Spectral Phenotyping.

Researchers

Journal

Modalities

Models

Abstract

The goal of the proposal is to develop spectral phenotyping, a non-destructive, label-free method to identify cell types and altered call states based on their spectral properties. Spectral signatures are acquired by scanning a cell with infrared (IR) light to computationally learn what features identify a cell type, a method conceptually akin to face recognition software. The technology integrates Fourier transform infrared (FTIR) spectromicroscopy and deep learning to distinguish among cell types (e.g., neuron versus astrocyte) or cell states (e.g., disease versus normal). FTIR signatures alone rapidly classify living unlabeled human disease cells.We propose that identification of cell subtypes by spectral phenotyping will be possible with a precision and reliability that has not previously been achieved using conventional approaches.
© Association of Biomolecular Resource Facilities.