Unsupervised data driven approaches to Raman imaging through a multimode optical fiber

Liam Collard,Mohammadrahim Kazemzadeh,Massimo De Vittorio,Ferruccio Pisanello

2024-05-22

Abstract:Raman spectroscopy is a label-free, chemically specific optical technique which provides detailed information about the chemical composition and structure of the excited analyte. Because of this, there is growing research interest in miniaturizing Raman probes to reach deep regions of the body. Typically, such probes utilize multiple optical fibers to act as separate excitation/collection channels with optical filters attached to the distal facet to separate the collected signal from the background optical signal from the probe itself. Although these probes have achieved impressive diagnostic performance, their use is limited by the overall size of the probe, which is typically several hundred micrometers to millimeters. Here, we show how a wavefront shaping technique can be used to measure Raman images through a single, hair thin multimode fiber. The wavefront shaping technique transforms the tip of the fiber to a sub-cellular spatial resolution Raman microscope. The resultant Raman images were analyzed with a variety of state-of-the-art statistical techniques including PCA, t-SNE, UMAP and k-means clustering. Our data-driven approach enables us to create high quality Raman images of microclusters of pharmaceuticals through a standard silica multimode optical fiber.

Optics,Chemical Physics

What problem does this paper attempt to address?

This paper attempts to solve several key problems encountered in Raman imaging through a single multimode fiber. Specifically, these problems include:

1. **Suppression of background signal**: When using a multimode fiber for Raman imaging, the fiber itself will generate a relatively strong background signal, which is usually 1 - 2 orders of magnitude higher than the analyte signal. This background signal will interfere with the detection and analysis of the analyte signal.

2. **Improvement of spatial resolution**: Due to the random spatial distribution of light in the fiber, traditional multimode fibers result in relatively low spatial resolution for imaging. The method proposed in the paper aims to convert the tip of the fiber into a Raman microscope with sub - cellular spatial resolution through the wavefront - shaping technique, thereby improving the imaging resolution.

3. **Optimization of data analysis**: In order to extract useful information from complex Raman signals, the paper adopts a variety of advanced statistical methods, including principal component analysis (PCA), t - distributed stochastic neighbor embedding (t - SNE), uniform manifold approximation and projection (UMAP), and k - means clustering. These methods are helpful for separating the background signal and the analyte signal and generating high - quality Raman images.

4. **Feasibility of surgical applications**: The paper also explores the potential application of these techniques in the surgical environment, especially for application scenarios requiring high sensitivity and high resolution, such as cancer diagnosis. By using a single multimode fiber as thin as a hair, high - resolution Raman imaging can be achieved without increasing the size of the probe, thus meeting the requirements of the surgical environment.

In summary, the main objective of this paper is to achieve high - resolution and high - sensitivity Raman imaging in a single multimode fiber by combining the wavefront - shaping technique and advanced data - processing methods, thereby providing new tools and technical means for biomedical research and clinical applications.