Projects

Micro-CT and deep learning were used to develop novel tools for the diagnosis and prognosis of thyroid tumors, addressing the sampling bias inherent to conventional histology and enabling faster, more confident evaluation of thyroid neoplasms.

A 3D U-Net for simultaneous denoising and sharpening of micro-CT scans, trained on 467 heterogeneous volumes across multiple scanners and sample types.

AI-assisted lymph node screening and metastasis detection in whole slide images, enabling automated N-staging for multi-cancer pathology workflows.

A differential phase-contrast microscope with a low-coherence light source

A web-based tool for patchwise inference of tumor diagnostics and BRAF V600E mutation status on 3D micro-CT volumes of thyroid tumors.