On the Estimation of Optoacoustic Waves in Retinal Laser Therapy Using Gaussian Processes

In this paper, we study potentials of machine learning in terms of Gaussian processes (GPs) to learn optoacoustic (OA) waves generated in eye globe during retina laser therapy. This can be utilized to improve temperature measurements for feedback laser control and to understand the tissue properties of the irradiated sites on the retina. The treatment of each site is performed in two phases: The first one is used for GP learning based on measurements from an ultrasonic transducer and its dynamical model, then, the learned GP is employed for estimating the waves during the second phase where the treatment process is activated. The results demonstrate the effectiveness of the approach.