Work

A non compartmental method for functional quantitative imaging with Positron Emission Tomography and irreversible tracers

Year: 2013
Type: dissertation
Abstract: In dynamic Positron Emission Tomography (PET) studies the term "Spectral Analysis" indicates a time-invariant single input/single output model, used for the data quantification [Cunningham and Jones, 1993]. Despite the name and its common use in the engineering field, SA does not indicate an analysis in the frequency domain but, instead, it represents a method from which the radioactivity concentration measured with PET can be related to the underlying physiological processes of the investigated system. SA is so-called, because it provides a “spectrum” of the kinetic components from which it is possible to derive a large variety of physiological parameters, depending on the characteristics of the analyzed tracers. In the last years SA has been widely used with a large number of PET tracers to study brain and non brain tissues, demonstrating to be a very flexible method. Differently from the most used PET quantification approaches, like the compartmental modelling [Godfrey, 1982] or the graphical methods [Patlak, 1983; Logan et al., 1990], SA can be applied to homogeneous as well as to heterogeneous kinetic tissues without any specific compartmental model assumptions. This characteristic makes it a high informative investigative tool especially for the analysis of novel PET tracers. The most critical aspect of SA is related to its sensitivity to the presence of noise in the data. This characteristic makes SA not properly indicated for the application to low signal-to-noise ratio (SNR) data [Turkheimer et al., 1994]. During the past several years, several solutions have been introduced to improve the robustness of SA in the presence of noise. The most famous example is represented by rank-shaping spectral analysis (RS) [Turkheimer et al., 2003]. However, even if RS has been shown to be a precise and accurate quantification method, its applicability is limited to tracers with reversible uptake. This is a severe restriction if we consider that one of the most used PET tracer for clinical research, 18F-Fluorodeoxyglucose ([18F]FDG), is irreversible. In this work we present SAIF, (Spectral Analysis with Iterative Filter), a SA-based method for the quantification of PET data investigated with irreversible-uptake tracers. SAIF has been designed in order to maintain the main advantages of SA but providing a superior robustness to measurement noise. The final aim was to create a reliable and flexible PET quantification tool, offering a valid alternative to standard methodologies for functional quantitative imaging with PET and irreversible tracers. The organization of this thesis is as follows: Chapter 1 offers a brief introduction to PET technique and its quantification methods. A comparison between compartmental modelling approaches and graphical methods is also presented, in order to provide the operative context in which SA is located. Chapter 2 contains the mathematical formalization of the SA model. Standard and filtered SA versions are presented with particular attention to novelty elements introduced by SAIF. In Chapter 3 and Chapter 4, SAIF will be tested with brain and non brain PET data. Several datasets obtained by using different PET tracers are considered. As an example for brain tissue quantification, SAIF application to L-[1-11C]Leucine and [11C]SCH442416 data is presented. For non brain tissues, instead, analysis of three datasets is reported: 1) [18F]FDG PET studies applied to skeletal leg muscle, 2) [18F]FLT PET studies applied to breast cancer patients and 3) [18F]FDG PET studies applied to normal control and acute lung injury patients. For each dataset SAIF results are compared with those provided by already validated methods and used in the literature as reference for the quantification. This analysis allows to compare SAIF performances with those offered by the current state of the art. Chapter 5 investigates the conditioning of the kinetic heterogeneity to PET quantification. The relationship between this problem, the spatial resolution of the imaging technique and the noise level of the data is also considered. This aspect is a critical point for PET quantification because when it is not taken into account it can lead to heavily biased results. Particular attention is given to how SAIF addresses this issue. In Chapter 6 we present SAKE, a software application in-house developed which implements the major SA algorithms. SAKE manages the whole process of PET quantification: from data pre–processing to the result analysis. No other program or additional tool is required. Chapter 7 discusses the most relevant criticalities of the SA approach and of SAIF method in particular. Considerable attention is given to the definition of the setting algorithm as well as to the model assumptions used by SAIF to describe the data. In Chapter 8 an overall discussion is presented with a conclusive summary about strengths and weakness of SAIF method. The appendix of the thesis is dedicated to the some additional works, not directly related to the main argument of this PhD project, but of interest for the PET field. This research concerns 1) the development of voxelwise quantification methods for [11C](R)Rolipram PET data, 2)the use of non linear mixed effects modelling for plasma metabolite correction, and 3) the evaluation of the sensitivity of PET receptor occupancy studies to the experimental design.
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W2289013388
Cites: 148
Cited by:
Related to: 10