The studies of improved quantification techniques for positron emission tomography

Abstract

Positron emission tomography (PET) technology provides a sensitive, informative and quantitative approach for non-invasively visualization of biological processes with specific molecular probes. Although PET has been widely used in clinical and preclinical studies, there are still some challenges for quantification techniques to provide more reliable information about the underlying biological processes from PET imaging. This thesis is thus devoted to advance appropriate quantification techniques in order to improve the reliability of quantification and to facilitate image analysis, emphasizing on the PET imaging with [¹⁸F]Fluoro-2-deoxy-2-D-glucose (FDG-PET) in small-animal studies and dual-time scans for human. The investigations of this thesis began with the improvements in the reliability of quantification for PET imaging. An automatic image-based method with improved standard uptake values (SUVs) was developed to reduce the influences of the error measurements and the excreted tracer by renal function. The improved SUVs, an image-derived SUV (iSUV) and a modified SUV (mSUV), were proposed and then demonstrated in a set of static mouse FDG-PET studies. For the selected targets, the iSUV and mSUV were compared against their corresponding SUVs. The results indicated that the automatic image-based method could derive the iSUV to replace SUV when the actual measurements were missing or unreliable. Meanwhile, the estimated mSUV can reduce the inter-subject variability and enhance the tumor-to-background separation. In addition to the improvements in the calculation of SUV, a simulation framework was designed to derive generalized optimal quantitative index (QI) and its associated flexible dual-time imaging protocol for accurately differentiating the malignancy from benign lesions in FDG-PET for lung cancer diagnosis. In computer simulations, the generalized optimal QI can provide the higher accuracy in differentiating malignancy from benign lesions under each noise level than traditional RI for lung cancer diagnosis. The corresponding optimal dual-time imaging protocol was suggested to be flexible, which could be valuable for the high-throughput clinical applications. Furthermore, the frequent invasive arterial blood sampling to obtain input function poses many challenges and inconvenience for tracer kinetic modeling with dynamic PET imaging. Non-invasive approaches were proposed to address this issue. A simple non-invasive quantification method based on Patlak graphic analysis (PGA) that uses a reference region to obtain the relative influx rate without an input function has recently been proposed. This non-invasive Patlak (nPGA) method was extended to whole-body FDG-PET studies and systematically investigated using a set of mouse studies and computer simulations. In the mouse studies, a high linearity of relative influx rates was observed between the nPGA and PGA for most pairs of reference and target regions, when an appropriate underlying kinetic model was used. The simulation results demonstrated that the accuracy of the nPGA method was similar to that of the PGA method, with a higher reliability for most pairs of reference and target regions. The results proved that the nPGA method could provide a non-invasive and indirect way of quantifying the FDG kinetics in the tissues where the k₄ and vascular effect were negligible. The automatic non-invasive approach can further facilitate the analysis of dynamic PET imaging. A novel hybrid clustering method (HCM) was proposed to objectively delineate the regions of interest (ROIs) in dynamic PET images for the automatic estimation of image-derived input function (IDIF) and was demonstrated the application to the mouse FDG-PET studies. The results showed that the HCM achieved accurate ROIs delineation in both computer simulations and experimental mouse studies. In the mouse studies the predicted IDIF had a high correlation with the gold standard, the PTAC derived from invasive blood sampling. The results indicate that the proposed HCM has a great potential in ROI delineation for automatic estimation of IDIF in dynamic FDG-PET studies. In summary, this thesis has significantly contributed to the improvement of several critical quantification approaches for preclinical small-animal studies and clinical patient examinations.