Discrete-frequency infrared imaging with a quantum cascade laser and diffuse optical Raman tomography: methods toward improving clinically relevant label-free chemical contrast

Label-free chemical imaging provides the ability to probe chemical structure without interference from dyes or other contrast agents which potentially alter the system being studied. Two emerging, non-invasive spectroscopic methods with significant potential include the use of quantum cascade laser technology for mid-infrared spectral imaging and diffuse Raman tomography for three dimensional chemical analysis. In this thesis, we constructed and evaluated two analytical instruments which represent the first steps toward realizing these technologies. First, a discrete frequency imaging microscope that uses a tunable quantum cascade laser (QCL) source and an uncooled bolometer camera is described. Its performance is compared to a commercial Fourier transform infrared (FT-IR) instrument, and a significant signal to noise advantage over conventional sources was observed when using a spatial filter aperture on the order of the wavelength of light (10μm). We further demonstrate this advantage by illustrating supersampling point mapping which is not achievable with conventional sources. Second, we present experimental results and theoretical considerations for Raman tomography measurements. Using a fiber-coupled Raman spectrometer, spectra were acquired from Teflon spheres embedded in an Intralipid-based tissue-mimicking sample. The instrument’s source-detector collection angles were systematically varied to achieve multiple diffuse projections. Two light fluence modeling methods, radiative transport calculation and Monte Carlo simulation, were compared. Reconstruction of the size and position of buried targets was attempted via an iterative modified-Tikhonov minimization algorithm without the use of spatial priors. The accuracy of these reconstructions was validated using computed tomography (CT) images. We found that although radiative transport methods provide a good first approximation for reconstructions, Monte Carlo methods may have a higher capacity for more accurately modeling the propagation of directionalized light sources over short distances (