Cancer remains one of the most important contributors to premature mortality at the global level. The elastic properties of cells and tissue have been shown to correlate with normal, dysplastic, and cancerous states. In this work, we rely on time-resolved Brillouin scattering to characterise cancerous and normal cells with contrast provided by their elastic properties. In doing so, we achieved proof of concept that artificial intelligence can be used to differentiate between cancerous and normal cell lines with a low number of highly localised measurements. A differentiation accuracy of 93%, was obtained probing in a volume of a few microns corresponding to a single phonon measurement. Our findings suggest the possibility of potential applications for diagnostics.
|