeuro] → [ds000244]
    • Click on [Select Matching] button → gqi.fz (After entering the filter, move the scroll bar a bit to see the selected data)
    • Specify the Save Directory and Click on [Download 12 File(s)]
  2. Screen capture whole brain tractography on downloaded files
    • At main window’s [Tractography] tab, click on [Step T3: Fiber Tracking] and select the downloaded file sub-01_ses-00_dwi.gqi.fz
    • Click on the [Fiber Tracking] button on the right to get whole brain tractography
    • Adjust zoom value in the 3D window to get a good view
    • On the top menu, save whole brain tractography using [Screen][Save 3D Screen] → save screen shot as sub-01_ses-00_dwi_whole_brain_screen.jpg
    • On the top menu, select [Records][Command History] → Select all steps from open_fib to save_screen → Click [Apply to Others…] → Select other .fz files
    • On the top menu, select [File][Open FIB Directory] to see if the screenshots are generated.

Assignment 2: Batch Processing NIfTI Files to Reduce File Size

Objective:

Reduce the size of T1-weighted images by adjusting intensity, data type, and spatial resolution.

Step-by-Step Instructions:

  1. Download the Original OpenNeuro ds001378 Dataset
  2. Collect T1-Weighted Images
    • Navigate to the downloaded dataset’s SCA2 folder.
    • Search for all files named *T1w.nii.gz and copy them into a new working folder.
  3. Open Images in DSI Studio
    • Launch DSI Studio.
    • Go to [Tools] → [Step O1: View Images]
    • Select all T1w.nii.gz files in the folder.
    • Note: DSI Studio will preview only the first image, but the actions will later apply to all.
  4. Adjust Image Contrast
    • In the top menu, go to:
      • [Views] → [Normalize Otsu Median]
      • [Views] → [Upper Threshold], enter 1
  5. Reduce Data Precision
    • In the top-right dropdown menu, change the image type from 32-bit floating point to 8-bit integer
  6. Downsample Image Volume
    • Go to [Volume] → [Resize]
    • Enter new dimensions: 200 256 160
  7. Save the Processed Images
    • Go to [File] → [Save as…]
    • Use a new filename with a suffix, e.g., *_T1w_reduced.nii.gz
  8. Apply Settings to All Files
    • When prompted, click [Yes] to apply the same changes to all selected images.

Assignment 3: Command Line Batch Processing NIfTI Files to Reduce File Size

  1. Open the console window in DSI Studio

  2. Repeat the steps in Assignment 2 to identify the commands and parameter used in NIFTI tool.
    • The command usually has this format name of the command:parameters
    • For example, if regrid at 1 1 1, the name of the command is regrid and the parameters are 1 1 1

  3. Use command line to repeat the task of the assignment 2

    Example command:

    dsi_studio --action=img --source=*.nii.gz --cmd="flip_x+upper_threshold:1+change_type:1" --output=../*_modified.nii.gz

    This command batch-processes all .nii.gz images in the current folder by flip the image at x direction, thresholded values beflow 1, and change the pixel type to 16-bit integer before sacing the results with a _modified suffix in the parent directory.