Broadband CARS hyperspectral classification of single leukemia cells

Broadband CARS is a coherent Raman scattering technique that provides access to the full biological vibrational spectrum within milliseconds, facilitating the recording of widefield hyperspectral Raman images. In this work, BCARS hyperspectral images of two different unstained leukemic blood cells were recorded and analysed using multivariate statistical algorithms in order to determine the spectral differences between the species. A classifier was trained, which could distinguish the known cells with a 97 % out-of-bag accuracy. The classifier was then applied to unlabelled samples containing a mixture of the two cell types on the same coverslip. This work demonstrates the effective label-free high-throughput single-cell analysis of blood using BCARS. A key feature of this work is the use of an image-based deep-learning cell segmentation algorithm that enables the spectra recorded within a given cell boundary to be integrated producing a high quality single cell spectrum for classification.