Fast Optical Coherence Tomography Image Enhancement using Deep Learning for Smart Laser Surgery: Preliminary Study in Bone Tissue

One of the most common image denoising technique used in Optical Coherence Tomography (OCT) is the frame averaging method. Inherent to this method is that the more images are used, the better the resulting image. This approach comes, however, at the price of increased acquisition time and introduced sensitivity to motion artifacts. To overcome these limitations, we proposed an artificial neural network architecture able to imitate an averaging method using only a single image frame. The reconstructed image has an improvement in the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) parameters compared to the original image. Additionally, we also observed an improvement in the sharpness of the denoised images. This result shows the possibility to use this method as a pre-processing step for real-time tissue classification in smart laser surgery especially in bone surgery.