Abstract
1297
Objectives Cerebral blood flow (CBF) and oxygen extraction fraction (OEF) are important physiological parameters that reflect the functional activity of the brain. PET imaging with 15O tracers is one approach to measure these parameters. While the classic autoradiographic approach has been widely applied, it does not give information on the reliability of the estimated parameters nor does it take advantage of the dynamic capability of modern PET scanner. On the other hand, nonlinear fitting based kinetic modeling uses dynamic PET data, but it is difficult to implement. In this work, we use linearized models to estimate these parameters.
Methods Linearized models were derived analogous to that done in Logan graphical analysis based on compartment models for CBF and OEF. Six healthy subjects each received four repetitions of dynamic 15O-H2O scans and two MRI scans. The MRI included high resolution 3D T1 weighted MPRAGE and 3D time-of-flight (TOF) MR angiography for co-registration and input function extraction. Whole brain CBF was estimated using both the linearized model and the autoradiographic approach. In three other subjects, MRI scans, 15O-H2O, 15O-CO, and 15O-O2 PET scans were performed. Whole brain CBF and OEF was estimated for these three datasets using both the linearized models and the conventional autoradiographic approach.
Results For the six subjects receiving only water scans, the mean CBF averaged over all the subjects and repetitions was 52.1 ml/100g/min using the linearized model, and 53.8 ml/100g/min using the conventional method. For the three subjects with all three 15O PET scans, the mean CBF was 62.4 ml/100g/min for the linearized models and 64.6 ml/100/g/min for the conventional method. The mean OEF was 0.50 for the linearized models and 0.42 using the conventional method.
Conclusions The proposed linearized models generate comparable CBF and OEF estimations as the autoradiographic approach. Further study of the linearized models is undergoing to evaluate its performance.
Research Support P50NS006833-39; P30NS048056-0