VMDiff: A novel hybrid CNN-Mamba generative framework for retinal phenotype image enhancement.

Abstract excerpt

Retinal phenotype is crucial for the diagnosis of ophthalmic diseases. Although deep learning models exhibit strong generative capability and detailed reconstruction, they struggle to simultaneously capture fine local details and long-range global dependencies, resulting in poor phenotypic enhancement. To overcome these limitations, we propose VMDiff, a hybrid generative framework based on diffusion models that...

Topics

Share this publication in a Topic to start or enrich a Post.

• Share to Algorithms
• Share to Deep Learning
• Share to Image Enhancement