SNR in nearby galaxies: machine learning segmentation and optical properties

The high spatial resolution observation of nearby galaxies with MUSE has allowed the study of the ionized interstellar medium (ISM) at scales of tens of parsecs. Revealing the different ISM components: HII regions, diffuse ionized gas and supernova remnants (SNR); their properties and how they interact with each other. In this work, we present our recent results for the MUSE observations of the NGC 300 galaxy and the PHANGS-MUSE survey. To unravel the complex ISM structure exposed at these small spatial scales, we have developed a machine learning algorithm to automatically classify the interstellar medium without imposing ad hoc prescriptions such as the [SII]/Ha > 0.4 criteria. The code uses an unsupervised machine learning algorithm to perform a Bayesian Gaussian mixture analysis of the gas component of the galaxy, taking advantage of the rich emission line spectra of these star forming systems. We used different diagnostics diagrams, such as the BPT diagram, in order to check the good performance of our classification, and in addition, we propose new diagnostics diagrams to separate HII regions from SNR. We compare these results with the state of the art Cloudy and Mappings theoretical model in order to extract the physical parameters that control the photoionization and shocks emission.