The aim of the present study was to reproduce previous findings [6] and to evaluate additional phantom measurements relevant for DAT imaging. Therefore, in addition to typical NEMA testing setup for gamma cameras (NU 1-2012) [7], several tests were adapted from NU 2-2012 [8], originally designed to evaluate performance of PET systems, to augment conventional SPECT testing. It should be noted that extended modifications had to be made to NEMA testing, in order to accommodate for the geometry of the InSPira HD system. In addition to NEMA testing, striatal phantom measurements were performed and a DAT scan from a healthy control and a patient were retrospectively included to illustrate the clinical applicability of the InSPira HD system for DAT imaging. DAT imaging and processing were executed according to guidelines published by the European Association of Nuclear Medicine [9]. See below for details on the phantom measurements (Additional file 1).
The InSPira HD system
High-resolution imaging is achieved by the unique design of the detector ring of the InSPira HD (Fig. 1). The detector ring consists of two clamshells, each containing 12 fanbeam collimators. At start position, the two clams touch and the collimators are focused at the center of the ring, achieving a focal point of 3 mm in diameter. During acquisition, the gantry rotates and simultaneously the two clams are moved outward leading the two focal points of the clam’s collimators to follow spiral trajectories over the field of view (FOV). The spatial resolution in the resulting image slice is therefore determined by the speed at which the clams move outward. After acquiring the first slice, the camera moves in axial direction to the position for acquiring the adjacent slice, and so on. A proprietary iterative reconstruction algorithm tailored to this unique method of spatial sampling is used to reconstruct the data into 3D images. This iterative reconstruction algorithm is based on a maximum a posteriori (MAP) estimation. It includes a point spread function (PSF), which is defined as the detector response to an impulse activity source point placed in the scanner FOV. Attenuation correction is performed using either a CT previously obtained from the patient/phantom or a deformable CT template. Scatter correction was not available on the system and therefore not performed.
Acquisition and reconstruction parameters
For the striatal phantom and clinical DAT imaging, a slice time of 180 s and slice spacing of 4 mm was used, resulting in a ~ 30 min scan duration, while maintaining sufficient spatial resolution. For all other scans, a slice time of 240 s and a slice spacing of 3.125 mm were used. Energy windows were 20% centered around 140 and 159 keV for the 99mTc and 123I energy windows. Three reconstruction approaches were investigated. A “clinical” reconstruction approach, with 60 iterations and voxel size of 2.083 × 2.083 x slice thickness mm3, as recommended by the vendor. A “research” reconstruction using a developers program (80 iterations with an increased sampling rate in projection space; voxel size 3.125 × 3.125 × slice thickness mm3). To assess maximum achievable resolution of the system a “high-resolution” (1000 iterations) reconstruction approach was examined, which did not include attenuation correction, which was only used for point and line sources (voxel size of 2.083 × 2.083 × slice thickness mm3). For all other measurements, attenuation correction was used in the reconstruction by importing a CT-scan of the phantom.
Phantom preparation and positioning
3D resolution in air
The tips of three capillaries with an inner diameter of 0.8 mm were filled with 99mTc (~ 1.5 MBq) to create point sources to assess resolution in air in x/y in-plane and z axial direction. The sources were positioned in the same coronal plane such that the middle source was positioned in the center of the FOV, one source was positioned approximately 7.3 cm left and 5.0 cm cranial of the central source, and one source was positioned 7.3 cm right and 5.0 cm caudal. All capillaries were attached to the table such that the point source was surrounded by air. A scan of 40 slices was acquired. Resolution was determined by calculating FWHM in three dimensions.
Resolution in air
Spatial resolution in air across the FOV was calculated using line sources in air. Ten capillaries with an inner diameter of 0.2 mm were filled with 99mTc (~ 1.2 GBq/ml). The capillaries were inserted into holes drilled into a round acrylic disk from where they suspended in air, located at the center of the FOV, 2, 4, 6, and 8 cm to the left and right, 3, 6, and 9 cm above and below the center. A single slice scan was recorded. FWHM in x and y direction was calculated.
Resolution in water
Three line sources in water were used to determine the spatial resolution in a scattering medium. Three capillaries with inner diameter of 0.8 mm were filled with 99mTc (~ 23 MBq) and inserted into a cylindrical water-filled phantom with a diameter of 14 cm. One capillary was placed in the center of the cylinder and the two capillaries were placed at 5 cm distance from the center with a 90° angle between them. First, a scan with no background activity concentration in water was performed. Next, two scans with background activity concentration of 0.1 and 1.0% of total activity in the capillaries were performed. Average and standard deviation (SD) FWHM in x and y direction was calculated for three consecutive slices. For scans with background activity, mean background was defined in three circular ROIs with diameter of 12.5 mm in the slice with peak values, and subsequently subtracted from all pixels, before calculating FWHM.
Contrast
A Jaszczak phantom with a diameter of 140 mm was custom made by Neurologica to fit the size of the InSPira HD. Either fillable spheres (31.2, 24.9, 19.7, 15.7, 12.4, and 9.8 mm inner diameter) or rods (11, 9.5, 7, 6, 4.5, and 3 mm diameter) were inserted. Spheres were filled with a 37.5 kBq/ml 99mTc solution and the background compartment with a 8.4 kBq/ml 99mTc solution, resulting in a sphere-to-background ratio of 4.4 to 1. A scan of 30 axial slices centered at the hot spheres was acquired. On the reconstructed image, a series of circular regions of interest (ROIs) were projected onto the spheres, selecting the slice with peak value. Six ROIs were projected between the spheres to determine background concentration (all 17.6 mm diameter, except one with 13.6 mm diameter in order to fit between the two largest spheres). Contrast recovery coefficients (CRC) were then calculated using Eq.
1
below, where pixcountsphere,J represents the measured activity concentration in sphere J and activitysphere,J represents the actual activity concentration in the sphere J. Similarly, pixcount_background represents the mean activity concentration from the six background ROIs, and activity_background is the actual activity concentration in the background.
$$ \mathrm{CRC}=\frac{\mathrm{pixcoun}{{\mathrm{t}}_{\mathrm{sphere}}}_J/\mathrm{activit}{{\mathrm{y}}_{\mathrm{sphere}}}_J}{\mathrm{pixcoun}{\mathrm{t}}_{\mathrm{background}}/\mathrm{activit}{\mathrm{y}}_{\mathrm{background}}} $$
(1)
Visual assessment of spatial resolution
The custom Jaszczak phantom with the cold rods insert was used to assess visual resolution. The phantom was filled with a solution of 99mTc at a concentration of 46.76 kBq/ml. From the reconstructed image, four adjacent slices that showed the rods section with the best contrast were selected and summed. The resulting image was examined visually to identify the smallest rods that could be discerned.
Uniformity
The uniform area of the custom Jaszczak phantom was used for assessment of uniformity. Four slices containing the uniform area were summed. Uniformity was assessed qualitatively by plotting a 6.25 mm wide profile across the phantom. A quantitative measure of uniformity (coefficient of variation (CV)) was calculated using Eq.
2
, where SDNROI is the standard deviation of the voxel values in a 100-mm-diameter circular ROI, placed within the central slice and MeanNROI the mean voxel value.
$$ \mathrm{CV}\ \left(\%\right)=\frac{{\mathrm{SD}}_{\mathrm{N}}\mathrm{ROI}}{{\mathrm{Mean}}_{\mathrm{N}}\mathrm{ROI}}\times 100\% $$
(2)
Striatal phantom
A striatal head phantom (anthropomorphic striatum phantom; Radiology Support Devices Inc., Long Beach, CA, USA [RS-901 T]) was filled with 123I dissolved in water at a concentration of 9.1 kBq/ml for the background, and 42 and 32 kBq/ml for the left and the right striatal compartment, respectively, yielding a striatal-to-background ratio of 4.6:1 and 3.5:1 [10, 11]. In addition, measurements with background concentrations of ~ 5 and ~ 15 kBq/mL were included to assess linearity of quantitative values. The Brain Registration and Analysis Software Suite (BRASS™, HERMES Medical Solutions, Sweden) was used, which fits the data to a template containing a number of volumes of interest (VOIs). CRC was calculated for left and right striatal VOIs using Eq.
1
. The specific to non-specific binding ratio was calculated in bilateral striatum as follows: (total binding in striatum–non-specific binding in occipital cortex) /(non-specific binding in occipital cortex). Linearity of quantitative values (i.e. the association of the specific to non-specific binding ratio, with the ratio of the true activity concentration) was assessed using Pearson’s correlation.
Typical clinical example
To illustrate the clinical applicability of the InSPira HD system, we retrospectively included a DAT scan from a healthy control and a patient. The patient was scheduled for a routine 123I-ioflupane SPECT scan on the NeuroFocus system for clinical evaluation of possible parkinsonism, and the healthy control participated in a research study on the NeuroFocus. For both subjects, the scan on the InSPira was obtained after the NeuroFocus scan was completed (approx. 4 h after the injection of ~ 111 MBq 123I-ioflupane). Acquisition parameters for the InSPira HD were adopted from the standard clinical acquisition parameters on the NeuroFocus; a slice timing of 180 s and slice spacing of 4 mm was used for both systems resulting in a ~ 30-min scan duration, whilst maintaining sufficient spatial resolution. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional committee.