Improving Retinal Disease Detection: Turning OCT Images into Confocal Microscopy

Optical coherence tomography (OCT) and confocal microscopy are crucial tools in eye imaging. OCT gives quick, non-invasive images but can have clarity issues and motion artifacts. Confocal microscopy provides high-resolution, detailed color images but is invasive and raises ethical concerns. To combine the best of both, researchers have created a new method using a 3D CycleGAN. This translates OCT images into the detailed color images of confocal microscopy.

They also made a special dataset called OCT2Confocal, which includes mouse eye images from both methods. This helps in developing and testing cross-modal image translation. Their model scored well in tests like Fréchet inception distance (FID) and Kernel Inception Distance (KID), and people liked the images. It worked better with limited data compared to other methods. This suggests it could be really useful for diagnosing and monitoring eye diseases.

Actions