Optical characterization of tissue for medical diagnostics

Abstract: This thesis deals with the development and evaluation of some different optical techniques for tissue characterization. The main aim has been to improve and evaluate such investigative techniques in in vivo studies. In this attempt a key issue has been to gain understanding in the underlying mechanisms of laser-based diagnostics of premalignant and malignant lesions and to evaluate methods for efficient analysis of acquired data. In particular, photon migration-based methods, aiming at the characterization of thick tissues such as the female breast, and laser-induced fluorescence for detection of superficial malignant and premalignant lesions have been studied. Novel time and frequency domain techniques have been employed to determine the optical properties at multiple wavelengths in healthy breast tissue of volunteers. These techniques have proven to be capable of simultaneously determining the fat, water and haemoglobin concentrations in the tissue. This might be essential in the effort to find a reliable intrinsic contrast for optical breast cancer detection. Another technique, based on interference between photon density waves, is improved and its potential for detection of small inhomogeneities in turbid media is evaluated. The potential of laser-induced fluorescence (LIF) spectroscopy to distinguish between malignant, premalignant and benign lesions in vivo in the colon and in the larynx has been studied. This is an interesting technique, since it could provide information about tissue pathology in real-time during endoscopic examination, thus reducing sampling error in the biopsy procedure. LIF could also help the physician characterize and demarcate lesions for optimal therapy. The technique is shown to be a powerful tool during colonoscopy – fluorescence could to a high degree of accuracy distinguish neoplastic polyps from non-neoplastic polyps and normal mucosa. Also the results from the larynx are promising, lesions could be distinguished from normal, and different lesion types could to some degree be separated. Both autofluorescence and the characteristic fluorescence from d-aminolevulinic acid induced protoporphyrin IX have been studied. Different statistical tools for evaluation of fluorescence emission spectra have been used. In particular, multivariate analysis tools such as multivariate linear regression and partial least squares analysis have been employed. Fluorescence imaging of basal cell carcinomas on the skin, using a multi-wavelength and a fluorescence lifetime system has been performed.