Tumor detection in various parts of the body by optical tomographic methods is often a limited data problem with only back-scattered light available for analysis of the subsurface. The reconstruction challenge is to obtain reconstructions for small tumors (characteristic of early cancer stages), while facing problems of clutter in the tissue environment in addition to data SNR issues and the inherent ill-posedness of the reconstruction. This prompts the approximate-tomographic reconstruction of the shape and typical optical parameters of the affected tissue rather than a full pointwise estimate.

For the reconstruction problem in cervical cancer detection, we propose an alternative level set based reconstruction approach to obtain the optical-property profile of the subsurface tissue from above-surface measurements of the back-scattered light, while keeping the number of reconstruction unknowns small. For this problem, we investigate the suitability of the diffusion model of propagation with respect to that based on the radiative transfer equation. A spatially resolved fluorescence set up in frequency domain and DC modes will be developed. The fluorescence spectra at various positions will be used to reconstruct a layered inhomogeneity in solid layered phantoms as well as cervical tissue, subsequent to working with a homogeneous liquid phantoms to verify the model used.