RKSPProjectOverviewToolsused

RKSP Project Overview and Tools used

Tools and Technologies Explored

1 MONAI SDK

Overview: Open-source framework built on PyTorch for developing AI models in healthcare imaging.

Resources: Tutorials and examples available on GitHub for various use cases, including DICOM to image conversion apps​ (GitHub)​​​​ (MONAI)​​ (MONAI)​.

2. 3D Slicer for MRI/CT Data

• Explored The 3D Slicer UI, imported MRI data, used the Segment Editor, Level Tracing, and Scissoring tools, constructed 3D models, and applied Gaussian and median smoothing.

Quick tutorial : 3D Slicer \ Resources : Initial exploration of CT data using 3D slicer

3. InVesalius

Details and Performance: Similar to 3D Slicer Excellent for creating detailed models; takes around 8 minutes to load DICOM data.

4. Medivis

Medivis provides technology that converts 2D medical images, like MRI and CT scans, into 3D holographic visualizations.

• On-premise IT Deployment: Ensures patient data remains within the hospital firewall.

• HIPAA Compliance: Seamless, secure PACS integration.

• No Cloud Storage: No PHI stored on the cloud.

5. Weasis

• Cross-Platform DICOM Viewer: Standalone and web-based viewer with a highly modular architecture.

• Integration: Flexible integration with PACS, RIS, HIS, or PHR.

• Performance: High-quality renderings with high performance through the OpenCV library.

6. Horos Project :

-Open-source medical image viewer for MacOS.

- 3D Reconstruction of CT Scans: Allows rotating and zooming, adjusting transparency, and interacting with the model tissue using the cursor to cut or remove tissue.

- Platform: Available for macOS only.

- Resources: For a brief tutorial, you can watch this video, and the source code is available on GitHub.

Deepdive analysis of The repository for the Horos project

7. VolView

• VolView is a fast, lightweight radiological viewer supporting multiple DICOM options, with upcoming support for DICOM SEG and RTSTRUCT. It offers high-quality, interactive 3D cinematic volume renderings with presets, lighting adjustments, transparency control, cropping, and data interaction.

• VolView requires no installation and runs entirely on the user's machine using local CPU, GPU, and disk resources, ensuring data security by keeping all data on the user's machine.

• source code is available on Github

Libraries and Tools Implementation

Used 3D slicer to Produce a 3D volume mesh of the CT scan outputs in STL format

STL to FBX Conversion for Unreal Engine

Blender: Smoothed out rough edges and valleys in STL files before exporting to FBX format.

Performance Notes : Blender : Swift handling of STL files.

• Unreal Engine : Takes 10-12 minutes to load FBX files.

Dicom to 3D Using ITK and VTK using python libraries

Loaded DICOM images into a 3D volume using SimpleITK, resampled to new size and spacing, applied a binary threshold to highlight relevant voxels, cropped to a specific region, and rendered the volume with VTK for visualization.

VTK.js

VTK.js: Utilized for developing a web-based app

Examples Worked On:

standalone.html: Renders a cone.

src/cone.js: Renders a cone using a webpack build approach.

src/cone-filter.js: Demonstrates VTK.js filters.

src/volume.js: Demonstrates basic volume rendering without transfer functions.

src/volume-transfer.js: Demonstrates basic volume rendering with transfer functions.

src/image-slicing.js: Demonstrates volume slicing and interaction.

src/widgets.js: Demonstrates a volume cropping widget.