A fully featured studio for reconstructing CT-scans using Marching Cubes and Gaussian Splatting
GaussLab is a fully-featured studio for reconstructing CT-scans using classical approaches like Marching Cubes and SOTA approaches, Gaussian Splatting. GaussLab is fully GPU-accelerated.
This project was developed as our Bachelor's Graduation Project at Ain Shams University for the academic year 2024/25, You can read the full thesis here: GaussLab - Bachelor's Thesis
- 3D reconstruction using Marching Cubes and Gaussian Splatting
- Multi-viewport viewing
- AI Assistant for Diagnosis
- Object slicing
GaussLab uses the CMake build system. Install the following prerequisites first:
- glm
- LibTorch, PyTorch's C++ frontend
Then create a build directory and run:
cmake ..
make- CUDA backend for rasterization
- Segmentation models