High-Quality 3D Tissue Reconstruction in Digital Pathology via Radiance Fields
Master projects/internships - Leuven | Just now
Explore how radiance fields can enable high-fidelity 3D reconstruction of pathological tissue from microscopy data.
Digital pathology is rapidly moving from traditional 2D whole-slide imaging toward 3D tissue reconstruction, enabling better visualization of tissue morphology, tumor micro-structure, and spatial cellular organization. However, current 3D pathology techniques (serial sectioning, confocal microscopy, light-sheet microscopy) suffer from limitations such as sparse sampling, optical distortions, and high acquisition cost.
Radiance Fields approaches such as 3D Gaussian Splatting (3DGS) and Neural Radiance Fields (NeRFs) provide a powerful framework for reconstructing continuous 3D volumetric representations from multi-view 2D images. While such methods have shown impressive results in computer vision and graphics, their potential for microscopy-based and pathology-driven 3D imaging remains underexplored — especially under realistic constraints such as limited views, noise, and optical aberrations.
This thesis proposes to investigate radiance-field-based 3D reconstruction for digital pathology, focusing on reconstruction quality, visual fidelity, and perceptual evaluation of pathological structures.
Type of Internship: Combination of internship and thesis
Master's degree: Master of Engineering Science
Required educational background: Computer Science; Biomedical engineering
For more information or application, please contact the supervising scientist Saeed Mahmoudpour (Saeed.Mahmoudpour@imec.be).
Imec allowance will be provided for students studying at a non-Belgian university.