[Abstract] This work explores the use of paired and unpaired data for training deep neural networks in the multimodal reconstruction of retinal images. Particularly, we focus on the reconstruction of fluorescein angiography ...