Generative Adversarial Learning for OCT-TPM Vascular Domain Translation

Modeling of microscopic cerebrovascular networks is essential for understanding cerebral blood flow and oxygen transport. High-resolution imaging modalities, e.g., Optical Coherence Tomography (OCT) and Two-Photon Microscopy (TPM), are widely used to capture microvascular structure and topology. Despite TPM angiography providing better localization and image quality than OCT, its use is impractical in studies involving fluorescent dye leakage. Here, we exploit generative adversarial learning to produce high-quality TPM angiographies from OCT vascular stacks. This will serve as a complementary tool to enhance vascular analysis when only OCT imaging is involved. We investigate the use of 2D and 3D cycle generative adversarial networks (CycleGANs) trained on unpaired image samples. Our results demonstrate a successful generative ability, with a high structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR), of the 2D adversarial learning over that relying on 3D learning.