Examination of assumptions for local cerebral blood flow studies in PET.

Two common assumptions made in most positron emission tomography (PET) cerebral blood flow techniques have been examined in detail. These are (1) that the blood-borne radioactivity component in the measured PET data is negligible, and (2) that differences in arrival time of the arterial bolus across the brain cause insignificant biases in the estimated cerebral blood flow (CBF) values. Biases in CBF values due to partial failure of these assumptions have been predicted by computer simulation studies and also quantitated for both dynamic and single scan PET methods using H2 15O. Both computer simulations and measured PET data indicate that these assumptions can sometimes cause significant errors in the estimated flow values. The magnitude of these errors depends on the PET technique used (dynamic or static) and on the interval of data included in the flow calculations. The bias caused when these assumptions fail can be considerably reduced by omitting approximately 40 sec of data immediately following tracer administration from the CBF calculations.