- Researchers at The University of British Columbia used denoising diffusion probabilistic models (DDPMs) to generate synthetic retinal images for classifying retinal diseases.
- Augmenting training data with synthetic images improved the performance of a deep convolutional neural network (CNN) ensemble.
- While they noted some issues with image quality, the investigators noted the potential of generative models to enhance deep learning-based medical diagnostics and address data scarcity.
Source: British Journal of Ophthalmology
