Related Experiment Video
Updated: Jan 13, 2026
Cross-Modal Multivariate Pattern Analysis
Published on: November 9, 2011
Spatial-frequency dual-constrained Mamba diffusion model for cross-modal generation from CFP to FFA
Qing Liu1, Hongqing Zhu1, Tianwei Qian2
1School of Information Science and Engineering, East China University of Science and Technology, Shanghai, 200237, China.
Abstract:
Fundus Fluorescein Angiography (FFA) is a critical imaging technique for visualizing retinal vascular dynamics and diagnosing various retinal pathologies. However, its reliance on intravenous fluorescein dye may pose risks of nausea, allergies, and even life-threatening complications, with potential adverse reaction hazards. To overcome these limitations, we propose a novel Spatial-Frequency Dual-Constraint Mamba Diffusion Model (SFDC-MambaDiff) for cross-modal generation of FFA images from non-invasive Color Fundus Photography (CFP). Specifically, we design a Learnable Wavelet Frequency-domain Extractor (LWFE) that integrates spiral Mamba scanning to capture frequency-domain vascular features, serving as structural priors. In parallel, a Dual-pyramid Spatial-domain Feature Extractor (DSFE) is developed by combining convolution and average pooling operations with row-column bidirectional scanning to model both global and local spatial representations, which serve as lesion-aware priors. These two domain-specific priors are further integrated through a Dual-domain Conditional Constraint Mamba module (DCCM), which employs a dual attention mechanism to guide the progressive denoising process of the diffusion model. Experimental results on both public and private datasets demonstrate that SFDC-MambaDiff can synthesize high-fidelity, non-invasive FFA images, offering a promising alternative for safe retinal screening and low-cost diagnostic support. The code for this project is available at https://github.com/MISlq/SFDC-MambaDiff.
Insights
This study introduces a new AI model, SFDC-MambaDiff, to create non-invasive retinal images from standard eye photos. This innovation offers a safer alternative to traditional dye-based angiography for diagnosing eye conditions.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Fundus Fluorescein Angiography (FFA) is vital for diagnosing retinal diseases but carries risks associated with intravenous dye injections.
- Adverse reactions to fluorescein dye, including allergies and severe complications, necessitate safer diagnostic alternatives.
Purpose of the Study:
- To develop a novel AI model, Spatial-Frequency Dual-Constraint Mamba Diffusion Model (SFDC-MambaDiff), for generating non-invasive FFA images from Color Fundus Photography (CFP).
- To provide a safe and cost-effective method for retinal screening and diagnostic support, mitigating the risks of traditional FFA.
Main Methods:
- A Learnable Wavelet Frequency-domain Extractor (LWFE) with spiral Mamba scanning captures frequency-domain vascular features as structural priors.
- A Dual-pyramid Spatial-domain Feature Extractor (DSFE) uses convolutions and bidirectional scanning for spatial representations as lesion-aware priors.
- Integration of these priors via a Dual-domain Conditional Constraint Mamba module (DCCM) with dual attention guides the diffusion model's denoising process.
Main Results:
- SFDC-MambaDiff successfully synthesizes high-fidelity, non-invasive FFA images from CFP.
- The model demonstrates strong performance on both public and private datasets, validating its efficacy.
- Generated images maintain crucial diagnostic information without requiring invasive dye injection.
Conclusions:
- SFDC-MambaDiff presents a promising, non-invasive alternative to conventional FFA for retinal imaging.
- This AI-driven approach enhances patient safety and offers potential for widespread, low-cost diagnostic applications in ophthalmology.
- The developed model paves the way for safer, more accessible retinal disease screening.
