Positron emission tomography (PET) and single photon emission computed tomography (SPECT) are in vivo tracer kinetic techniques that employ specific, venously injected radiotracers to image molecular interactions and pathways. PET and SPECT imaging processes involve multiple steps, including radionuclide production, radiosynthesis, preclinical imaging, signal and image processing and, ultimately, clinical application. These technologies provide quantitative information about physiological and biological properties, such as cerebral blood flow, glucose metabolism and receptor concentration, that have the potential to answer a large number of important clinical questions. However, the raw data obtained by PET and SPECT do not correspond directly to the biological parameters of interest, but instead represent variations in tracer concentrations and states in tissue. Post-processing is therefore required to accurately transform this raw information into an interpretable form. This is accomplished using tracer kinetic models that simulate the kinetic behavior of the radioactive tracer, expressed in terms of functional compartments represented by mathematical expressions. This thesis explores the mathematical modeling of neurotransmitter systems in the human and rat brain using PET and SPECT, and its application to parametric imaging. The main goal was to propose reliable and simple methods for extracting biological parameter values for a given brain region or for the brain as a whole to build parametric images. We have firstly shown that benzodiazepine receptor concentration and ligand affinity could be obtained using a non-invasive and very simple protocol that is appropriate for routine clinical use. This development highlighted the importance of modeling in PET and SPECT, particularly for brain receptor imaging. For example, we have shown that both 11C-labeled flumazenil ([11C]FMZ) and 123I-labeled iomazenil ([123I]IMZ) could be used to map benzodiazepine receptor distribution, but each required different experimental protocols that depended mainly on the kinetic properties of the injected tracer. A new approach is then proposed to study tracer kinetics using SPECT data. This new method, termed the dual-ligand approach, uses an adaptation of the multi-injection protocol and enables all model binding parameters to be measured with acceptable standard errors. Modeling is then applied to the entire SPECT (or PET) dataset to build parametric images at the pixel level. The proposed modeling method is based on the application of wavelet filtering to dynamic data. This work, which represents the first attempt proposed in the literature to improve the signal-to-noise ratio using a wavelet filtering method, provides a theoretical framework for analyzing data and presents an initial application of PET and SPECT to imaging of biological parameters at the whole brain level. In the next section, imaging techniques are used to evaluate quantitative protocols in the rat. Here, we have used the β-microprobe, which, because of its high temporal and spatial resolution, is an appropriate tool for kinetic analysis, and the YAP-(S)PET system, which is a dedicated small-animal scanner for the non-invasive analysis of radioligand distribution in the whole brain. Using 2-deoxy-2-18F-fluoro-D-glucose ([18F]FDG) and the 18F-labeled 5-HT1A receptor antagonist, [18F]MPPF, to assess glucose consumption and 5-HT1A receptors, respectively, we found that the β-microprobe was suitable for quantitative studies, but some correction factors should be included with YAP-(S)PET data to improve quantitative results.