Responsibilities include:
- Engineered advanced digital pathology image and segmentation mask generative models utilizing shape-based constraints, with architecture designed for seamless extension to 3D volumetric applications.
- Pioneered high-fidelity digital pathology image synthesis techniques implementing denoising diffusion probabilistic models, achieving superior texture and clinical feature preservation.
- Developed robust evaluation metrics for synthetic medical images to assess their utility in downstream machine learning tasks, leveraging large-scale foundation models to ensure reliable performance benchmarking.