In the past twenty years, far-reaching studies of molecular and cellular processes have reached a milestone in their maturation, and the knowledge from these studies was ready to apply at higher organizational levels. At that time, rodent models were long established. However, methods were inappropriate to image a whole rodent organ, such as the mouse brain, which drove the emergence of a new range of imaging techniques, later gathered under the name mesoscopy. Mesoscopic techniques filled a gap between classical microscopy and medical imaging techniques, such as magnetic resonance imaging, and X-ray computed tomography. They allow the acquisition of centimeter-sized samples. In this thesis, we focus on one of these mesoscopic imaging techniques called optical projection tomography, or OPT, and its potential application to Alzheimer's disease (AD) research. We review the fundamentals of OPT and describe the filtered back-projection algorithm, which is the primary tomographic reconstruction method of this technique. We also go through the implementation of OPT for whole mouse brain imaging, including sample preparation. We show that OPT is suitable to image the whole brain anatomy based on endogenous fluorescence, and the whole neural vasculature as well as amyloid plaques (a hallmark of AD) with adequate fluorescent markers. Then, we dwell on the characterization of OPT instruments. We give some insights on the instrument point spread function and discuss the influence of the number of projections on the quality of the reconstructed image. Afterward, we illustrate the application of OPT to study amyloidosis progression in a preliminary cross-sectional study, where we have used supervised learning to quantify the amyloid plaque load. In this study, we show that OPT can be used to quantify amyloidosis in whole mouse brains and that comparison between individuals of different age can be performed. Imaging of a whole mouse brain is unquestionably necessary. At this scale though, it has some constraints. We present the limitations of OPT, and we share how we think they can be circumvented by combining this modality with another microscopy technique, namely structured illumination microscopy. We see that this other microscopy technique has the potential to produce high-resolution zooms in selected regions of interest based on a prior OPT acquisition. The results presented in this work have led to the duplication of our OPT instrument in Lund University, and we hope they will help to foster advances in OPT and broaden its range of application. We also hope that this work will contribute to making OPT more accessible and user-friendly.
EPFL2019
