Design and implementation of a calibrated hyperspectral small-animal imager: Practical and theoretical aspects of system optimization
Abstract
Pre-clinical imaging has been an important development within the bioscience and pharmacology fields. A rapidly growing area within these fields is small animal fluorescence imaging, in which molecularly targeted fluorescent probes are used to non-invasively image internal events on a gross anatomical scale. Small-animal fluorescence imaging has transitioned from a research technique to pre-clinical technology very quickly, due to its molecular specificity, low cost, and relative ease of use. In addition, its potential uses in gene therapy and as a translational technology are becoming evident. This thesis outlines the development of an alternative modality for small animal/tissue imaging, using hyperspectral techniques to enable the collection of fluorescence images at different excitation and emission wavelengths. In specific, acousto-optical tunable filters (AOTFs) were used to construct emission-wavelength-scanning and excitation-wavelength-scanning small animal fluorescence imagers. Statistical, classification, and unmixing algorithms have been employed to extract specific fluorescent-dye information from hyperspectral image sets. In this work, we have designed and implemented hyperspectral imaging and analysis techniques to remove background autofluorescence from the desired fluorescence signal, resulting in highly specific and localized fluorescence. Therefore, in practice, it is possible to more accurately pin-point the location and size of diagnostic anatomical markers (e.g. tumors) labeled with fluorescent probes. Furthermore, multiple probes can be individually distinguished. In addition to imaging hardware and acquisition and analysis software, we have designed an optical tissue phantom for quality control and inter-system comparison. The phantom has been modeled using Monte Carlo techniques. The culmination of this work results in an understanding of the advantages and complexities in applying hyperspectral techniques to small animal fluorescence imaging. In conclusion, we have identified potential further applications of hyperspectral methods in pre-clinical and clinical imaging.
Degree
Ph.D.
Advisors
Robinson, Purdue University.
Subject Area
Biomedical engineering|Optics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.