We prospectively and consecutively included patients in the diagnostic work-up for Parkinsonism at neurological departments or outpatient neurological clinics referred for DAT imaging at the Department of Clinical Physiology and Nuclear Medicine, Copenhagen University Hospital Bispebjerg from October 2017 to August 2018. Exclusion criteria were severe motor or cognitive disabilities, and more than 2 h of transport time to avoid additional scanning procedures for this vulnerable group. Patients were asked whether they were undergoing treatment with selective serotonin reuptake inhibitors or amphetamine-like medication, but this was not an exclusion criterion. Consents to participate were obtained from all individuals after receiving oral and written information according to national regulations, and the study was registered at the The Committees on Health Research Ethics, Capital Region of Denmark (ID: 17,026,292) that waived need for approval. All data were handled according to regulations by The Danish Data Protection Agency.

Radiosynthesis of [

A one-step, one-pot radiosynthesis was used for the production of [18F]FE-PE2I based on the method reported by Stepanov and co-workers [26]. Briefly, nucleophilic substitution of a tosyl group with [18F]fluoride was followed by high performance liquid chromatography (HPLC) purification (HPLC column:(Onyx™ Monolithic C-18, 100 × 10 mm; flow: 5.5 mL/min). The [18F]FE-PE2I containing fraction was collected through a 0.22 µm sterile filter directly in the final product vial containing 15 ml sterile sodium phosphate buffer, thus providing the final product solution (23 mL) containing approximately 6% ethanol. Further details are reported elsewhere [27].

I]FP-CIT acquisition, reconstruction and semiquantitative analysis

Patients received 200 mg of sodium perchlorate i.v. 10–15 min before the radiotracer injection in order to block uptake of 123I in the thyroid. The SPECT scan was performed 3 h after injection of 185 MBq (5 mCi) [123I]FP-CIT and SPECT image acquisition of 30 min was carried out with a PRISM 3000XP (Marconi, Phillips) triple-headed gamma camera equipped with low-energy, ultra-high-resolution fan-beam collimators. SPECT acquisitions were performed using a full 360° rotation. A 153Gd source was used for transmission scan. Image reconstruction was performed using iterative reconstruction with corrections for scatter and non-uniform attenuation. The number of iterations was 20 for the transmission data and 4 for the emission data. Pixel size after reconstruction was 3.1 mm in each direction. A 3D low-pass filter (cut-off 0.40, order 4.0) was employed and the data reformatted in oblique slices along the orbitomeatal line with a slice thickness of 6.2 mm. Semiquantitative analysis was performed as previously described [28, 29]. The slice with the highest maximum pixel value was selected and the neighbouring slice with the highest maximum pixel value. The two selected slices were added yielding a single, 12.4 mm thick slice, which was used for further analysis. Four regions of interest (ROI) representing striatum (caudate nucleus and putamen) bilaterally were placed. The shape and size of the ROIs were unchanged, but the ROIs could be rotated and translated to fit the location of the basal ganglia (Fig. 1). Further, a fixed region for non-specific uptake was placed in the occipital lobe as this is a standard DAT SPECT reference region visible within the same slice. The set of these five regions was created to mimic the ROIs defined in the initial clinical verification of the [123I]FP-CIT [30]. The ratio of specific striatal and putamen uptake to non-specific uptake, the specific binding ratio (SBR), and the putamen-to-caudate ratio were calculated.

Fig. 1
figure 1

Delineations of anatomical structures. Top row: Five regions of interest (ROI) representing caudate nucleus and putamen bilaterally as well as the occipital lobe were placed on a fused section of the [123I]FP-CIT SPECT. Bottom: Three consecutive axial slices (superior—> inferior) of a [18F]FE-PE2I PET dataset in MNI space illustrating the automated volume of interest (VOI) delineation of putamen, caudate nucleus and cerebellar grey matter

F]FE-PE2I acquisition and reconstruction

PET acquisition was performed on a separate day in a Discovery 710 or MI PET/CT (GE Healthcare, Milwaukee, USA). Patients received a median dose of 192 MBq [18F]FE-PE2I (range 40–239 MBq), and 17 min after administration, a 25-min list-mode PET acquisition was performed. All PET scans were preceded by a low-dose CT for attenuation correction and a diagnostic CT-scan. Data were reconstructed into 3D datasets using a commercial implementation (“Q.Clear”) of a block-sequential regularized expectation maximization algorithm with a regularization parameter “β” of 250. In addition to the recommended interval (17–42 min [20]), a second dataset was reconstructed covering the interval from 30–40 min. The two datasets were compared to evaluate if a shorter scan time is feasible without loss of diagnostic accuracy. Table 1 lists practical differences between the clinical setups/workflows on SPECT and PET.

Table 1 Comparison between the two clinical setups/workflows

F]FE-PE2I volumes of interest

Delineation of volumes of interest (VOI) on static [18F]FE-PE2I PET images was performed using an in-house developed automated segmentation algorithm targeting the caudate nuclei, putamina and cerebellar grey matter including vermis in order to compute imaging metrics comparable to [123I]FP-CIT SPECT. Cerebellum has previously been validated as a reference region without specific binding in an autoradiographic blocking study using [125I]PE2I [31] and was therefore chosen instead of the occipital lobe used for SPECT. These metrics were the SBR’s of the caudate and putamen relative to cerebellar grey matter and the putamen/caudate ratios for each hemisphere. The automatic segmentation method was based on an in-house created atlas with delineations of the above structures from 37 healthy elderly subjects using Joint Label Fusion [32]. These subjects all had matching CT and high resolution T1-weighted MRI which allowed for the target structures to be segmented from MRI using FreeSurfer [33] and transferred rigidly onto the co-registered CT scans. The cerebellar grey matter VOIs were used directly, while the putamen and caudate delineations were filtered with a 6 mm Gaussian kernel to mimic PET resolution. The resulting segmentations were manually corrected when necessary and verified by a nuclear medicine expert specialized in neuroimaging (IL). Individual delineations from the 37 subjects were transformed to MNI space constituting the templates for the automatic segmentation.

For each study patient, the brain was initially extracted from CT using Brain Extraction Tool [34] of the FSL software suite [35] after thresholding and smoothing [36] and was affinely registered to an MNI template using correlation ratio as cost function. Applying the STEPS algorithm [37] (Markov Random field prior value: 5, kernel size: 5), the initial segmentations of putamen, the caudate nucleus, and cerebellar grey matter were applied. To obtain the final segmentations, the PET image was rigidly registered to the CT image, and the transformation aligning CT image to CT template was applied to the resulting PET image. The resulting spatially normalized PET image was then given as input to the Joint Label Fusion algorithm as implemented in ANTs [32] with a search and patch radius of 5 (Fig. 1). All quantitative measurements were completed in native space after reverse transformation. Median values were used to reduce the influence of voxels with extreme values.

Interpretation of images

SPECT scans were categorized as either normal, PS, vascular, or mixed independently by two experienced specialists in nuclear medicine with 10 and 25 years of experience in neuroimaging, respectively (KK,LF). As PET is the new modality, the scans were categorized into the same classifications by two additional readers with 5 and 15 years of experience in neuroimaging (MRJ, LM) to assess interreader agreement. Reading of images was based on visual assessment in accordance to practice guidelines [38, 39], blinded for the results of the other reader and the other ligand. For SPECT assessment, semiquantitative measures were incorporated in the clinical reading as it has previously been described to influence clinical interpretation [40] and structural changes on CT/MRI preceding the SPECT scan were taken into account if available mimicking clinical routine. Specific attention was paid to structural lesions in the basal ganglia and to the occipital lobe that served as reference region. Clinical cut-off values for normal and abnormal binding were not available for [18F]FE-PE2I and thus semiquantitative values were not included in the interpretation. Although six of the patients received injected doses below 100 MBq, the readers deemed all the scans to be of diagnostic quality for visual interpretation. Structural changes on the CT performed with the PET scan were included in the visual assessment of [18F]FE-PE2I as described above. Neurodegenerative disease was suspected in case of significant posterior-anterior gradient and left–right asymmetry, while cerebrovascular cause was suspected in case of larger infarctions on structural imaging co-localized with reduction in tracer uptake. In case of disagreement between readers and modalities, consensus reading was performed.


A significance level of 0.05 was used throughout. Cohen’s kappa was used to compare agreement between modalities, and Fleiss’ kappa was used to compare agreement between readers using SPSS (IBM SPSS Statistic version 25). For analyses of semiquantitative values, the six patients with injected activity below 100 MBq were excluded. The effect sizes for discrimination between abnormal and normal scans using putamen/caudate nucleus ratio and putamen SBR for the worst hemisphere were calculated for the two PET reconstructions (30–40 min versus 17–42 min) and for SPECT versus PET using Glass’ ∆ calculated as (Meanabnormal-Meannormal)/SDnormal (R Core Team, 2017; R Foundation for Statistical Computing, Vienna, Austria; https://www.R-project.org). The study complies with the standards of reporting diagnostic accuracy studies (Additional file 1: Table S1).

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.


This article is autogenerated using RSS feeds and has not been created or edited by OA JF.

Click here for Source link (https://www.springeropen.com/)