6.01] Prostate cancer diagnostics using spatially resolved fluorescence-based optical imaging

Abstract

Objective Prostate cancer (PCa) is the second leading cause of cancer-related deaths in men. An ideal imaging modality should identify the location of the aggressive cancer cells with 100% accuracy. So far this has not been achieved. The method we propose suggests the use of “smart probes” that specifically activate multiple fluorophores when found in the presence of PCa cells. Fluorescence detection can be performed with high sensitivity using fairly cheap equipment compared to other imaging modalities. Material and methods Fluorophores emitting between 600 and 850 nm are ideal labels. The tissue absorbs little and scatters most of this light, exhibiting a high signal-to-noise ratio. Different wavelengths offer a deeper, spatially resolved resolution. Optical imaging is impaired by the penetration depth of the excitation light and the geometrical restrictions of the detectors. Results Determination of the optical properties of the involved tissue is necessary for precise calculations of the light propagation. Goniometric measurements retrieve the scattering function and collimated transmission experiments provide the attenuation coefficient. Forward model: the light propagation through tissue is determined analytically using the diffusion model approximation and the a priori known optical properties. Results are available for simple parallelepiped geometry. The calculations will be generalized using a Finite element model (FEM) for more complex geometries found in the real environment. Inverse problem: derives the locations of the tumor, i.e. the fluorophores, within the prostate. The numerical 3D-reconstruction algorithm is based on the Adjoint theorem that extracts the exact locations of the fluorescing light sources by a single step linear calculation. The method relies on the equivalent readings when the detector and the light source exchange positions in space. Using the simplified parallelepiped model, the exact locations of the fluorescent light sources have been identified analytically for a determined system and partially determined for an underdetermined system. Conclusion The absolute determination of the tumor locations is dependent on the information gathered by the recorded data. For a determined system, i.e. the number of detectors equals the number of fluorescent light sources in the phantom; the locations have been extracted accurately. In general, the number of unknowns (i.e. fluorescent locations) is greater than the number of detectors; so the system is underdetermined. In this case the data have to be linearly independent so maximum information output is achieved. In the real world we are faced with underdetermined or ill-posed systems. Mathematical methods like Tikhonov regularization are employed for the calculation of the inverse solution. Depending on the sought spatial resolution, the percentage of the retrieved locations can be 100% or less. The spatial arrangement for detectors and the choice of emission wavelength of the fluorophores are the key factors in obtaining higher information content for the measured data.