@misc{oai:repo.qst.go.jp:00070447,
 author = {Ikoma, Youko and E., Turkheimer Federico and Edison, P. and Ramlackhansingh, A. and Brooks, D.J. and 生駒 洋子},
 month = {May},
 note = {Objectives: Positron emission tomography with [11C]PIB has been utilized for imaging amyloid plaque deposition in patients with Alzheimer's disease (AD). The specific binding of [11C]PIB is usually quantified by an accumulated radioactivity ratio of target and reference regions or a graphical analysis method using the cerebellum as a reference region where the specific binding is negligible. Recently, however, the retention of [11C]PIB in the cerebellum was reported to increase in familial AD compared with normal volunteers (1), suggesting that this region may not be suitable as a reference region. In this study, we devised a supervised clustering procedure for extracting the reference region automatically in [11C]PIB PET studies, and validated this method in the [11C]PIB studies of control and AD subjects.
Methods: First, the supervised clustering approach of automatic reference extraction for [11C]PIB PET study was established by applying the procedure developed for [11C]PK11195 study (2). A set of kinetic classes that represent the typical kinetic pattern of normalized radioactivity in the gray matter with and without specific binding and in the blood pool was defined from the time-activity curves (TACs) of control and AD subjects. In PET dynamic image, the combination of each kinetic class was calculated by the linear-least squares for each voxel in the gray matter. The TAC of reference region was derived from voxels in which the ratio of normal gray matter class was high. Next, this extraction method was applied to [11C]PIB PET studies of normal volunteers and AD patients. For the validation of proposed method, the TACs of extracted reference region were compared with those of the cerebellum for each subject, and distribution volume ratio (DVR) estimated by the Logan Plot using the input function of extracted reference TAC was compared with DVR using the cerebellum TAC and with DVR calculated by the Logan Plot using an arterial input function.
Results: The TACs of extracted reference region agreed well with those of the cerebellum region in both control and AD patients. There was little difference between DVR of the Logan Plot with the extracted reference region and with the cerebellum, and they were correlated with the DVR estimated with the arterial input function.
Conclusions: Reference tissue extraction with supervised clustering could detect the voxels of reference region without specific binding in [11C]PIB PET study, and it would be a useful tool for the quantification of amyloid plaque deposition, especially for studies in which the retention in cerebellum becomes higher.
References: 
1. Knight WD et al. Carbon-11-Pittsburgh compound B positron emission tomography imaging of amyloid deposition in presenilin 1 mutation carriers. Brain, in press
2. Turkheimer FE et al. Reference and Target Region Modeling of [11C]-(R)-PK11195 Brain Studies. J Nucl Med 48:158-167, 2007, Brain 2011},
 title = {Automatic Extraction of Reference Region using Supervised Clustering for PET Study with [11C]PIB},
 year = {2011}
}