The positive results of the feasibility study of the EDICAM visible camera system on JT-60SA [1] motivated further study of possible synchrotron radiation images from this camera system. A convolutional neural network is being developed for the detection and analysis of synchrotron radiation in camera images. The neural network is trained on images combined from simulated synchrotron radiation images from SOFT [2] and background plasma emissions from CHERAB. The SOFT images are simulated similarly to the EDICAM feasibility study, with plasma parameters relevant to the JT-60SA tokamak, but the electron distribution function is taken as an analytical distribution function as derived in [3]. In this presentation, I show the first results from the neural network aimed at indicating the presence of runaway electrons and possibly gaining information on the q-profile of the tokamak plasma. References: [1] S. Olasz et al. FED, 195 113940 (2023) [2] M. Hoppe et al. NF, 58 026032 (2018) [3] Fülöp et al., PoP, 13 (6), 2006