Validation of \(^{99m}\)Tc and \(^{177}\)Lu quantification parameters for a Monte Carlo modelled gamma camera

This study was conducted in two parts: experimental data acquisition and Monte Carlo simulations. In each part both \(^\)Tc and \(^\)Lu radioisotopes were studied, for a total of 40 experimental scans and 140 simulation runs. The information about the isotopes’ half-lives, their main gamma emissions and the maximum energy of their beta emission are summarized in Table 1.

Table 1 Decay characteristics of both Tc-99m and Lu-177; data from [25, 26]

The SPECT/CT scanner used for the experimental measurements is a Siemens Symbia Intevo Excel [13] provided by Nuclear Medicine Unit, University Hospital of Ferrara (Italy). The system consists of two gamma camera detectors with NaI crystals (FOV 53.3x38.7 cm). The gamma camera parameters are listed in Table 2. The so-called “step and shoot” technique was used for the tomographic studies. The CT was performed after the SPECT acquisition, with a 110 kVp voltage and Care Dose 4D system, an automated exposure control which ensures constant image quality over all body regions at the lowest possible dose. The Symbia Intevo Excel was equipped with a Low Energy High Resolution (SY-LEHR) collimator for \(^\)Tc studies and with a Medium Energy Low Penetration collimator (SY-MELP) for \(^\)Lu studies. All measurements and simulations were performed with a 20% energy window centred over the 140.5 keV photopeak of the \(^\)Tc, and with two 15% energy windows centred over the 113 keV and 208 keV photopeaks of the \(^\)Lu. A Mec Murphil MP-DC-Chamber dose calibrator has been used for the activity measurement. The accuracy of the dose calibrator is better than 5% as stated from the last quality control test performed on it. All activity measurements were repeated five times to improve the statistical inaccuracy of the measurement. The \(^\)Tc radioisotope has been obtained as sodium perthecnetate (\(Na[^Tc]O_4\)) from \(^Mo/^Tc\) generator (Ultratechnekow, CURIUM, Netherlands), while the \(^\)Lu has been obtained as Lutetium chloride (\([^Lu]Cl_3\)) (EndoLucinBeta, ITM, Munich, Germany).

Table 2 Main Symbia Intevo specifications, taken from Symbia T series data sheet.Phantom experiments for \(^\)Tc and \(^\)Lu: planar imaging

Planar measurements aim at the evaluation of the fundamental SPECT features: spatial resolution and sensitivity, which are defined by the scintillation crystal, the collimator and the photodetector. The planar imaging procedures were performed according to the recommendations found in the report no.177 of AAPM “Acceptance Testing and Annual Physics Survey Recommendations for Gamma Camera, SPECT, and SPECT/CT Systems” [14]. All planar acquisitions were performed on a single detector only, as the gamma camera acceptance tests showed a slight difference between the system’s two detectors. All measurements have been repeated three times.

Extrinsic spatial resolution measurement

The system spatial resolution was measured using two capillary tubes with an inner diameter of 1 mm. The first tube was filled with 30 ± 1 MBq of a \(^\)Tc solution and placed on a low density support at a distance of 10.0 ± 0.5 cm from the collimator. Planar images were acquired in a 512x512 image matrix with a pixel size of 1.2x1.2 \(mm^2\) until the highest pixel value in the line image exceeded 1000 counts.

A second capillary tube was filled with a 130 ± 7 MBq of a \(^\)Lu solution, and planar images were acquired, as before. The spatial resolution was measured by drawing a horizontal profile across the image of the capillary tube in three different positions in order to compensate for the possible non-uniformity in the tube filling. The line profile was fitted with a Gaussian function and the full width at half maximum (FWHM) and the full width at tenth maximum (FWTM) values were calculated. The reference value provided by SIEMENS for the extrinsic spatial resolution with a LEHR collimator and a capillary tube filled with a \(^\)Tc source is 7.5 mm at 10 cm.

System sensitivity measurement

A Petri dish with an inner diameter of 10 cm was filled with 25.0 ± 1.3 MBq of a \(^\)Tc solution to a depth of 4 ± 1 mm. The dish was placed on low-density support made of polystyrene foam at a distance of 10.0 ± 0.5 cm from the collimator. Planar images were acquired in a 128x128 image matrix with a pixel size of 4.8x4.8 \(mm^2\) until the total counts in the image exceeds 1 million. A background image was acquired for the same time after removing the radioactive source. A second Petri dish was filled with 30.0 ± 1.5 MBq of a \(^\)Lu solution, and planar images were acquired, as before. The total net counts over the detector’s useful field of view (UFOV) was obtained and the sensitivity was calculated as follows,

$$}\;[}/MBq] = \frac}\; }\; }}}}(MBq) \cdot } \;} (s)}}}$$

(1)

The reference value provided by SIEMENS for the sensitivity with a LEHR collimator for \(^\)Tc source is 91.8 cps/MBq at 10 cm.

Dead time

Dead-time count loss may result in significant quantitation inaccuracy in SPECT imaging. In fact, at high count rates, the scintillation camera cannot be able to separate temporally all the incoming events, hence the count rate will decrease. This means that the gamma camera sensitivity is a diminishing function of the count rate. To estimate the dead-time effects on the gamma camera quantitation, additional measurements of the sensitivity have been performed for \(^Tc\) by using values of activity between 4 MBq to 3500 MBq. The TEW scatter correction were applied to the measured photopeak counts to remove the scatter or pile-up events could accumulate under the photopeak [15].

Phantom experiments for \(^\)Tc and \(^\)Lu: tomographic imagingCalibration factor measurement

A cylindrical Jaszczak SPECT Phantom (Fig. 1) deprived of all inner inserts has been employed to obtain the CF. The cylinder was filled with a 6800 ml solution of distilled water, 350 MBq of \(^\)Tc.

Fig. 1figure 1

Experimental Setup Example of experimental configuration: a the uniform phantom is shown; b the acquisition geometry for Tc-99 m

Similarly, the Jaszczak phantom was filled with a 6800 ml solution of distilled water (6720 ml), 430 MBq of \(^\)Lu from a certificated vial (accuracy of ±10%), and 67 ml of HCl (37%). The hydrochloric acid was added to prevent lutetium accumulation on phantom surfaces and to ensure a homogeneous radioactive solution.

For both \(^\)Tc and \(^\)Lu, the SPECT/CT acquisitions were performed via the Siemens Symbia Intevo Excel with the step-and-shoot technique. Each tomographic acquisition consisted of 64 projections over 360\(^\) projections performed maintaining a constant distance of 25 cm between the center of the cylinder and the lower part of the detector head. The acquisition time was 20 s for \(^\)Tc and 30 s for \(^\)Lu respectively.

The reconstruction of the projected images was performed with the built-in software from the vendor, Siemens Flash-3D, based on the OSEM-3D iterative reconstruction technique [16]; employing 10 iterations and 8 subsets were chosen. CT-based attenuation-, window-based scatter- and CDR corrections were applied during the reconstruction process. Flash-3D models CDR in both transverse and axial directions.

The scatter correction for Technetium was performed via the DEW (Double Energy Windows) technique with the use of the PW (Photopeak Window) and the LSW (Lower Scatter Window).

The scatter correction for Lutetium was performed via the TEW (Triple Energy Windows) method for the 113 keV peak and the DEW method for 208 keV peak; the widths of each photopeak window are reported in Table 3. A Gaussian post-reconstruction filter with 4.8 mm FWHM was applied to the reconstructed volume.

Table 3 Photopeak Window (PW), Low Scatter Window (LSW), Upper Scatter Window (USW) for \(^\)Tc and \(^\)Lu main peaks are listedRecovery coefficients measurement

To obtain the RC coefficients, the absolute quantification of \(^\)Tc was performed via a Jaszczak SPECT Phantom with six hot spheres.

The phantom was placed in the centre of the field of view. Acquisitions were performed with the same settings as those of the uniformly filled Jaszczack phantom previously described and conducted with 64 projections of 20 s scan time.

The energy windows were the same as those set for the CF evaluation and are listed in Table 3. The spheres’ volume and the background activity are listed in Table 4. Each activity value reported in Table 4 is the mean of five different measurements.

Table 4 The six spheres of the Jaszczack phantom with their respective activity and the background are shown. Each of the value reported in this Table is the mean value of five different measurements, with a standard deviation less than 1%. These errors must be added to the 10% error on the activity as stated by the dose calibrator manufacturer

For the evaluation of the RC coefficients of \(^\)Lu a NEMA image quality PET phantom with five spheres of different diameters was used. Spheres diameters, volumes, injected and background activities are listed in Table 5. Measurement settings were the same as those used for the CF evaluation and are listed in Table 2 and Table 3. The number of projections was 64, each of 30 s duration.

Table 5 The five spheres of the NEMA PET phantom with their respective activity and background are shown. Each of the value reported in this Table is the mean value of five different measurements, with an associated error of less than 1%

The ratio between activity concentration in the background and the activity concentration in the spheres was not constant but ranged from 0.2% to 7% starting from the smallest sphere to the largest. Each value reported in Table 5 is the mean of five different measurements, with an associated standard deviation of less than 1%.

CT data has been used for delineation of the volume of interest (VOI) of each sphere in SPECT studies.

The curve fitting the RC values was performed using the Igor software [Igor Pro, version 4.01, Wavemetrics, Inc, 1988-2000, Oregon, USA]. RC data errors were evaluated by taking into account the Poisson distribution of the SPECT acquired counts and the errors in activity measurement, volume and time interval estimation.

Monte Carlo simulation for \(^\)Tc and \(^\)Lu

Monte Carlo simulations of the experiments performed with \(^\)Tc and \(^\)Lu have been performed via SIMIND v6.1. The Monte Carlo simulation code SIMIND is a photon-tracking program developed by Professor Michael Ljungberg (Medical Radiation Physics, Department of Clinical Sciences, Lund, Lund University, Sweden). SIMIND models a standard clinical SPECT camera, then simulates projection images from user-defined attenuation maps and activity distributions.

Both \(^\)Tc and \(^\)Lu were studied via SIMIND: the main parameters set for the Monte Carlo simulations are listed in Table 6.

Table 6 Main parameters inserted in SIMIND’s CHANGE program for horizontal cylinder uniformly filled with radionuclides activity

In SIMIND, we simulated all the planar and tomographic acquisitions reported previously, and we added the simulation of the system spatial resolution for distances from the source to collimator front-end ranging from 5 cm to 40 cm in 5 cm steps both for \(^\)Tc and \(^\)Lu. These curves are useful to estimate the compensation for system spatial resolution in the reconstruction process. To obtain the three-dimensional studies, the projected images produced via SIMIND were reconstructed using CASToR (Customizable and Advanced Software for Tomographic Reconstruction [17]), an open-source toolkit for tomographic reconstruction for both emission and transmission exams. CASToR applies the OSEM-3D iterative reconstruction technique [18], 10 iterations and 8 subsets were chosen. Attenuation correction was performed using the SIMIND generated density maps, including window-based scatter correction, while the CDR was modelled as a stationary 2D isotropic Gaussian.

留言 (0)

沒有登入
gif