Unsupervised Hyperspectral Stimulated Raman Microscopy Image Enhancement: Denoising and Segmentation via One-Shot Deep Learning

TL;DR

This paper introduces UHRED, an unsupervised deep learning method that denoises hyperspectral SRS images and performs automatic segmentation with minimal data, enhancing biomedical and mineralogical imaging.

Contribution

The novel UHRED network enables one-shot denoising and segmentation of hyperspectral SRS images without prior training or labeled datasets.

Findings

Effective denoising of hyperspectral SRS images with a single image.

Automatic segmentation producing chemical species maps.

No need for pre-labeled training data.

Abstract

Hyperspectral stimulated Raman scattering (SRS) microscopy is a label-free technique for biomedical and mineralogical imaging which can suffer from low signal to noise ratios. Here we demonstrate the use of an unsupervised deep learning neural network for rapid and automatic denoising of SRS images: UHRED (Unsupervised Hyperspectral Resolution Enhancement and Denoising). UHRED is capable of one-shot learning; only one hyperspectral image is needed, with no requirements for training on previously labelled datasets or images. Furthermore, by applying a k-means clustering algorithm to the processed data, we demonstrate automatic, unsupervised image segmentation, yielding, without prior knowledge of the sample, intuitive chemical species maps, as shown here for a lithium ore sample.