Multi-wavelength and time-domain diffuse optical tomography data processing by using a material basis and Mellin-Laplace transform

In order to increase sensitivity in the depth of diffusive media and to separate chromophores with distinct spectral signatures, we developed a method to process time-domain/multi-wavelength diffuse optical acquisitions: 3D Reconstructions of chromophore concentrations are performed with an algorithm based on the use of Mellin-Laplace Transform and material basis. A noise weighted data matching term is optimized by using the conjugated gradients method without expressing the Jacobian matrix of the system. As the algorithm uses reference measurements on a known medium, it does not require measurements or computations of the instrument response function of the system. Validations are performed in the reflectance geometry on a tissue-mimicking phantom composed of intralipid and black ink and a cylindrical blue dye inclusion with a radius of 4mm located at 15mm in depth. The optical tomography setup includes a laser whose picosecond pulses are injected via an optical fiber to the probed diffusive medium and the light collected by two fibers (located 15mm apart from the source), is sent to a Single-Photon Avalanche Diode (SPAD) connected to a Time-Correlated Single-Photon Counting (TCSPC) board. The source and two detectors scan the surface of the medium so as to provide 30 source-detector couples, 900 time-bins and 5 wavelength signals. 3D reconstructions performed on the black ink and blue dye materials on a mesh of around 10000 nodes show that we are able to detect, localize and determine the composition of the inclusion and the background.