Implemented the Marching Cubes algorithm from scratch by reading the original paper and translating the mathematical formulation into a working Python pipeline.

• Processed volumetric CT data (>200 slices per scan) and reconstructed surfaces with ~0.5 mm voxel resolution, generating watertight 3D meshes in STL/OBJ formats.

• Achieved <2 minutes per dataset for full surface reconstruction, reducing manual segmentation effort by ~60% and improving surface consistency across scans.

• Validated mesh fidelity against ground truth slices, maintaining surface error within ±1–2 voxels (~1 mm) while preserving fine geometric detail.

• Enabled downstream applications in visualization, defect localization, and simulation by producing high-fidelity meshes suitable for rendering and finite-element preprocessing.