skip to primary navigationskip to content
 

The potential of hyperspectral imaging in fluorescent contrast enhanced imaging

last modified Jun 18, 2015 12:18 PM
Anna Siri Luthman, Department of Physics and Cancer Research UK Cambridge Institute

The potential of hyperspectral imaging in fluorescent contrast enhanced imaging

A. Siri Luthman, Department of Physics and Cancer Research UK Cambridge Institute

Abstract

Hyperspectral imagers (HSI) -or imaging spectrometers- combine morphological and spectral information to provide detailed information for high specificity readout in biological and medical applications. As HSI provide simultaneous detection in several spectral bands, the technology has significant potential for use in real-time multiplexed fluorescent contrast agent studies. Examples include tumour detection in intraoperative and endoscopic imaging, as well as histopathology. A multiplexed readout from multiple pathological targets, including cell surface receptors overexpressed in cancer cells, could improve both sensitivity and specificity of tumour identification. However, for the most efficient clinical use of HSI several challenges emerge in the optimization of acquisition, analysis and representation of the spatial-spectral data at video rate.

This work evaluates the clinical applicability of a compact, low-cost commercial HSI operating in the near infra-red spectral range, which makes it well suited for fluorescent contrast enhanced imaging. The HSI has been integrated into a platform for wide-field fluorescent contrast agent enhanced reflectance mode imaging with fluorophore specific LED excitation. The hyperspectral imaging performance of the platform has been evaluated, including sensor calibration, wavelength band response and spectral response non-uniformity. Experimentally, a nanomolar detection limit for spectral unmixing of multiple fluorescent contrast agents have been demonstrated. In-vivo evaluation is currently in progress. In parallel with the hardware integration, the implementation of spectral unmixing algorithms and data representation methods are being investigated to establish the most suitable way to rapidly extract clinically relevant information from the hyperspectral data cube. Optimized data management, analysis and hardware integration promises to showcase the potential for improved sensitivity and specificity provided by hyperspectral imaging in a clinical setting.