LDASeg
Latent diffusion for ambiguous medical image segmentation.
Project
LDASeg: Latent Diffusion for Ambiguous Medical Image Segmentation was a collaborative course research project in Dartmouth’s Deep Learning Robustness course (COSC 89.31/189.31).
The project studied probabilistic segmentation when multiple plausible annotations exist. I contributed evaluation experiments and analysis, with emphasis on comparing predictive diversity and segmentation quality under ambiguity.
Status
This work received a Best Class Project Citation. It is listed as course research—not as a peer-reviewed publication—to keep the public record precise.