Non-invasive arterial input function estimation using an MRI atlas and machine learning

Abstract

Abstract Background Dynamic positron emission tomography (PET) scanners can generate images of parameters that reflect the kinetics of the administered radiotracer. Kinetic parameter estimation requires an arterial input function (AIF) conventionally obtained from arterial blood samples. The AIF can also be non-invasively estimated from blood pools in PET images, often identified using co-registered MRI images. Deploying methods without blood sampling or the use of MRI generally requires total body PET systems with a long axial field-of-view (LAFOV) that includes a large cardiovascular blood pool. However, the number of such systems in clinical use is currently much smaller than that of short axial field-of-view (SAFOV) scanners. Methods We propose a data-driven approach for AIF estimation for SAFOV PET scanners, which is non-invasive and does not require MRI or blood sampling. Dynamic 18F-fluorodeoxyglucose (18F-FDG) total body PET data were acquired over 62 min in 10 subjects. A probabilistic vascular MRI atlas was registered to each subject’s PET image to identify cerebral arteries in the brain, from which the AIF was estimated. To correct for partial volume effects, a variational inference machine learning approach was implemented. The estimated AIF using brain PET images (AIF-Brain) was compared to that obtained using data from the descending aorta of the heart (AIF-DA). Kinetic rate constants (K1, k2, k3) and net tracer influx (Ki) for both cases were computed and compared. Results Qualitatively, the shape of AIF-Brain matched that of AIF-DA, capturing information on both the peak and tail of the AIF. The area under the curve (AUC) of AIF-Brain and AIF-DA were similar, with an average relative error of 9%. The mean Pearson correlations between kinetic parameters (K1, k2, k3) estimated with AIF-DA and AIF-Brain for each voxel were between 0.92 and 0.99 in all subjects, and for Ki, it was above 0.97. Conclusion This study introduces a new approach for AIF estimation in dynamic PET using brain PET images, a probabilistic vascular atlas, and machine learning techniques. The findings demonstrate the feasibility of non-invasive and subject-specific AIF estimation for SAFOV scanners.