Quantification of Brain Oxygen based on Time and Space Optimization of Diffuse Optics: Monte-Carlo Inversion of Infrared Spectroscopy on Phantoms

Optical spectroscopy techniques are frequently used for the quantitative analysis of substance
composition, physical properties, and the phenomena risen from energy-matter interactions.
Near-infrared absorption spectroscopy of human tissues is particularly of medical interests for
its ability to selectively acquire physiological and neurological information of certain
chromophores in the human body, with the help of photons that are non-invasively propagated
through organs such as brain or kidney. The implementation of these techniques, despite many
research activities in this area, still encounters massive difficulties primarily due to one simple
fact: Light no longer travels in a straight line in tissues but rather spreads on stochastic
trajectories due to the strong scattering. The optical scattering effect dominates in these media
and to some extent is entangled with the optical absorption. While the optical absorption
spectroscopy is what most quantitative analyses rely on, the interpretation and analysis of the
results are difficult due to this entanglement with scattering and this problem has not yet been
satisfactorily solved. Biomedical applications of the techniques are further complex by the
geometry and heterogeneity of tissues at almost every length scale, which results in the illposedness
of solving the underlying inverse problem and thereafter usually the solution’s nonuniqueness
when retrieving the diagnostical information.
The present thesis is devoted to disentangle the effects from absorption and scattering in
human brain and purpose an innovative approach on quantifying the optical absorption and
scattering coefficients with improved accuracy. Especially, a new concept of integrating
disparate data types from various measurement domains is proposed and verified. The work is
based on a fundamental fact: The absorption and scattering, despite heavily entangled, are
essentially independent. And the complementarity encoded in the measurements of different
domains can be advantageously used to increase the retrieval accuracy of the unknowns and
reduce the complexity of the inversion.
The thesis realizes the concept in the term of spatial-enhanced time domain diffuse optics.
By deploying picosecond pulse laser and time-correlated single photon counting technique, the
approach is validated on homogeneous solid and two-layered liquid phantoms mimicking
human brain’s optical properties. Monte-Carlo simulations are applied to imitate photon
random transport in turbid media and are incorporated into the spatio-temporal optimization of
the inversion process. The estimation accuracy of absorption and scattering coefficients is
demonstrated at level of 5%. The examined and presented concept and computational method have the potential to overcome the challenges of the inverse problem in diffuse optics such as
solution’s non-uniqueness and deep scattering neutrality.