• Monte carlo simulation
• Cerenkov luminescence imaging
• quantitative imaging
• CCD

Introduction and motivations
Functional imaging is a method used by physicians to detect a certain type of physiological activity related to the physical distribution of the radioisotope used. The imaging method should allow to image and to localize the activity distribution (imaging), but it should also allow to quantify the same distribution (quantification). In functional imaging techniques such as positron emission tomography (PET) these requirements are fulfilled by combining PET with computed tomography (CT) and by calibrating the response of the PET imaging system with a reference standard.
Cerenkov luminescence imaging (CLI) is a novel functional imaging technique that has been introduced to study beta-emitting radiotracers and radiopharmaceuticals through the Cerenkov radiation they produce in biological tissue. Up to date, several studies have demonstrated the imaging capabilities of CLI in different applications field, but a complete quantitative reconstruction of the activity distribution has not been demonstrated yet and very few bibliographical references are available on the topic. In order to achieve this goal, one should be able to calibrate the detector response in terms of radiance at the tissue surface (that is the measured quantity) and to relate this quantity to the activity distribution deep in tissue through the distribution of the produced Cerenkov radiation. This reconstruction is difficult for several reasons: the lack of a tissue-equivalent material able to mimic the tissue interactions with optical light, the heterogeneity of the optical properties of different tissues and the limited amount of available data, and the dependence of the Cerenkov radiation production on factors like the properties and geometry of the material and the isotope used. A Monte Carlo simulation can be a powerful tool to shed some light on such a complex scenario, which would not only be useful to help in the reconstruction, but would also be a valid means to use in experiment planning and in studying potential applications for CLI.
Work done
The main goals of this thesis were the development of a Monte Carlo code to model all the physical processes involved in a CLI experiment (the radioactive decay, the Cerenkov radiation production and transport as well as the detector response) and the test of the predictive capabilities of the code.
To this aim, the simulation output was compared with experimental results obtained with different systems. The imaging system representing the gold standard for CLI (the IVIS imaging system) was used first for measurements in tissue to compare the simulated and measured light distribution at the tissue surface, and a custom imaging system featuring a charge coupled device was built to compare quantitatively the simulation output and the measured signal. In addition, a feasibility study with digital silicon photomultipliers was performed to evaluate the capabilities of this technology as an alternative to charge coupled devices.
Results and Conclusions
The comparison of the simulation results with the acquisitions performed in tissue with the IVIS imaging system demonstrated that the Monte Carlo is able to predict the spatial distribution of the Cerenkov light at the tissue surface within a 25% precision. Very modest results were obtained in terms of the transmitted spectrum, with differences as high as 40% between the predicted and measured spectral components, because this aspect depends on the spectral response of the detector which could not be fully modeled.
So far, the custom CCD-based imaging system was the only imaging system that allowed an absolute quantitative comparison of the experimental and simulated results, and provided an acceptable and constant discrepancy between measurements and simulations in water. In fact, the Monte Carlo code was able to predict the electron count rate produced in the CCD in a known geometry with different isotopes and different activity levels, but the simulated results typically underestimated the real results by a factor 0.8. However, this factor was a constant offset among different measurements, thus it is reasonable to assume that it is due to a fixed contribution that was not considered in the simulation and that could presumably be corrected. The experimental results with this system also suggest that for quantitative CLI measurements a conventional CCD is preferable over an EMCCD because the measured signal is less influenced by noise factors that depend on the acquisition settings.
The experimental results obtained with the digital silicon photomultipliers suggest that this technology might be an appealing alternative to CCDs for quantitative CLI, because the detected signal showed a dynamic response to the source activity and depth in tissue, but more work is needed to understand and to quantify the experimental factors (e.g. afterpulse) to correct when recovering quantitatively the measured signal.