1) Imaging protocol and equipment
The University of Fukui Hospital was the only one institution in Japan that collected 123I-IMP-SPECT data using Siemens’s SPECT with LEHR collimator. According to their procedure of brain perfusion scan, the acquisition time was determined by the SPECT count, that would be the condition of using SMS LEHR with 6 mCi of 123I-IMP administration and counting equivalent to 30 minutes of collection. matrix size128 ×128, zoom = 1.45. The energy window center is 159keV; window width is 20%. Image reconstruction method is FBP, no attenuation correction (AC), no scatter correction (SC). We performed SPECT studies using the SPECT/CT Symbia T6 (Siemens, Erlangen, Germany), equipped with LEHR collimators. The image-processing workstations for SPECT reconstruction is Siemens Syngo.via for MI. SPECT projection data sets were obtained using the following phantom: the Line source Phantom, Cerebral blood flow brain phantom with acrylic cover ( without bone), adult male and elderly female skull phantom, Cold Rods phantom with adult male skull phantom, and 6 chambers phantom with adult male skull phantom. The phantom conditions and the rotation radius (14, 15, 16cm) were changed and collected. And then, we imported these combined projection data into MATLAB (MathWorks, Natick, MA, USA).
For statistical parametric mapping analysis, a reliable repeatability and accuracy SPECT value need to be provided by SPECT image, and the SPECT value is affected by difference in spatial resolution, attenuation by the skull, and scatter of gamma-rays emitted from inside the body. So the quantitative ability of 123I-IMP SPECT were evaluated through the visual judgement, validity of spatial resolutions recovery, linearity of the radioactivity and SPECT value, and the gray-to-white matter count ratio. Finally, we determined the image post-processing parameters most suitable for statistical parametric mapping.
2) Study phantoms
Flash3D reconstruction with Gaussian filters: 6mm, 7mm, 8mm, 10mm, 12mm,13mm, and 14mm were used for the four phantoms (Fig. 1). Spatial resolution was evaluated using the transaxial images acquired from the Line source phantom. Visual inspection was evaluated by the use of a Cold rods phantom. Linearity of the radioactivity and SPECT value was evaluated by the use of a Six chambers phantom. And the gray-to-white matter count ratio was evaluated by the use of Cerebral blood flow brain phantom with skull phantoms.
2 − 1) Line source Phantom
The phantom consists of five thin-walled plastic capillary tubes were used and the capillary tubes were filled with 123I-IMP solution (Fig. 1A).
2–2) Cold Rods phantom
This phantom contains six arrays of cold rods with diameters of 6mm, 8mm, 10mm, 12mm, 14mm, 16mm (Fig. 1B). An123I-IMP solution was injected into the phantom. System spatial resolution is evaluated according to the number of cold rod arrays that can be resolved visually.
2–3) Cerebral blood flow brain phantom and skull phantoms
A cerebral blood flow brain and two skull phantoms (IB-20, Kyoto Kagaku) simulating adult male and elderly female were used. The thickness of each skull was 1 cm (Fig. 1C). An 123I-IMP solution was injected into the cerebral blood flow brain phantom to provide a radioactivity concentration of gray matter to white matter ratio of 4:1. The Cerebral blood flow brain phantom was enclosed in either skull phantom or an acrylic cover for all tests.
2–4) Six chambers phantom
There are sections 1–6 in this phantom, each volume is around 120ml, respectively. An 123I-IMP solution was injected into the phantom to provide a radioactivity concentration ratio of three regions of 1:2:3 (Fig. 1D).
3) Phantom preparation, image acquisition and processing
To prevent the IMP from adsorbing to the phantom wall, a solution containing 80 g of disodium hydrogen citrate per 3 Liters of solution, were used in the preparation for all phantoms.
For image acquisition. The energy window recommended by Siemens was used, the window center energy is 159keV ± 7.5%, lower scatter window width is 15%, and upper scatter is 15% (Fig. 2). The data of University of Fukui Hospital was converted after 3D-SSP analysis, so the pixel size is 2.25mm rather than original pixel size 3.3mm. Besides, the image acquisition setting and processing without SC and CT-AC (Table 1), was using an energy window of 159kev ± 10%, that different to our study. So we need to convert the counts per pixel. One based on the pixel size difference, the counts per pixel can be derived from Eq as:
Table 1
| Siemens_LEHR_NDB_AC-_FBP cutoff 0.28Nyquist | | | |
| Mean counts of areas (ACA, MCA, PCA) in brain | |
| corrected to noon standard | | | | |
| Teritory | pixel size | slice thickness | number of pixels | n | Mean counts | SD |
| cACA_R | 2.25mm | 2.25mm | 139 | 17 | 72.50 | 20.77 |
| cACA_L | 2.25mm | 2.25mm | 139 | 17 | 73.20 | 21.15 |
| cMCA_R | 2.25mm | 2.25mm | 377 | 17 | 78.17 | 22.79 |
| cMCA_L | 2.25mm | 2.25mm | 377 | 17 | 76.30 | 22.66 |
| cPCA_R | 2.25mm | 2.25mm | 210 | 17 | 82.53 | 26.49 |
| cPCA_L | 2.25mm | 2.25mm | 210 | 17 | 82.27 | 24.47 |
| mean | | | | | 77.5 | 23 |
{ (3.30)3/(2.25)3 } × (77.5 ± 23)/pixel = 244 ± 72 /pixel
And the other based on energy window difference (Table 2). The counts per pixel can be derived from Eq as:
Table 2
Total counts are difference due to main energy window width is different.
| 2min Tomo scan, | Image reconstruction by FBP without SC and CT-AC. | Image reconstruction by Flash3D with SC and CT-AC. |
| Energy window width | 20% | 15% |
| Total counts | 1232k | 1037k |
| Rotio | 1 | 0.84 |
244 ± 72 /pixel × 0.84 = 205 ± 60.5/pixel.
Finally, the image was reconstructed by Flash3D with CTAC.
3 − 1) Acquisition and processing of line source phantom
We injected 95MBq (1.5 ml) 123I-IMP solution into the phantom. Line source images were acquired following the National Electrical Manufacturers Association (NEMA) recommendations, the detector rotation radius setting used in the image acquisition were: 14cm, 15cm, and 16cm. Line source images were reconstructed using Flash3D with Gaussian filters: 6mm,7mm,8mm, 10mm, 12mm,13mm, 14mm, respectively. The SPECT and CT data acquisition was carried out by Flash3D of Siemens recommendation, and processing with the addition of the Flash3D and CTAC method.
The Line spread function (LSF) was measured, the full width at half maximum (FWHM) was calculated.
3 − 2) Acquisition and processing of cold rods phantom
Sodium hydrogen 21g, and 84MBq 123I-IMP solution were mixed with 790ml of water, and then injected into phantom. The acquisition time was determined as same as Cerebral blood flow brain phantom and skull phantoms.
3–3) Acquisition and processing of cerebral blood flow brain phantom and skull phantoms
We prepared Sodium hydrogen 16g into 600ml of water for gray matter, 9g into 340ml of water for white matter. And injected 70MBq 123I-IMP solution into 600ml of water for gray matter, 10MBq into 340ml of water for white matter.
So the gray matter to white matter ratio is: \(\:=\:\frac{70\text{M}\text{B}\text{q}/600\text{m}\text{l}}{10\text{M}\text{B}\text{q}/340\text{m}\text{l}}\) = \(\:=3.96\)
In addition, the acquisition time was determined as follows; First, we calculated the count per pixel from the NDB of the cerebral blood flow in Fukui University. Second, we performed the test scan of the phantom for two minutes, using the same acquisition parameters as in the NDB of Fukui University. And then calculated the number of counts per pixel. Finally, we determined the required acquisition time, using the equation as follow:
Acquisition time: T\(\:\text{a}\text{c}\text{q}=(\frac{Cndb}{Ctest}\times\:Ttest\))
T\(\:acq\): acquisition time
\(\:Cndb\) : (Count of gray matter in NDB)
\(\:Ctest\) : (Counting gray matter in test scans)
\(\:Ttest\) : (Imaging time of test scan)
Acquisition using the cerebral blood flow brain phantom with acrylic cover, skull adult male cover, and skull female cover are performed. After all, two 1 ml samples were taken from the phantom’s gray and white matter after each acquisition, to measure the accurate gray-to-white matter radioactivity count ratios using a dose calibrator.
3–4) Acquisition and processing of sections 1–6 phantom
Section 1: Sodium hydrogen 4g and 14.65MBq 123I-IMP solution in 130ml of water;
Section 2: Sodium hydrogen 4g and 29.5MBq 123I-IMP solution in of water 130ml of water;
Section 3: Sodium hydrogen 4g and 49.2MBq 123I-IMP solution in 130ml of water. Concentration ratio for section 1, section 2 and section 3 were, 1:1, 2:1 & 3.3:1, respectively. Six chambers phantom with skull adult male cover was performed, the acquisition time was determined using the method, as in the Cerebral blood flow brain phantom. After all, two 1 ml samples were taken from the phantom’s three regions after each acquisition, to measure the accurate three regions radioactivity count ratios using a dose calibrator.
4) Phantom data analysis
Data in DICOM (Digital Imaging and Communications in Medicine) format were imported into MATLAB (MathWorks, Natick, MA).
4 − 1) Analysis of Line source phantom
The Line spread function (LSF) was measured, and the full width at half maximum (FWHM) was calculated.
Comparing the differences of FWHM value with different rotation radius using Gaussian filter from 6mm to 14mm, to find out the best constant value of Gaussian filter.
4 − 2) Analysis of Cold Rods phantom
Analysis were performed by two independent nuclear medicine physicians using visual judgement, according to image quality and constant under different rotation radius.
4 − 3) Analysis of Cerebral blood flow brain phantom and skull phantoms
To evaluate the correction accuracy of the Siemens’s TEW and CTAC methods, regions of interest (ROI) are placed on the gray and white matter of each SPECT image, as shown in Fig. 3.
The number of counts in each region was measured and the ratio of gray-to-white matter was calculated. Gray-to-white matter counts ratio were calculated for Siemens’s TEW + CTAC, and for the acrylic cover (no bone) and each skull phantom under different rotation radius and gaussian filter, respectively.
4–4) Analysis of 6 chambers phantom
To evaluate the link between SPECT Value (counts) and radioactivity (MBq) of 123I-IMP, regions of interest (ROI) are placed on the three regions of each SPECT image, as shown in Fig. 4. Using different size of ROI mask, and were drawnROI mask. The counts for each region was measured.
The radioactivity values of 123I-IMP are plotted on the x-axis. SPECT values are plotted on the y-axis.