Skip to main content
NCBI home page
Journal of Cerebral Blood Flow & Metabolism logoLink to Journal of Cerebral Blood Flow & Metabolism
. 2010 May 5;31(1):155–165. doi: 10.1038/jcbfm.2010.64

Reproducibility of striatal and thalamic dopamine D2 receptor binding using [11C]raclopride with high-resolution positron emission tomography

Kati Alakurtti 1,*, Sargo Aalto 1, Jarkko J Johansson 1, Kjell Någren 1, Terhi Tuokkola 1, Vesa Oikonen 1, Matti Laine 2, Juha O Rinne 1
PMCID: PMC3049480  PMID: 20442726

Abstract

Positron emission tomography (PET) imaging of small striatal brain structures such as the ventral striatum (VST) has been hampered by low spatial resolution causing partial-volume effects. The high-resolution research tomograph (HRRT) is a brain-dedicated PET scanner that has considerably better spatial resolution than its predecessors. However, its superior spatial resolution is associated with a lower signal-to-noise ratio. We evaluated the test–retest reliability of the striatal and thalamic dopamine D2 receptor binding using the HRRT scanner. Seven healthy male volunteers underwent two [11C]raclopride PET scans with a 2.5-hour interval. Dopamine D2 receptor availability was quantified as binding potential (BPND) using the simplified reference tissue model. To evaluate the reproducibility of repeated BPND estimations, absolute variability (VAR) and intraclass correlation coefficients (ICCs) were calculated. VAR values indicated fairly good reproducibility and were 3.6% to 4.5% for the caudate nucleus and putamen and 4.5% to 6.4% for the lateral and medial part of the thalamus. In the VST, the VAR value was 5.8% when the definition was made in the coronal plane. However, the ICC values were only moderate, in the range of 0.34 to 0.66, for all regions except the putamen (0.87). Experimental signal processing methods improved neither ICC nor VAR values significantly.

Keywords: dopamine receptors, PET, raclopride, reproducibility, test–retest, ventral striatum

Introduction

In quantification of striatal D2 receptor binding in vivo in humans using positron emission tomography (PET), [11C]raclopride is a well-validated tracer. Previous studies with different test–retest time intervals and quantification methods indicate that the region of interest (ROI) measurement of striatal [11C]raclopride binding is highly reliable (Volkow et al, 1993; Mawlawi et al, 2001; Hirvonen et al, 2003). However, PET measurements have bias because of the partial-volume effect (PVE) caused by the low spatial resolution, especially when imaging the small, D2 receptor-rich striatal brain structures such as the caudate nucleus and the ventral striatum (VST) (Sureau et al, 2008). Brain imaging has recently benefited from the development of scanners such as the high-resolution research tomograph (HRRT), which has a spatial resolution in the range of about 2.5 to ∼3 mm in the 10-cm field of view that covers most of the brain (de Jong et al, 2007). The ability to reliably measure dopamine D2 receptor binding in striatal subregions is important, because these regions are differentially involved in cognitive and motor performance both in healthy volunteers and in patients with Parkinson's disease (DeLong, 1983; Evarts and Wise, 1984; Volkow et al, 1993; Bäckman et al, 2000; Kaasinen and Rinne, 2002). In addition, the VST has recently attracted attention because of its role in human behavior and its disorders such as addiction (Floresco et al, 2008; Belin and Everitt, 2008). However, applying a high-resolution scanner is associated with a lower signal-to-noise ratio, which may impair the reliability of the measurement (Sureau et al, 2008; Johansson et al, 2007). Evaluation of the test–retest variability of PET measurement is particularly important in studies in which the same individuals are scanned repeatedly before and after intervention or when small brain structures, such as the VST, are analyzed. Knowledge about the test–retest variability also allows to estimate the number of subjects needed to detect a change in tracer binding, for instance after experimental intervention.

In this study, we evaluated the test–retest reliability of the striatal and thalamic D2 binding by administering a single bolus of [11C]raclopride twice on the same day in seven healthy males and collecting data using the HRRT scanner. A novel approach in data sampling was implemented to account for potential bias in statistical image reconstruction of the HRRT data, and a recently introduced resolution-modeling image reconstruction algorithm (Comtat et al, 2008) was applied for evaluation purposes.

Materials and methods

The study protocol was approved by the Ethics Committee of the Hospital District of Southwestern Finland. The study subjects were given written information about all the relevant issues involved in this study. A written consent was obtained from each subject. This study was performed according to the ethical guidelines given by the Declaration of Helsinki.

Subjects

Seven healthy male subjects free of any somatic or psychiatric illness volunteered to take part in the study. All were right handed and nonsmokers. To exclude any structural brain abnormalities and to obtain anatomic reference, all subjects underwent 1.5 T magnetic resonance imaging. The age, height, and weight of the subjects were 24.5±3.5 years, 185.5±12.5 cm, and 74±14 kg, respectively (mean±s.d.). Each subject underwent two [11C]raclopride PET scans during the same day between 10.00 a.m. and 6.00 p.m., at least 2.5 hours apart.

Preparation of [11C]raclopride

The precursor O-desmethylraclopride free base and raclopride were obtained from ABX advanced biochemical compounds (Radeberg, Germany). [11C]methane was produced at the Accelerator Laboratory of Åbo Akademi with a 103-cm isochronous Efremov cyclotron using the 14N(p,α)11C reaction. [11C]methyl iodide with high specific radioactivity was prepared from [11C]methane (Larsen et al, 1997; Någren et al, 2003). [11C]raclopride was prepared from [11C]methyl triflate according to a published procedure (Langer et al, 1999). The radiochemical purity and the specific radioactivity of the product were determined using high-performance liquid chromatography and ultraviolet detection at 214 nm. The volume of the final product solution was calculated by weighing the sterile product vessel before and after sterile filtration and by dividing by the density of the sterile solvent.

Positron Emission Tomography Imaging

Positron emission tomography experiments were performed using a brain-dedicated high-resolution PET scanner, the ECAT HRRT (Siemens Medical Solutions, Knoxville, TN, USA). The HRRT is a dual-layer, crystal-detector scanner allowing a depth-of-interaction measurement for the coincident photons. It is characterized by an isotropic 2.5-mm intrinsic spatial resolution. Spatial resolution in the reconstructed images varies in the radial and tangential directions from about 2.5 to ∼3 mm and in the axial directions from 2.5 to 3.5 mm in the 10-cm field of view covering most of the brain (de Jong et al, 2007).

The left antecubital vein was cannulated and a single [11C]raclopride bolus injection was administered intravenously and flushed with saline. The injected dose, specific radioactivity, and the injected mass for the test and retest scans were 413±91 MBq, 222±154 MBq/nmol, and 2.47±4.68 μg versus 393±99 MBq, 257±142 MBq/nmol, and 0.68±0.35 μg, respectively, (mean±s.d.), see Table 1. There were no statistically significant differences in the injected doses (paired t-test, P=0.56), in the specific radioactivities (paired t-test, P=0.15), or in the injected masses (paired t-test, P=0.33) between the test and the retest scans. However, one outlier was detected in the boxplot analysis of the injected masses. In the outlier case, the injected mass was 13.03 μg, which is considerably higher than the average of others. Although the injected mass was higher than the others, it was considered tolerable with regard to pharmacological effects and thus this subject was not excluded from the analysis. Radiochemical purity was over 99.5% in all cases. Before each emission scan, a transmission scan was performed using 137Cs point source. Emission scan was started at injection time and data were collected for 55 minutes in list mode format.

Table 1. [11C]raclopride injections and their test–retest characteristics.

[11C]raclopride injection Scan 1
 Scan 2
 Both scans
 VAR (%) (Scan 2−Scan 1)/ Scan 1
 t-Test sign. ICC
  M±s.d. COV (%) M±s.d. COV (%) M±s.d. M±s.d. Range    
Injected dose (MBq) 413±91 22.03 393±99 25.26 403±92 19.87±11.62 −0.32–0.25 0.56 0.60
Specific radioactivity (MBq/nmol) 222±154 69.40 257±142 55.32 240±144 42.89±53.98 −0.24–1.56 0.15 0.91
Injected mass (μg) 2.47±4.68a 189.60 0.68±0.35 51.86 1.57±3.32 43.41±56.44 −1.65–0.24 0.33 0.11
Open in a new tab

COV, coefficient of variation; ICC, intraclass correlation coefficient; M±s.d., mean±standard deviation; VAR, absolute variability.

a

One outlier (case 7, injected mass 13.03 μg).

An individually shaped thermoplastic mask was used with each subject to minimize the head movement. To monitor the head movement, an external position detector (Polaris Vicra, Northern Digital, Waterloo, Ontario, Canada) was used. The position information was converted into motion information by subtracting the average reference position (transmission scan position) coordinates from the position data. A representative motion parameter was obtained by taking the norm of the translation vector. The external motion data were used to estimate three features of motion, which hamper the quality of PET imaging. First, translocation from the transmission scan (TX) position may cause erroneous attenuation correction. To measure distance from the TX position, the mean of the norm vector was calculated within each frame. Second, within-frame motion, which may cause signal blurring, was measured as the amplitude of the norm vector within each frame. Finally, frame-to-frame motion, which may cause sudden changes in the time–activity curve, was measured as the mean of the norm from the previous frame position (transmission scan position for the first frame). Distance from the TX position, the within-frame motion, and the frame-to-frame motion for the test and retest scans were 1.11±0.58, 0.46±0.47, 0.21±0.25 mm versus 1.09±0.56, 0.47±0.46, and 0.21±0.19 mm, respectively (mean±s.d.), and there were no systematic differences in motion between the test and retest scans. Thus, the amount of motion was negligible and in average tolerable compared with scanner resolution. Frame-to-frame coregistration was deemed unnecessary based on external motion recordings.

Quantification of [11C]raclopride Binding

All images were corrected for attenuation, scattering, and random events, scanner dead time, and detector normalization, and finally calibrated to kBq/mL within decay correction. Prompt and random events, attenuation, and scatter correction and normalization factors were processed in 3D sinogram mode, with an axial compression of span 9 and a maximum ring difference of 67, resulting in 2,209 sinograms in 16 segments. Tissue attenuation maps were reconstructed using the maximum a posteriori for transmission data (MAP-TR) algorithm with the standard human brain priors for air, bone, noise, water, or soft tissue (Nuyts et al, 1999). The 3D attenuation correction sinograms were obtained by forward projecting the tissue attenuation maps. Scattered events were estimated using the single scatter simulation algorithm (Watson, 2000), and randoms were estimated from the block singles using a variance reduction algorithm (Byars et al, 2005).

The [11C]raclopride uptake was measured continuously for 55 minutes after the injection. Histogramming was made in the following sequence of frames: 2 × 30, 9 × 60, 3 × 120, 3 × 180, and 6 × 300 seconds, resulting in altogether 23 frames. Image estimates with isotropic voxel dimensions of 1.22 mm × 1.22 mm × 1.22 mm were generated using a speed-optimized version of ordinary poisson ordered subsets expectation maximization (OP-OSEM-3D) reconstruction with 16 subsets and 8 iterations, if not otherwise mentioned (Hong et al, 2007).

The [11C]raclopride images were processed with SPM2 in Matlab 7.0.4 as follows. First, to correct the head positioning between two PET scans, PET-to-PET realigment was made between the sum images. Second, for anatomic reference, T1-weighted magnetic resonance images were coregistered to a mean image of the realigned PET images. Regions of interest were drawn using Imadeus software, version 1.2 (Forima, Turku, Finland) on coregistered magnetic resonance images. Thus, an identical set of ROIs was applied for both [11C]raclopride scans of the same subject. The regions of interest were defined in a transaxial plane on the caudate nucleus, the putamen, the VST, the cerebellum, and the thalamus. We divided the thalamus into a medial and a lateral part because there is a gradient in D2 receptor distribution between these sections (Kessler et al, 1993). In addition, VST was drawn in the coronal plane according to Mawlawi et al (2001) (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Activity distribution of [11C]raclopride in coronal view (A) and in axial plane (B). A mean image of positron emission tomography (PET)-to-PET coregistered summed PET images is overlaid on a T1-weighted magnetic resonance image of one subject at the level of the striatum. Regions of interest are defined on the caudatus, putamen, thalamus, and the ventral striatum.

The mean activity concentration values in the ROIs were calculated and used to generate the regional time–activity curves. No partial-volume correction was made to the mean ROI values. These time–activity curves were used to calculate the regional [11C]raclopride binding potential (BPND) using the simplified reference tissue model with the cerebellum as reference tissue (Lammertsma and Hume, 1996). In this model, it is assumed that free plus nonspecific and specific binding equilibrium are reached so rapidly that they can be treated as a single compartment. Binding potential is defined according to the following equation: BPND=fND BAvail(1/KD), where BAvail is the total receptor concentration in tissue, 1/KD is the apparent affinity, and fND is the fraction of free radiotracer in the nondisplaceable tissue compartment (Innis et al, 2007). In the model fitting, the data points were weighted with the estimate of the inverse variance for each sample.

Statistical Analyses

Test–retest variability was evaluated as follows:

graphic file with name jcbfm201064e1.jpg

where T is the first (test) estimate and RT is the second (retest) estimate. We define VAR(%)=100% VAR.

Test–retest consistency was measured with an intraclass correlation coefficient (ICC). Intraclass correlation coefficient can have a value between −1 and 1. Values closer to 1 indicate that most of the variance is due to between-subject variation rather than within-subject variation, which indicates good reliability, whereas values under 0 indicate greater within-subject than between-subject variation, indicating poor consistency. Values are expressed as mean±s.d.

Paired two-tailed t-test was applied to the injected doses, specific radioactivities, injected masses, and BPND estimates between the test and retest scans and additionally for the BPND between the hemispheres. Moreover, the paired t-test was applied to the areas under the curve (AUC) (see below), BPND estimates, and VAR values between the reconstructions, so that the quantification method described above was always the reference.

Image Reconstruction With Fixed Counting Statistics

Conventionally, in the histogram mode image reconstruction, discrete time points are fixed in the sinogram space while counting statistics are let to vary. This procedure originated from old generation scanners that were not able to record all events as a stream in the acquisition disk. Modern scanners such as the HRRT record all events in a so-called list mode format, which allows us to determine framing after scanning, based on the number of acquired events for instance.

In PET, it is convenient to treat the acquired events as so-called noise-equivalent counts (NECs).

graphic file with name jcbfm201064e2.jpg

where T denotes the number of trues, S the number of scattered, and R the number of random events. In conventional framing with fixed time points, the NEC per frame was 4.15±1.72 × 106 (mean±s.d.) in the range of ∼106 to ∼107 counts, except for the first frame. A new set of sinograms with more homogeneous noise characteristics was created by fixing the minimum counts at 7 × 106 NEC per frame. NEC-based framing yielded 10 to 17 frames per scan, depending on the injected dose and the subject weight. Furthermore, parts of each scan were rebinned into frames of extremely low count statistics, yielding 5 × 105 to 2 × 106 NECs in 268 frames.

The rebinned sinogram data were reconstructed using the method described above. The head positioning between repeated PET scans was corrected using the coregistration parameters estimated from the original data, and thus the same ROIs were applied to yield new time–activity curves. Areas under the curve were calculated from both the original and the new sets of time–activity data and relative changes from original data were calculated as follows:

graphic file with name jcbfm201064e3.jpg

where f0 is the time–activity course estimated with the conventional framing. Integrals of the discrete time–activity course data were estimated using an in-house software with linear interpolation. We define ΔAUC(%)=100% ΔAUC. For the second full set of image data the BPs were reestimated as descibed above.

Image Reconstruction With Resolution Modeling

Additional image data were generated using a resolution model within the statistical reconstruction (Sureau et al, 2008). Resolution modeling exploits a measured point-spread function (PSF) of the scanner, which is assumed to be space invariant within the field of view. Implementation of the algorithm into OP-OSEM-3D has been described earlier (Comtat et al, 2008). The same PSF parameters, iteration number (10), and number of subsets (16) were used as in an earlier paper (Comtat et al, 2008). The same PET-to-PET coregistration parameters and ROIs as with the original data were applied, and the resulting time–activity course data were evaluated using the ΔAUC defined above. The BPs were reestimated using the above described method.

Results

The mean BPND values of the test and retest scans ranged from 0.65±0.04 to 4.41±0.47 with the conventional framing using the OP-OSEM-3D reconstruction (Table 2). The BPND was highest in the putamen, followed by the caudate nucleus and the VST. The BP values were lowest in the thalamus. Thalamic BPND values were nearly identical for the medial and lateral sections (BPND values 0.65±0.04 and 0.67±0.04, respectively). No noticeable lateralization of the [11C]raclopride binding was observed (t-test, P=0.18 to 0.98 for test, and P=0.08 to 0.94 for retest). Moreover, the VAR values were nearly identical in both the left and the right hemisphere (t-test, P=0.30 to 0.93). There was an increase in the mean of BPND values from the test to the retest scans in all regions except the VST. However, the increase was statistically significant only in the lateral (t-test, P=0.02) and medial thalamus (t-test, P=0.01). The mean left and right hemisphere values were used for the rest of the analyses.

Table 2. Regional [11C]raclopride BPND values and their test–retest characteristics.

Region Scan 1 (BPND)
 Scan 2 (BPND)
 Both scans (BPND) VAR (%) 100 × (Scan 2− Scan 1)/Scan 1
 t-Test sign. ICC
  M±s.d. COV (%) M±s.d. COV (%) M±s.d. M±s.d. Range    
Putamen 4.33±0.45 10.32 4.49±0.51 11.36 4.41±0.47 4.70±3.91 −2.76–12.08 0.104 0.87
Caudate nucleus 3.81±0.22 5.83 3.88±0.23 6.05 3.85±0.22 3.71±4.72 −4.24–12.37 0.402 0.59
Ventral striatum (cor) 3.40±0.28 8.34 3.44±0.26 7.51 3.42±0.26 5.93±5.34 −8.84–15.37 0.731 0.50
Ventral striatum (ax) 3.27±0.43 13.13 3.17±0.36 11.41 3.22±0.38 9.72±9.67 −20.06–24.96 0.599 0.34
Medial thalamus 0.65±0.03 5.03 0.68±0.05 7.12 0.67±0.04 4.59±3.22 0.49–9.85 0.010 0.66
Lateral thalamus 0.63±0.05 7.27 0.67±0.04 5.67 0.65±0.04 6.69±4.82 −1.77–11.51 0.024 0.44
Open in a new tab

BPND, binding potential according to the simplified reference tissue model; COV, coefficient of variation; ICC, intraclass correlation coefficient; M±s.d., mean±standard deviation; VAR, absolute variability.

Variability and Consistency of the Test and Retest Scans

The test–retest VAR and ICC values for the combined ROIs are presented in Table 2. Mean VAR (%) values were 3.7% to 4.7% for the caudate nucleus and the putamen, and 4.6% to 6.7% for the lateral and medial part of the thalamus, indicating fairly good reproducibility in these regions. The BPND values measured from the putamen yielded high ICC values both for the separate hemispheres and for the combined hemispheres (0.87 for the combined putamen). In the caudate nucleus, the hemispheric ICC values varied considerably and the ICC value for the combined ROI was only moderate (0.59). Thalamic BPND values resulted in moderate ICC values of 0.44 for the lateral part and 0.66 for the medial part. For the VST, the ROIs defined in the coronal plane yielded higher BPND values than the ROIs drawn in the axial plane (3.42±0.23 versus 3.22±0.32, respectively). In addition, the VAR (5.9%) and the ICC (0.50) values were superior for the coronally defined VST ROIs as compared with those drawn on the axial plane (9.7% and 0.34, respectively).

Image Reconstruction With Fixed Counting Statistics

Figure 2 shows a typical set of time–activity course data obtained with the different framings and the relative changes in the AUC (ΔAUC(%)) are presented in Table 3.

Figure 2.

Figure 2

Open in a new tab

A typical example of regional time–activity course data in the putamen and the cerebellum estimated by the different framings and the point-spread function (PSF) reconstruction. Solid line is for original framing with non-PSF reconstruction, dashdot line is for high statistics framing, dashed line is for PSF reconstruction, and circles are for low statistics framing.

Table 3. Regional ΔAUC(%) values in the combined ROIs, for high statistics, low statistics initial part, low statistics end part, and PSF reconstruction.

Region Scan 1 ΔAUC(%), M±s.d.
 Scan 2 ΔAUC(%), M±s.d.
  High Low start Low end PSF High Low start Low end PSF
Putamen 1.36±1.04† −2.52±2.21 −6.21±3.56‡ 7.06±1.02‡ 1.48±1.08‡ −2.80±2.48 −6.43±4.18‡ 7.50±0.92‡
Caudate nucleus 1.39±1.14† −3.85±2.91‡ −6.66±4.31‡ 8.29±1.17‡ 1.65±1.29‡ −3.13±2.69 −5.90±5.22‡ 8.66±0.86‡
Ventral striatum (cor) 0.88±1.03 −1.43±1.31‡ −5.04±3.20‡ 5.58±0.53‡ 1.12±0.97† −0.08±1.35 −6.04±3.93‡ 6.16±0.82‡
Ventral striatum (ax) 0.89±0.99 −1.66±1.91 −6.42±4.54‡ 6.45±0.94‡ 1.05±1.08 −0.51±1.23 −6.36±4.62‡ 6.94±0.98‡
Medial thalamus −0.05±1.27 0.19±1.26 3.27±3.77 1.53±2.11 0.00±1.50 −0.52±1.42 4.65±4.14‡ 2.19±0.39‡
Lateral thalamus −0.74±1.31 1.99±1.51† 5.70±3.07‡ −1.41±2.25 −0.93±0.92 2.24±1.47‡ 4.88±3.13‡ −0.77±0.86†
Cerebellum 0.16±0.53 −1.65±0.74‡ 0.09±2.91 0.78±1.70 0.17±0.59 −1.53±1.35† 1.63±1.95† 1.04±0.74‡
Open in a new tab

AUC, area under the curve; M±s.d., mean±standard deviation; PSF, point-spread function; ROI, region of interest.

†Statistically significant difference as compared with original framing with P<0.05 in Student's t-test.

‡Statistically significant difference as compared with original framing with P<0.01 in Student's t-test.

Regional ΔAUC(%) values have nonzero means for both the high and low statistics. High counting statistics yielded a positive mean in both the scans in the caudate nucleus (1.39±1.14 and 1.65±1.29), in the putamen (1.36±1.04 and 1.48±1.08), in the VST (transaxial ROI, 0.89±0.99 and 1.05±1.08), and in the cerebellum (0.16±0.53 and 0.17±0.59). However, negative (or zero) means were observed in both the scans in the medial (−0.05±1.27 and 0.00±1.50) and in the lateral parts of the thalamus (−0.74±1.31 and −0.93±0.92). Thus, high counting statistics yielded larger AUC estimates as compared with the original in the regions with high radioactivity concentration (RC) and smaller AUC estimates in the regions with low RC. For the low statistics, the behavior of the mean ΔAUC(%) was inversed: the caudate nucleus and the putamen had negative means in both the scans in the beginning and at the end of the scan, whereas the thalamus had positive means in all cases except for the start of the second scan. In the cerebellum, the mean ΔAUC(%) was −1.65±0.74 and −1.53±1.35 in the beginning of the scan but 0.72±3.02 and 1.49±1.94 at the end of the scan. However small, the direction of the change in the AUC is clear—high counting statistics yielded larger estimates in the high RC regions and smaller estimates in the low RC regions as compared with low counting statistics. Although statistically significant in the paired t-test, the amount of change was minimal between the original and the high statistics framing, and between the first part of the low statistics framing and the original framing. As regards the end part with low statistics, the bias (ΔAUC(%)) was noticeably larger in both the scans, with mean values of −6.66±4.31 and −5.90±5.22 for the caudate nucleus, and 4.65±4.14 and 3.96±3.65 for the medial part of the thalamus.

BPND estimates and the VAR and ICC values for combined ROIs with the high statistics framing are presented in Table 4. Changes in the mean BPND were positive in the caudate nucleus (2.86%), in the putamen (2.95%), in the axially drawn VST (2.17%), and in the coronally drawn VST (2.05%), and negative or zero in the lateral (−1.59%) and medial parts of thalamus (0.00%) as compared with the conventional framing. In the VAR and ICC, the changes were more ambiguous, as high counting statistics yielded higher VAR in the caudate nucleus, the putamen, and in the lateral part of thalamus, but lower VAR in the medial part of thalamus and in the VST. None of the regions had significantly different mean VAR as compared with the original framing. The high counting statistics improved the ICC in the thalamus and in the coronally defined VST, whereas in the other regions, the change was negative or zero.

Table 4. Regional [11C]raclopride BPND values with high counting statistics reconstructions and their test–retest characteristics.

Region Scan 1 (BPND)
 Scan 2 (BPND)
 Both scans (BPND) VAR (%) 100 × (Scan 2− Scan1)/Scan 1
 t-Test sign. ICC
  M±s.d. COV (%) M±s.d. COV (%) M±s.d. M±s.d. Range    
Putamen 4.46±0.49‡ 10.89 4.62±0.54‡ 11.72 4.54±0.50 5.03±3.84 −2.93–12.27 0.138 0.87
Caudate nucleus 3.92±0.24‡ 6.16 3.99±0.23‡ 5.88 3.96±0.23 4.00±4.70 −3.95–12.37 0.430 0.57
Ventral striatum (cor) 3.47±0.27‡ 7.84 3.52±0.28‡ 7.86 3.49±0.26 5.65±5.73 −7.34–16.22 0.674 0.51
Ventral striatum (ax) 3.34±0.42‡ 12.45 3.25±0.37‡ 11.51 3.29±0.38 9.53±9.90 −19.37–25.45 0.639 0.31
Medial thalamus 0.65±0.04 6.82 0.68±0.06 8.59 0.67±0.05 4.51±3.10 1.94–9.32 0.011 0.78
Lateral thalamus 0.62±0.05 7.85 0.65±0.04† 5.69 0.64±0.04 6.81±3.63 −3.82–12.04 0.077 0.51
Open in a new tab

BPND, binding potential according to the simplified reference tissue model; COV, coefficient of variation; ICC, intraclass correlation coefficient; M±s.d., mean±standard deviation; VAR, absolute variability.

†Statistically significant difference as compared with original framing with P<0.05 in Student's t-test.

‡Statistically significant difference as compared with original framing with P<0.01 in Student's t-test.

Image Reconstruction With Resolution Modeling

In Figure 2, a typical set of time–activity course data obtained with the PSF reconstruction along with the different framings using the non-PSF reconstruction is seen. Regional ΔAUC(%) values for PSF reconstruction are presented in Table 3.

Regional ΔAUC(%) values have positive means in both the scans except for the lateral thalamus (−1.41±2.25 and −0.77±0.86), with largest increases in the caudate nucleus (8.29±1.17 and 8.66±0.86) and the putamen (7.06±1.02 and 7.50±0.92), whereas the smallest changes were seen in the cerebellum (0.78±1.70 and 1.04±0.74).

The BPND estimates, the mean VAR, and the ICC values for combined ROIs with the PSF reconstruction are presented in Table 5. The PSF reconstruction yielded higher BPND estimates in all other regions expect the lateral thalamus when compared with the original framing using the non-PSF reconstruction. Increases were largest in the caudate nucleus and in the putamen. However, the increased BPND estimates were coupled with increased VAR and thereafter decreased ICC in many of the regions. In striatal regions, the largest increase in the VAR was seen in the caudate nucleus and in the putamen, whereas the VAR was somewhat smaller in the VST. The ICC values were decreased in all regions except the lateral thalamus, with largest decrease in the caudate nucleus.

Table 5. Regional [11C]raclopride BPND values with PSF reconstruction and their test–retest characteristics.

Region Scan 1 (BPND)
 Scan 2 (BPND)
 Both scans (BPND)
 VAR (%)
 100 × (Scan 2− Scan 1)/Scan 1
 t-Test sign. ICC
  M±s.d. COV (%) M±s.d. COV (%) M±s.d. M±s.d. Range    
Putamen 4.75±0.49‡ 10.41 4.93±0.57‡ 11.67 4.84±0.52 5.76±4.32 −3.90–14.08 0.158 0.82
Caudate nucleus 4.21±0.24‡ 5.65 4.29±0.25‡ 5.92 4.25±0.24 4.78±5.44 −4.68–15.26 0.497 0.35
Ventral striatum (cor) 3.65±0.27‡ 7.48 3.71±0.27‡ 7.37 3.68±0.26 5.56±5.57 −6.83–14.77 0.601 0.48
Ventral striatum (ax) 3.57±0.45‡ 12.7 3.46±0.38‡ 10.96 3.52±0.41 9.17±9.60 −20.18–23.31 0.587 0.33
Medial thalamus 0.66±0.04 6.28 0.70±0.06† 8.92 0.68±0.06 6.07±3.88 0.71–10.29 0.007 0.64
Lateral thalamus 0.59±0.06‡ 9.55 0.63±0.05‡ 7.54 0.61±0.05 9.09±3.37 −5.09–13.77 0.021 0.52
Open in a new tab

BPND, binding potential according to the simplified reference tissue model; COV, coefficient of variation; ICC, intraclass correlation coefficient; M±s.d., mean±standard deviation; PSF, point-spread function; VAR, absolute variability.

†Statistically significant difference as compared with original framing with P<0.05 in Student's t-test.

‡Statistically significant difference as compared with original framing with P<0.01 in Student's t-test.

Discussion

In this study, we set out to evaluate the test–retest reliability of the striatal, ventrostriatal, and thalamic D2 binding using [11C]raclopride and the HRRT scanner. The reproducibility of striatal [11C]raclopride binding has been explored earlier in various studies (Volkow et al, 1993; Hietala et al, 1999; Mawlawi et al, 2001; Hirvonen et al, 2003), but not with a high-resolution tomograph. In this study, with the conventional framing and non-PSF reconstruction, the VAR values were 3.7%±4.7% in the caudate nucleus, 4.7%±3.9% in the putamen, and 5.9%±5.3% in the VST (coronal plane). These values are comparable to those reported in an earlier study in which striatal and ventrostriatal D2 receptors were quantified (5.3%±5.2% for the caudate nucleus, 3.5%±3.9% for the putamen, and 8.6%±6.2% for the VST) (Mawlawi et al, 2001). However, the ICC values from our study were slighty lower for every region except for the putamen. This difference may, at least partly, be due to methodological differences such as the bolus plus infusion method used in the previous study (Mawlawi et al, 2001).

A previous study evaluating the reproducibility and validity of the thalamic D2 receptor binding measurements in a test–retest setting found the VAR values to be 8.5%±6.1% for the medial part and 8.9%±5.3% for the lateral part of the thalamus (Hirvonen et al, 2003). In this study, we obtained slightly superior VAR values, with 4.5%±3.1% for the medial and 6.4%±4.5% for the lateral part of the thalamus. The earlier results (Hirvonen et al, 2003) for the caudate nucleus (4.0%±2.2%) and for the putamen (5.3%±3.4%) are well in line with this study. Nonetheless, except for the putamen (0.87), the ICC values were less than those reported earlier for the caudate nucleus (0.96), the putamen (0.84), the medial thalamus (0.84), and the lateral thalamus (0.86) (Hirvonen et al, 2003). The most likely explanation to the lower ICC values in this study is the small between-subject variability in our study group (see Table 2, coefficient of variation (%)) when compared with the earlier study (Hirvonen et al, 2003). Altogether, the VAR values in our study indicate good reproducibility and are well in line with the previous studies.

In this study, we observed an increase in the mean BPND values from the test to the retest scans in all regions except the VST, which might also impair the ICC. However, the increase was statistically significant only in the lateral (t-test, P=0.02) and medial (t-test, P=0.01) thalamus. There were no statistically significant differences either in the specific radioactivities or in the injected masses between the test and retest scans. However, one case was an outlier in the analysis of the injected masses. Theoretically, low specific activity of [11C]raclopride would yield underestimation of the BPND. Exclusion of the low specific radioactivity case from the analysis decreased the difference between mean BPND in the test and retest scans, yielding slightly superior VAR and ICC values (data not shown).

This is the first study measuring the reproducibility of [11C]raclopride binding using the HRRT. There are some previous reports with other radioligands such as [18F]-FDG (fluorodeoxyglucose) showing that the better spatial resolution of the HRRT scanner decreased the PVE when determining metabolic rates in small brain nuclei (Heiss et al, 2004). Dopamine transporter binding was measured using the selective dopamine transporter radioligand [11C]-PE2I, which resulted in a significant increase (30%) in striatal BP values when using the HRRT versus a more conventional scanner (Leroy et al, 2007), and a recent study reported a good test–retest reliability of dopamine transporter binding using [11C]-PE2I with the HRRT (Hirvonen et al, 2008). Accordingly, we also discovered a significant benefit from the better spatial resolution of the HRRT scanner in noticeably higher BPND values compared with previous [11C]raclopride studies with more conventional scanners. For example, we obtained BPND values 20% higher for the caudate nucleus, the putamen, and for the thalamus than the BPND values reported using a lower-resolution scanner (Hirvonen et al, 2003).

The measured [11C]raclopride BPs in this study were high throughout the whole striatum, ranging from 3.22 (VST) to 4.41 (putamen), and moderate in the thalamus (0.65 to 0.67), using the conventional quantification methods. These results are in a good agreement with the previous in vivo studies (Mawlawi et al, 2001; Hirvonen et al, 2003) and also comparable with various in vitro studies (Hall et al, 1988, 1994). The BPND values found in the VST were 77% of that in the putamen, which is greater than the 63% reported earlier (Mawlawi et al, 2001). This could be explained by the increased signal because of the better spatial resolution of the HRRT scanner. Although BPND was higher in the VST, the VAR value (5.93%±5.34%) in the VST (coronal) did not significantly differ from those observed in the larger caudate nucleus and putamen (3.71%±4.72% and 4.70%±3.91%, respectively). Furthermore, the BPND values were higher and the reproducibility was better when delineating VST ROI in the coronal plane than in the axial plane. This is probably because of a more accurate anatomic delineation of the VST in the coronal than in the axial plane.

Instrumentational and Signal Processing Considerations

The small size of the VST renders the signal highly vulnerable to head movement. Therefore, in our study we used individually shaped thermoplastic mask with each subject to minimize head movement. The external motion detector (Polaris Vicra) used in this study showed very little motion during the scans, despite a trend toward increased movement during the course of the scan. Even when being greatest, the movement was less than the spatial resolution of the scanner, thus no movement correction was applied as this potential confound was not obviously affecting the results.

It has been reported that the HRRT image reconstruction suffers from quantification bias in low count statistics, such as we saw in this study (Johansson et al, 2007). Despite the development of algorithms such as variance reduction on randoms (Byars et al, 2005) and ordinary poisson OSEM reconstruction (Comtat et al, 2004), a recent study still reported a noticeable bias in statistical reconstruction as compared with analytical reconstruction in low count statistics (van Velden et al, 2008). However, analytical image reconstruction is not optimal for the HRRT image reconstruction because of gaps in the sinogram data. In this study, we implemented a novel approach in PET data sampling to overcome the bias issue. Our aim was to collect a fixed number of counts in each frame, which is made possible by the list mode event recording. Noise-equivalent count was applied because it takes into account the varying randoms rates and scatter fractions. We extrapolated an NEC threshold per frame from a previous phantom study in which we studied the quantification bias in low statistics (Johansson et al, 2007). In the conventional framing, the threshold of 7 × 106 NECs was not reached during the first 600 seconds of the scan because of dense framing depicting the bolus peak. After that NECs per frame varied below and above the threshold, depending on the subject mass and injected dose. Absolute difference in the injected doses between the test and retest scans ranged between 3 and 113 MBq, with the mean injected dose of 403 MBq. Scans with lowest dosage may yield lower BPND as compared with their counterpart, which would impair the ICC and VAR values. Our NEC-based framing was aimed at removing this biasing effect if it was present in this study. First, to evaluate the effect of the NEC-based framing the bias was estimated directly from the time-activity course (TAC) data (Table 3). It was concluded that the NEC-based framing yielded higher contrast images as compared with the original framing, with statistically significant difference in the high RC regions. Second, the BPND estimates were recalculated from the NEC-based data and the ICC and VAR values were estimated (Table 4). As expected, the observed changes in the AUC were transferred into BPND so that in the high RC regions the BPND estimates were higher as compared with the conventional framing. However, there was no improvement either in the ICC or in the VAR values, which could be expected because of relatively small within-subject differences in injected doses (Table 1). To study the bias further, we used NEC-based framing with thresholds of 5 × 105 to 2 × 106 NECs per frame. Parts of the data were histogrammed and reconstructed with the low count statistics, and bias was estimated directly from the TAC data (Table 3). Now, a statistically significant decrease in the image contrast was seen in the high and low RC regions when the NEC-based framing was compared with the original framing. The BPND estimates were not calculated for low statistics data because there were only a few samples from the beginning and end of each scan. Our observations add evidence to the claim that the current statistical reconstruction suffers from bias, which has significant magnitude, for instance, in follow-up studies if the injected doses vary substantially between scans. Thus, to cope with the bias the injected doses should be equalized, if not succeeded, the NEC-based framing should be used to maintain similar conditions for the statistical reconstruction. Ultimately, new bias-free reconstruction schemes must be explored to remove the count statistics limitation in the HRRT studies.

Detector design of the HRRT aims at removing the PVE from the PET measurement. Conventionally, the PVE is corrected from the collected data and a variety of algorithms have been proposed. An implementation of a scanner-specific PSF-based method was recently made for the HRRT and delivered to the HRRT users (Comtat et al, 2008). It has been reported that the PSF-based method is able to generate higher BP estimates as compared with the conventional reconstruction (Sureau et al, 2008; Comtat et al, 2008; Mourik et al, 2010; Varrone et al, 2009). Approximately, 25% higher BPND estimates in the caudate nucleus, 23% in the putamen, and 10% in the VST using [11C]-PE2I with listmode reconstruction and PSF modeling has been reported (Sureau et al, 2008). In a study including one human subject imaged with [11C]-raclopride ∼10% higher BPND estimates in the caudate nucleus and 6% higher in the putamen were obtained (Varrone et al, 2009). In a study using [11C]-flumazenil in five subjects no noticeably higher distribution volume estimates with PSF as compared with non-PSF reconstruction were obtained (Mourik et al, 2010). The authors concluded that the improvement obtained through the usage of PSF reconstruction is tracer dependent. In this study, we obtained ∼10% increase in the BPND estimates in the caudate nucleus and putamen and 8% increase in the VST, which are in good agreement with an earlier report (Varrone et al, 2009). Moreover, we obtained ∼ 7% lower BPND estimate in the lateral thalamus, and <2% increase in the medial thalamus, which would suggest that the PSF reconstruction alleviates the spill-in effect as well. In this study, our main objective was to evaluate the methods with regard to measurement reproducibility. Although PSF reconstruction provided BPND estimates that should be less contaminated by the PVE, its VAR values and, consequently, the ICC values were somewhat inferior to the non-PSF reconstruction. In the thalamus and VST, the VAR and ICC values were identical between the PSF and the non-PSF reconstructions.

As the line-of-response (LOR) of the HRRT is significantly thinner as compared with conventional scanners, the counting statistics rather than the spatial resolution becomes a limiting factor in the quantitative HRRT studies. In this report, we conclude that the statistical reconstruction of the HRRT is vulnerable to bias in low count statistics, but also, that the PSF reconstruction rather increases than decreases the VAR values. However, the PSF reconstruction was recently evaluated with regard to count statistics (Walker et al, 2009) and it was found to be less vulnerable to low count statistics bias. In this study, we did not evaluate the PSF reconstruction with regard to count statistics, but we observed that both the increment of count statistics and the inclusion of the resolution model in the reconstruction increase the contrast of the images. In addition, the preceding evaluation using list mode reconstruction and the resolution model (Planeta-Wilson et al, 2008) supports the usage of PSF reconstruction in the low statistics HRRT studies.

Conclusions

This study showed that the use of single bolus [11C]raclopride and the high-resolution research scanner HRRT results in good reproducibility of quantification of D2 receptor binding. The use of the HRRT scanner increased the signal from the striatum and the thalamus and resulted in higher BPND values compared with previous studies using scanners with lower resolution. The good test–retest reproducibility of the quantification of the VST is encouraging and shows the feasibility of the use of the HRRT in studying the dopaminergic system in small brain regions. Our results show that [11C]raclopride BP has good reproducibility even when using a high-resolution scanner that allows analysis of small brain structures.

Acknowledgments

The authors thank the personnel of Turku PET Centre for skillful assistance during the scanning.

The authors declare no conflict of interest.

Footnotes

This study was financially supported by the Academy of Finland (project #111879) and by clinical grants of the Turku University Hospital (EVO).

References

  1. Bäckman L, Ginovart N, Dixon RA, Wahlin TB, Wahlin A, Halldin C, Farde L. Age-related cognitive deficits mediated by changes in the striatal dopamine system. Am J Psychiatry. 2000;157:635–637. doi: 10.1176/ajp.157.4.635. [DOI] [PubMed] [Google Scholar]
  2. Belin D, Everitt BJ. Cocaine seeking habits depend upondopamine-dependent serial connectivity linking the ventral with the dorsal striatum. Neuron. 2008;57:432–441. doi: 10.1016/j.neuron.2007.12.019. [DOI] [PubMed] [Google Scholar]
  3. Byars LG, Sibomana M, Burbar Z, Jones J, Panin V, Barker WC, Liow J-S, Carson RE, Michel C. Variance reduction on randoms from coincidence histograms for the HRRT. IEEE Nucl Sci Conf R. 2005;5:2622–2626. [Google Scholar]
  4. Comtat C, Bataille F, Michel C, Jones JP, Sibomana M, Janeiro L, Trebossen R. OSEM-3D reconstruction strategies for the ECAT HRRT. IEEE Nucl Sci Conf R. 2004;6:3492–3496. [Google Scholar]
  5. Comtat C, Sureau FC, Sibomana M, Hong IK, Sjoholm N, Trebossen R. Image based resolution modeling for the HRRT OSEM reconstructions software. IEEE Nucl Sci Conf R. 2008;2008:4120–4123. [Google Scholar]
  6. de Jong HWAM, van Velden FHP, Kloet RW, Buijs FL, Boellaard R, Lammertsma AA. Performance evaluation of the ECAT HRRT: an LSO-LYSO double layer high resolution, high sensitivity scanner. Phys Med Biol. 2007;52:1505–1526. doi: 10.1088/0031-9155/52/5/019. [DOI] [PubMed] [Google Scholar]
  7. DeLong MR. The neurophysiologic basis of abnormal movements in basal ganglia disorders. Neurobehav Toxicol Teratol. 1983;5:611–616. [PubMed] [Google Scholar]
  8. Evarts EV, Wise SP. Basal ganglia outputs and motor control. Ciba Found Symp. 1984;107:83–102. doi: 10.1002/9780470720882.ch6. [DOI] [PubMed] [Google Scholar]
  9. Floresco SB, Onge JRS, Ghods-Sharifi S, Winstanley CA. Cortico-limbic-striatal circuits subserving different forms of cost-benefit decision making. Cogn Affect Behav Neurosci. 2008;8:375–389. doi: 10.3758/CABN.8.4.375. [DOI] [PubMed] [Google Scholar]
  10. Hall H, Farde L, Sedvall G. Human dopamine receptor subtypes—in vitro binding analysis using 3H-SCH 23390 and 3H-raclopride. J Neural Transm. 1988;73:7–21. doi: 10.1007/BF01244618. [DOI] [PubMed] [Google Scholar]
  11. Hall H, Sedvall G, Magnusson O, Kopp J, Halldin C, Farde L. Distribution of D1- and D2-dopamine receptors, and dopamine and its metabolites in the human brain. Neuropsychopharmacol. 1994;11:245–256. doi: 10.1038/sj.npp.1380111. [DOI] [PubMed] [Google Scholar]
  12. Heiss W-D, Habedank B, Klein JC, Herholz K, Wienhard K, Lenox M, Nutt R. Metabolic rates in small brain nuclei determined by high-resolution PET. J Nucl Med. 2004;45:1811–1815. [PubMed] [Google Scholar]
  13. Hietala J, Någren K, Lehikoinen P, Ruotsalainen U, Syvälahti E. Measurement of striatal D2 dopamine receptor density and affinity with [11C]-raclopride in vivo: a test-retest analysis. J Cereb Blood Flow Metab. 1999;19:210–217. doi: 10.1097/00004647-199902000-00012. [DOI] [PubMed] [Google Scholar]
  14. Hirvonen J, Aalto S, Lumme V, Någren K, Kajander J, Vilkman H, Hagelberg N, Oikonen V, Hietala J. Measurement of striatal and thalamic dopamine D2 receptor binding with 11C-raclopride. Nucl Med Commun. 2003;24:1207–1214. doi: 10.1097/00006231-200312000-00002. [DOI] [PubMed] [Google Scholar]
  15. Hirvonen J, Johansson J, Teräs M, Oikonen V, Lumme V, Virsu P, Roivainen A, Någren K, Halldin C, Farde L, Hietala J. Measurement of striatal and extrastriatal dopamine transporter binding with high-resolution PET and [11C]PE2I: quantitative modeling and test-retest reproducibility. J Cereb Blood Flow Metab. 2008;28:1059–1069. doi: 10.1038/sj.jcbfm.9600607. [DOI] [PubMed] [Google Scholar]
  16. Hong IK, Chung ST, Kim HK, Kim YB, Son YD, Cho ZH. Ultra fast symmetry and SIMD-based projection-backprojection (SSP) algorithm for 3-D PET image reconstruction. IEEE Trans Med Imaging. 2007;26:789–803. doi: 10.1109/tmi.2007.892644. [DOI] [PubMed] [Google Scholar]
  17. Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN, Holden J, Houle S, Huang S-C, Ichise M, Iida H, Ito H, Kimura Y, Koeppe RA, Knudsen GM, Knuuti J, Lammertsma AA, Laruelle M, Logan J, Maguire RP, Mintun MA, Morris ED, Parsey R, Price JC, Slifstein M, Sossi V, Suhara T, Votaw JR, Wong DF, Carson RE. Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 2007;27:1533–1539. doi: 10.1038/sj.jcbfm.9600493. [DOI] [PubMed] [Google Scholar]
  18. Johansson J, Oikonen V, Teras M. Quantitative brain imaging using the new, fast iterative histogram-mode reconstruction for the HRRT PET scanner. IEEE Nucl Sci Conf R. 2007;5:3463–3467. [Google Scholar]
  19. Kaasinen V, Rinne JO. Functional imaging studies of dopamine system and cognition in normalaging and Parkinson′s disease. Neurosci Biobehav Rev. 2002;26:785–793. doi: 10.1016/s0149-7634(02)00065-9. [DOI] [PubMed] [Google Scholar]
  20. Kessler RM, Whetsell WO, Ansari MS, Votaw JR, de Paulis T, Clanton JA, Schmidt DE, Mason NS, Manning RG. Identification of extrastriatal dopamine D2 receptors in postmortem human brain with [125I]epidepride. Brain Res. 1993;609:237–243. doi: 10.1016/0006-8993(93)90878-q. [DOI] [PubMed] [Google Scholar]
  21. Lammertsma AA, Hume SP. Simplified reference tissue model for PET receptor studies. Neuroimage. 1996;4:153–158. doi: 10.1006/nimg.1996.0066. [DOI] [PubMed] [Google Scholar]
  22. Langer O, Någren K, Dollé F, Lundkvist C, Sandell J, Swahn C-G, Vaufrey F, Crouzel C, Mazière B, Halldin C. Precursor synthesis and radiolabelling of the dopamine D2 receptor ligand [11C]raclopride from [11C]methyl triflate. J Labelled Comp Radiopharm. 1999;42:1183–1193. [Google Scholar]
  23. Larsen P, Ulin J, Dahlström K, Jensen M. Synthesis of [11C]methyl iodide by iodination of [11C]methane. Appl Radiat Isot. 1997;48:153–157. [Google Scholar]
  24. Leroy C, Comtat C, Trébossen R, Syrota A, Martinot J-L, Ribeiro M-J. Assessment of 11C-PE2I binding to the neuronal dopamine transporter in humans with the high-spatial-resolution PET scanner HRRT. J Nucl Med. 2007;48:538–546. doi: 10.2967/jnumed.106.037283. [DOI] [PubMed] [Google Scholar]
  25. Mawlawi O, Martinez D, Slifstein M, Broft A, Chatterjee R, Hwang DR, Huang Y, Simpson N, Ngo K, Heertum RV, Laruelle M. Imaging human mesolimbic dopamine transmission with positron emission tomography: I. Accuracy and precision of D(2) receptor parameter measurements in ventral striatum. J Cereb Blood Flow Metab. 2001;21:1034–1057. doi: 10.1097/00004647-200109000-00002. [DOI] [PubMed] [Google Scholar]
  26. Mourik JEM, Lubberink M, van Velden FHP, Kloet RW, van Berckel BNM, Lammertsma AA, Boellaard R. In vivo validation of reconstruction-based resolution recovery for human brain studies. J Cereb Blood Flow Metab. 2010;30:381–389. doi: 10.1038/jcbfm.2009.225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Någren K, Truong P, Helin S, Amir A, Halldin C. Experience from two systems for recirculating production of [11C]methyl iodide from target produced [11C]methane. J Labelled Comp Radiopharm. 2003;46 (Suppl. 1:S76. [Google Scholar]
  28. Nuyts J, Dupont P, Stroobants S, Maes A, Mortelmans L, Suetens P. Evaluation of maximum-likelihood based attenuation correction in positron emission tomography. IEEE Trans Nucl Sci. 1999;46:1136–1141. [Google Scholar]
  29. Planeta-Wilson B, Yan J, Mulnix T, Carson RE. Quantitative accuracy of HRRT list-mode econstructions: effect of low statistics. IEEE Nucl Sci Conf R. 2008;2008:5121–5124. doi: 10.1109/NSSMIC.2008.4774388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Sureau FC, Reader AJ, Comtat C, Leroy C, Ribeiro M-J, Buvat I, Trébossen R. Impact of image-space resolution modeling for studies with the high-resolution research tomograph. J Nucl Med. 2008;49:1000–1008. doi: 10.2967/jnumed.107.045351. [DOI] [PubMed] [Google Scholar]
  31. van Velden FHP, Kloet RW, van Berckel BNM, Wolfensberger SPA, Lammertsma AA, Boellaard R. Comparison of 3D-OP-OSEM and 3D-FBP reconstruction algorithms for High-Resolution Research Tomograph studies: effects of randoms estimation methods. Phys Med Biol. 2008;53:3217–3230. doi: 10.1088/0031-9155/53/12/010. [DOI] [PubMed] [Google Scholar]
  32. Varrone A, Sjöholm N, Eriksson L, Gulyás B, Halldin C, Farde L. Advancement in PET quantification using 3D-OP-OSEM point spread function reconstruction with the HRRT. Eur J Nucl Med Mol Imaging. 2009;10:1639–1650. doi: 10.1007/s00259-009-1156-3. [DOI] [PubMed] [Google Scholar]
  33. Volkow ND, Fowler JS, Wang GJ, Dewey SL, Schlyer D, MacGregor R, Logan J, Alexoff D, Shea C, Hitzemann R. Reproducibility of repeated measures of carbon-11-raclopride binding in the human brain. J Nucl Med. 1993;34:609–613. [PubMed] [Google Scholar]
  34. Walker MD, Julyan PJ, Talbot PS, Jones T, Matthews JC. Bias in iterative reconstruction of low-statistics PET data: benefits of a resolution model. IEEE Nucl Sci Conf R. 2009;2009:2857–2863. doi: 10.1088/0031-9155/56/4/004. [DOI] [PubMed] [Google Scholar]
  35. Watson CC. New, faster, image-based scatter correction for 3D PET. IEEE Trans Nucl Sci. 2000;47:1587–1594. [Google Scholar]

Articles from Journal of Cerebral Blood Flow & Metabolism are provided here courtesy of SAGE Publications

RESOURCES