Computational and experimental methods for the assessment of tomographic optical microscopy in the II Near Infrared Window and in Low Scattering Media

Abstract

Light microscopy in the second near infrared (NIR-II) window has shown to be a very promising approach to increase the depth of penetration and reduce the loss of resolution of deep tissue images in Light Sheet Fluorescence Microscopy (LSFM) and Optical Projection Tomography (OPT). In this spectral window the scattering of light in biological tissue samples is much weaker than in the visible, where this phenomenon limits the penetration to a few hundreds of microns for in vivo specimens and non cleared tissues. However, measurements in the NIR-II demonstrated to fall in the yet uncharacterized ballistic to diffusion transition, where neither the Diffusion Approximation (DA) nor the ballistic model can accurately predict the propagation of light. This thesis investigates the performance of OPT and LSFM in this spectral window and equivalent low scattering media through the development of a set of computational tools and experimental approaches to characterize the penetration andchallenges of imaging in this regime. The characterization begins with an evaluation of the validity of the DA in low scattering media using a new Monte Carlo (MC) simulation method to estimate the forward flux density of point and collimated sources in infinite media. The proposed tool allowed to demonstrate the weaknesses of the DA to model light propagation below a transport mean free path. In order to study LSFM in the NIR-II , this work presents a Monte Carlosimulator capable of mimicking the entire photon flow of a LS microscope. The software was developed with the combination of a modified version the validated MCX package with a novel algorithm that focuses the detected fluorescence and computes the optical sectioning according to the position of the LS in the volume. The tool was used to simulatelight sheet acquisitions of a distribution of fluorophores in a large volume with the optical properties of tissues from several spectral windows. The results showed that LSFM is expected to resolve structures at depths of at least one transport mean free path in scattering media, which demonstrates that the predictions of the theoretical framework can be translated into this optical imaging modality. The LSFM simulator showed to be also useful during the validation step of new optical imaging modalities. The package was modified to assess the performance of Statistical Projection Optical Tomography (SPOT), a new tomographic imaging technique in which projections from different views are acquired through the integration of the stack of z planes of a LSFM volume. The simulation tool was used to compare the results from this method against traditional LSFM images, demonstratingisotropic voxel size and a more stable resolution at deep z planes for SPOT. Moreover, the inhomogeneities of the illumination due to the attenuation of the LS where suppressed with this new technique. The last section introduces an experimental proof of concept version of a transmission OPT system in the NIR-II, exploring a polarization approach and a scanned OPT method to overcome the inherent limitations that the increase in the absorption coefficient of water in this window sets to the conventional illumination scheme.