Label-Free Intracellular Temperature Measurement by Integrating Raman Imaging and Deep Learning

Abstract We developed a label-free method for visualizing intracellular temperature distribution using Raman image analysis with deep learning. Analysis of the entire region of the O-H stretching band of water using deep learning achieves an accuracy of ∼0.2°C, allowing intracellular temperature mapping without any preprocessing such as fluorescent labeling. Experiments using photoexcitation of carbon nanotubes attached to living cells demonstrate that this method can obtain transient changes in the temperature distribution within a cell.