Master of Science (MS)


Physics and Astronomy

Document Type



The PET/CT scanner has been recognized as a powerful diagnostic imaging modality in oncology and radiation treatment planning. Traditionally, PET has been used for quantitative analysis, and diagnostic interpretations of PET images greatly relied on a nuclear medicine physician’s experience and knowledge. The PET data set represents a positron emitter’s activity concentration as a gray scale in each pixel. The assurance of the quantitative accuracy of the PET data is critical for diagnosis and staging of disease and evaluation of treatment. The standard uptake value (SUV) is a widely employed parameter in clinical settings to distinguish malignant lesions from others. SUV is a rough normalization of radioactive tracer uptake where normal tissue uptake is unity. The PET scanner is a sensitive diagnostic method to detect small lesions such as lymph node metastasis less than 1 cm in diameter, whereas the CT scanner may be limited in detecting these lesions. The accuracy of quantitation of small lesions is critical for predicting prognosis or planning a treatment of the patient. PET/CT uses attenuation correction factors obtained from CT scanner data sets. Non-biological materials such as metals and contrast agents are recognized as a factor that leads to a wrong scaling factor in the PET image. We challenge the accuracy of the quantitative method that physicians routinely use as a parameter to distinguish malignant lesions from others under clinical settings in commercially available CT/PET scanners. First, we verified if we could recover constant activity concentration throughout the field of view for small identical activity concentration sources. Second, we tested how much the CT-based attenuation correction factor could be influenced by contrast agents. Third, we tested how much error in quantitation could be introduced by object size. Our data suggest that the routine normalization process of the PET scanner does not guarantee an accurate quantitation of discrete uniform activity sources in the PET/CT scanner. Also, activity concentrations greatly rely on an object’s dimensions and object size. A recovery correction factor is necessary on these quantitative data for oncological evaluation to assure accurate interpretation of the activity concentration. Development of parameters for quantitation other than SUV may overcome SUV’s inherent limitations reflecting patient-specific physiology and the imaging characteristics of individual scanners.



Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Kenneth L. Matthews, II