Edit this page

NA-MIC Project Weeks

Back to Projects List

DeformView: Quantitative Visualization of Non-Linear Deformation Fields for Use in Image-Guided Neurosurgery

Key Investigators

Project Description

We have been developing DeformView, a visualization module for 3D Slicer that improves the interpretation of non-linear brain deformation (“brain shift”) during image-guided neurosurgery and as a training tool for inexperienced surgeons and researchers. DeformView provides two dense, intuitive visualization maps: (1) a dense displacement magnitude map (mm), and (2) a Jacobian determinant magnitude map representing local tissue expansion and compression (%).

The proposed module combines scientifically derived, intuitive colour maps and voxel-wise cursor pointer that directly displays displacement values on hover, a function not available in existing Slicer tools, to improve user understanding and confidence.

Objective

  1. Objective A. Improve user experience and stability by identifying and fixing bugs, refining interactions, and ensuring reliable performance across datasets.
  2. Objective B. Gather user feedback from researchers and clinicians to guide the design of additional features, including potentially adding features to visualize registration error and uncertainty within the module).
  3. Objective C. Integrate transform grid/ glyph visualizations directly into DeformView to provide complementary spatial context alongside dense deformation maps.

Approach and Plan

  1. We will systematically test DeformView across representative datasets (focusing on IGNS-focused data - ReMIND, RESECT, BITE, etc.) and use cases to identify and resolve software bugs. We will ask attendees to use the module to identify common workflows, areas of improvement. We will also perform stress and destructive testing.
  2. User-centered design and feedback: We will conduct structured feedback sessions with expert users, non-expert users, and clinicians, using our targeted questionnaires and short tasks to identify desired features and usability gaps. We will lead discussions with attendees to identify areas of improvement and feature prioritization.

Plan for Project Week

We have implemented the core functionality of the DeformView module, including dense deformation visualization, Jacobian-based expansion/compression maps, and voxel-wise readout on cursor hover. Initial testing confirms that primary visualization goals have been achieved, with only minor usability and stability issues remaining.

We conducted a user study with 10 non-expert participants (average 2.9 years of imaging research experience) to evaluate module functionality. Participants compared DeformView to the existing 3D Slicer Transform Visualizer across four attributes: helpfulness in comprehension, interpretability, intuitiveness, and user confidence, using Likert ratings and the System Usability Scale. On average, DeformView was rated higher across all categories (mean: 4.1/5.0 vs 3.2/5.0), with statistically significant improvements in helpfulness (p=0.008) and intuitiveness (p=0.027). Overall, 80% of participants preferred DeformView over the existing module, confirming the value of our visualization approach.

We plan to address the remaining bugs (that we know of), get user feedback, and engage in discussions with Slicer developers to optimize our module:

  1. Colour Map and Legend Modifications
    • Fix legend scale reload bug
    • Resolve remaining default colour map behaviour, consistent default colour levels when loading new maps
    • Interaction between legend and colour level/window controls
    • Clarity of colour map loading and switching- add descriptive text under the “Color Map” selector
  2. Jacobian-Specific Visualization Controls
    • Jacobian colour legend labels
    • Set Jacobian window and level to constant values to ensure consistent interpretation
  3. User Interface and Readability Improvements
    • Adjust cursor text size for improved readability
    • Implement a full reset of default settings, not limited to window/level

Progress and Next Steps

Updated Our Module README

We’ll make our GitHub public for interested parties in the coming days. See GitHub Repo here: https://github.com/elisedl1/BrainshiftModule

Introduced the ‘Increment Transform’ feature

The transformation is incrementally applied to the moving image over 10 discrete steps. This creates a sliding scale of the transform applied to the underlying image at discrete quantities (ie: 0.1x full transform, 0.2x final transform, etc.)

https://github.com/user-attachments/assets/a7def86b-730b-4f77-9f9a-d078bb9b40a0

Fixed Remaining (Known) Bugs and Functional Errors

  1. Colour Map and Legend Modifications
    • Fix legend scale reload bug
    • Resolve remaining default colour map behaviour, consistent default colour levels when loading new maps
    • Interaction between legend and colour level/window controls
    • Clarity of colour map loading and switching- add descriptive text under the “Color Map” selector
  2. Jacobian-Specific Visualization Controls
    • Jacobian colour legend labels
    • Set Jacobian window and level to constant values to ensure consistent interpretation
  3. User Interface and Readability Improvements
    • Adjust cursor text size for improved readability
    • Implement a full reset of default settings, not limited to window/level

https://github.com/user-attachments/assets/cc53157e-60be-44e1-93b6-ee63ef7c2e34

Started Integration of DeformView and Transforms Visualizer Module

Local UI changes have begun, no functionality is attached yet. Implementing a ‘sparse’ tab (current functionality) and a ‘dense’ tab (added functionality) - but open to feedback.


Sparse Visualization Tab UI (current Transforms Visualizer functionality)
Dense Visualization Tab UI (DeformView added functionality)

Slicer Build on Mac (Silicon) + Documentation

Worked with Steve to get a local build of 3D Slicer on OSX (Silicon). Also wrote documentation to configure, build, run, and debug 3D Slicer from source on macOS (ARM64) using Qt6 and Xcode.

Current Recipe for Slicer

My config:
frolick@IsabelMacBook % sw_vers
ProductName:		macOS
ProductVersion:		15.7.3
BuildVersion:		24G419

frolick@IsabelMacBook % qmake --version 
QMake version 3.1
Using Qt version 6.10.1 in /opt/homebrew/lib

frolick@IsabelMacBook % cmake --version
cmake version 3.31.5

CMake suite maintained and supported by Kitware (kitware.com/cmake).
Prerequisites
Install Qt6

Verify Qt6 Install

qmake --version
  1. If not installed, Install Qt6 (via Homebrew) 
    brew install qt@6
  2. Add Qt6 to shell environment: 
    echo 'export PATH="/opt/homebrew/opt/qt@6/bin:$PATH"' >> ~/.zshrc
echo 'export CMAKE_PREFIX_PATH="/opt/homebrew/opt/qt@6:$CMAKE_PREFIX_PATH"' >> ~/.zshrc
  3. Reload shell: 
    source ~/.zshrc
  4. Verify Qt6 installation: 
    qmake --version
Install Xcode Command Line Tools
xcode-select --install
Install XCode (not strictly necessary (?) but I couldn’t do it without XCode Desktop)

If installing full Xcode from the App Store:

  1. Install Xcode from App Store

  2. Set it as the active developer directory: 
    sudo xcode-select --switch /Applications/Xcode.app/Contents/Developer
  3. Accept the license: 
    sudo xcodebuild -license accept
  4. Verify: 
    xcodebuild -version
xcode-select -p

Else, check the SDK path used by Xcode:

xcrun --show-sdk-path
Directory Setup

Create build directories and set ownership:

sudo mkdir -p /opt/scmake /opt/scd
sudo chown -R $(whoami) /opt/scmake /opt/scd
Create Build Script
vim build_slicer.sh

Paste:

#!/bin/bash

# Configuration
SLICER_SOURCE_DIR="$HOME/slicer/latest/Slicer"
SLICER_BUILD_DIR="/opt/scd"
SLICER_SUPERBUILD_DIR="/opt/scmake"

# Create directories
mkdir -p "$SLICER_BUILD_DIR"
mkdir -p "$SLICER_SUPERBUILD_DIR"

# Clone Slicer source if it doesn't exist
if [ ! -d "$SLICER_SOURCE_DIR" ]; then
    echo "Cloning Slicer repository..."
    mkdir -p "$(dirname "$SLICER_SOURCE_DIR")"
    git clone https://github.com/Slicer/Slicer.git "$SLICER_SOURCE_DIR"
fi

cd "$SLICER_SUPERBUILD_DIR"

# Configure with CMake
cmake \
  -DCMAKE_OSX_ARCHITECTURES=arm64 \ 
  -DSlicer_REQUIRED_QT_VERSION="6.10" \ # CHANGEME: Qt version (qmake --version)
  -DCMAKE_BUILD_TYPE:STRING=Debug \
  -DSlicer_USE_SimpleITK:BOOL=OFF \
  -DSlicer_BUILD_I18N_SUPPORT:BOOL=OFF \
  -DSlicer_BUILD_DICOM_SUPPORT:BOOL=OFF \
  -DSlicer_VTK_SMP_IMPLEMENTATION_TYPE:STRING=Sequential \
  -DCMAKE_OSX_DEPLOYMENT_TARGET:STRING=15.7 \ # CHANGEME: OSX version (About this Mac -> macOS)
  -DCMAKE_OSX_SYSROOT:STRING="$(xcrun --show-sdk-path)" \
  "$SLICER_SOURCE_DIR"

# Build (use -j for parallel jobs, -k to keep going on errors)
make -j10 -k
Run Build
cd /opt/scmake
chmod +x ./build_slicer.sh
./build_slicer.sh
Running Slicer Locally
Option 1: Navigate to the build directory, make, launch:

Executable path: /opt/scmake/Slicer-build/Slicer

cd /opt/scmake/Slicer-build
make -j8 #Optional rebuild to show local changes
./Slicer
Option 2: Run on XCode with Debugging
  1. Create a Dummy Xcode Project Xcode requires an open project in order to attach a debugger.
  2. Open Xcode
  3. File → New → Project
  4. Create a Command Line Tool project (macOS)

  5. Enter an Organization Name (required!)

  6. Get the Slicer Process ID (PID)
    1. Open local Slicer through the build directory (following the steps in Option 1)
    2. In Python Interactor: 
      import os
os.getpid()

      Copy the PID

  7. Attach the Debugger In Xcode:
    1. Debug → Attach to Process by Name or PID…
    2. Paste the PID
    3. Now you can attach process and use debugger functionality (breakpoints, etc.) as usual

Illustrations

Displacement Magnitude Map
Voxel-wise magnitude of non-linear deformation between preoperative T2-FLAIR MRI and intraoperative tumour resection T2-FLAIR, from Case 50 of the ReMIND dataset. Warmer colours indicate larger tissue displacement.

Displacement magnitude map

**Overlay of Displacement Magnitude (Colour Map) & Current 3D Slicer Transform Visualizer Module (Glyphs)

The current Transform Visualizer module (core 3D Slicer module) visualizes deformation as glyphs (arrows), grid, and contour. When integrated together, it is more intuitive where deformation has occurred (DeformView) and the direction of deformation (Transform Visualizer).

NAMIC-Overlay

Jacobian determinant magnitude map
Visual of the Jacobian map, where red indicates tissue expansion and blue is tissue compression, as a percentage. This is the same data as the above displacement magnitude example.

Jacobian_mag

User Study Results
Comparison of DeformView with the existing Transform Visualizer module (n=10) across four attributes: helpfulness, interpretability, intuitiveness, and user confidence (1–5 scale; higher scores indicate better performance). DeformView is rated higher across all categories, with significant improvements in helpfulness and intuitiveness.

User study results

Background and References

Miner, R. C. (2017). Image-guided neurosurgery. Journal of Medical Imaging and Radiation Sciences, 48(4), 328–335.

Abhari, K., Baxter, J. S., Chen, E. C., Khan, A. R., Peters, T. M., De Ribaupierre, S., & Eagleson, R. (2014). Training for planning tumour resection: augmented reality and human factors. IEEE Transactions on Biomedical Engineering, 62(6), 1466–1477.

King, F., Lasso, A., & Pinter, C. (2015, August 4). TransformVisualizer (Documentation/Nightly/Modules). 3D Slicer Wiki. Link

Vlachogianni, P., & Tselios, N. (2022). Perceived usability evaluation of educational technology using the System Usability Scale (SUS): A systematic review. Journal of Research on Technology in Education, 54(3), 392–409.

Drouin, S., Kochanowska, A., Kersten-Oertel, M., Gerard, I. J., Zelmann, R., De Nigris, D., … & Collins, D. L. (2017). IBIS: an OR ready open-source platform for image-guided neurosurgery. International Journal of Computer Assisted Radiology and Surgery, 12(3), 363–378.

Chung, M. K., Worsley, K. J., Paus, T., Cherif, C., Collins, D. L., Giedd, J. N., … & Evans, A. C. (2001). A unified statistical approach to deformation-based morphometry. NeuroImage, 14(3), 595–606.

Juvekar, P., Dorent, R., Kögl, F., Torio, E., Barr, C., Rigolo, L., … & Kapur, T. (2024). REMiND: The brain resection multimodal imaging database. Scientific Data, 11(1), 494.

Crameri, F., & Hason, S. (2024). Navigating color integrity in data visualization. Patterns, 5(5), 100972. doi:10.1016/j.patter.2024.100972