| Summary: | Positron emission tomography (PET) allows the accurate quantitation of blood flow measurement non-invasively. Four mathematical methods of calculating blood flow from PET scan data of H215O bolus injections were implemented. No gold standard method for determining blood flow exists, so the accuracy and precision of estimates must be found using alternative methods. To this end a computer model of the scanner was used to study various effects on flow including the arterial time delay and dispersion, tissue inhomogeneity, incorrect partition coefficient estimation, scan duration and weighting functions. However, computer models do not adequately describe all of the physical problems. Physical models were therefore built and imaged in an analogous manner to patients. They highlighted some of the physical problems with data collection, including deadtime of both the arterial -detector and the PET scanner. Suitable correction procedures were developed and implemented. The physical models were used to determine the optimal scan duration. For the autoradiographic method, the optimal scan duration is 132 seconds and for the weighted methods, 264 seconds. The best weighting functions for the weighted methods were 1 and t . Sophisticated optimisation methods were unable to give consistent optimal weights. The estimates of flow for the physical models were typically within 5&'37 of the true values. When flows were calculated from patient data, the flow estimates fell as the scan duration increased. This was due to a blood volume component where the activity in the blood does not diffuse into the tissue spaces. The conclusion is that the single compartment tissue model enables blood flow to be estimated with an accuracy of about 10&'37, but for higher accuracy, the non-exchanging blood volume component must be considered.
|
|---|
