Fluorescein Angiography Transformation via Multimodal Generative Adversarial Network with Misaligned Data

Fluorescein angiography (FA) is a diagnostic method for observing the vascular circulation in the eye; however, it poses a risk to patients. Generative adversarial networks (GANs) have been used to convert fundus structure (FS) images into FA images. Current high-resolution image generation methods employ complex deep network models that are challenging to optimize, leading to issues such as blurred lesion boundaries and poor capture of microleakages and microvessels. In this study, we solved the problem by developing an novel multiple- ResNet GAN to improve model training, thereby enhancing the ability to generate high-resolution FA images. Based on a comparison with state-of-the-art methods, the results show that our method can improve the generation of detailed regions in high-resolution FA images. Thus, it shows a promising clinical diagnostic value in future research.