Patient study design
This study was approved by our institutional review board, and the need for written informed consent was waived. All data were and managed in compliance with the Health Insurance Portability and Accountability Act. A total of 111 [I-123] mIBG SPECT/CT examinations were reconstructed by using SUV SPECT® (HERMES Medical Solutions Inc., Montreal Quebec). The 111 examinations were obtained from the local archival database for the months April 2015 through March 2016. Of the examinations described in this study, the first 42 were acquired using a GE Infinia Hawkeye 4 (hereafter referred to as “Infinia”), April through July 2015, and the remaining 69, using a Siemens Symbia Intevo (hereafter referred to as “Symbia”), August 2015 through March 2016. Image analysis of two patients was extended through July 2016 for inclusion in Figs. 2 and 4 .
Per the institutional standard imaging protocol, each patient received [I-123] mIBG intravenously [370 MBq (10 mCi) per 1.73 m2 body surface area, minimum 74 MBq (2 mCi) and maximum 370 MBq (10 mCi)]. To block [I-123] uptake in the thyroid, a saturated solution of potassium iodide drops (1 drop three times daily) was administered 24 h prior to [I-123] mIBG injection and continued for a total of 3 days. Approximately 24 h post [I-123] mIBG administration, whole-body and spot images were acquired followed by SPECT/CT images. The SPECT field of view was determined based on the technologist’s consultation with the nuclear medicine physicians but generally included the neck, chest, abdomen, and pelvis; imaging of the cranium and extremities was not included in this analysis due to a paucity of data. SPECT acquisition parameters for both scanners included 180-degree configuration for a total of 96 views at 35 s/stop with a 128 × 128 matrix. The Infinia used medium-energy (ME) collimators for both planar and SPECT images, and the Symbia used low-energy high-resolution collimators (LEHR). Directly following the SPECT acquisition, a CT series was acquired for attenuation correction and anatomical co-localization purposes. For the Infinia, the CT acquisition parameters were 120 kV, 59 mAs, 2.6-s rotation time, no zoom, and collimation of 4 × 5 mm. Images were reconstructed using filtered back projection (FBP) with a standard soft tissue reconstruction kernel. The average examination volume-computed tomography dose index (CTDIvol) for the Hawkeye was 13.1 mGy (assuming a 32-cm diameter CTDI phantom). For the Symbia, the CT acquisition parameters were 80 kV, 30 quality ref mAs, 0.6-s rotation time, pitch of 0.8, no zoom, and collimation of 16 × 1.2 mm. Standard images were reconstructed by using FBP with a soft tissue reconstruction kernel (I31s). The average examination CTDIvol for the Symbia was 0.69 mGy (assuming a 32-cm diameter CTDI phantom).
Patient image reconstruction and SUV calibration
To render the Infinia and Symbia acquired SPECT images quantitative, each patient’s examination SPECT and CT image sets were reconstructed using HERMES’ SUV SPECT® software. The SPECT images were reprocessed using 5 iterations and 16 subsets of an ordered subset expectation maximization (OSEM) reconstruction followed by a series of corrections applied to the image raw data to correct for count losses due to attenuation from the patient (using the CT image set), partial volume resolution loss, and photon scattering contamination of the main photon peak. Following the SPECT reconstruction, a conversion factor (units: kBq cts− 1) was applied to convert pixel values (units: cts cm− 3) to represent a value of radioactivity concentration (units: kBq cm− 3). The conversion factor was derived using a Jaszczak cylinder phantom (without any inserts) filled with a known water volume. The water phantom was prepared with an activity of 111 MBq (3 mCi) of [I-123] mIBG calibrated using a Capintec CRC-15R dose calibrator; the dose calibrator accuracy was calculated to be better than 1.5% (as calculated for 8 different radioisotopes including [I-123]). The conversion factor was derived based on 111 MBq (3 mCi) to keep the count rate on the scanner < 20 kcts s− 1, per HERMES instructions. SPECT/CT images were acquired and processed within the HERMES Hybrid Recon domain to produce the conversion factor. A separate conversion factor was calculated for both the Infinia scanner, using ME collimators, and the Symbia scanner, using LEHR collimators. For each patient’s examination, the administered radiopharmaceutical dose (determined by subtracting postinjection activity from preinjection activity in the injection syringe), time of injection, patient’s weight, patient’s height, and scan start date/time were entered into the software.
Phantom study
Quantitative accuracy was measured using a National Electrical manufacturers Association (NEMA) International Electrotechnical Commission (IEC) body phantom with six fillable spherical and one lung insert. The lung insert (measured 50 mm outer dia.) was placed at the center of the phantom and was filled with low density Styrofoam balls (average ρ = 0.3 ± 0.1 g cm− 3; μlung ~ 0.026 cm− 1) and air to mimic lung attenuation. Six spherical inserts (inner diameter 10, 13, 17, 22, 28, and 37 mm) were filled with a mixture of [I-123] mIBG and water.
The phantom was prepared using an ~ 3:1 ratio of water to [I-123] mIBG. A 3000 cm− 3 volume of water was prepared with a calibrated net activity of 106.2 MBq (2.9 mCi) which led to an activity-by-volume of 35.4 kBq cm− 3. Radioactive water was drawn from the 3000 cm3 solution and used to fill the six spherical inserts: 10-mm sphere received 0.5 cm3, 13-mm sphere received 1.0 cm3, 17-mm sphere received 2.8 cm3, 22-mm sphere received 5.8 cm3, 28-mm sphere received 11.7 cm3, and 37 mm received 29 cm3 of radioactive water. The body phantom, with inserts in place, was filled to full capacity (9650 cm3). The phantom preparation-to-scan interval was approximately 2 h. The phantom was scanned using the same acquisition techniques described in section II.A using only the Symbia.
Once acquired, the images from the Symbia were processed using HERMES SUV SPECT® for quantitation. SUV values were measured by placing ROIs within the area of the observed radioactive spheres on the images using the following formulation:
$$ SUV=\frac{A_v}{\frac{A}{w}}, $$
(1)
where Av is the activity-by-volume measured with the ROI, A is the net activity measured by the dose calibrator, and w is the weight of the phantom/patient entered at the scanner console at the time of the study; the weight of the phantom with water was 16 kg and used for the SUV calculation. Units of SUV calculated in the phantom were g cm− 3; note SUV values calculated in tissue (i.e., patients) have no units because the activity-by-volume is measured using an ROI over tissue that has a mass density of 1 g cm− 3, thus, the activity-by-volume is converted to tissue tracer activity by dividing Av by g cm− 3 to yield units of kBq g− 1. Additionally, ROIs were placed throughout the phantom volume to measure activity in the background water. The ROI measurements from the spheres and the background were combined to calculate the contrast recovery coefficient (CRC):
$$ CRC=\frac{\frac{SUV_{\mathrm{sphere}}}{SUV_{\mathrm{bkgd}}}-1}{\frac{A_{\mathrm{sphere}}}{A_{\mathrm{bkgd}}}-1}, $$
(2)
where SUVsphere and SUVbkgd are the measured sphere and average background SUV (g cm− 3) respectively, and Asphere and Abkgd are the known sphere and background activity-by-volume (kBq cm− 3), respectively. The measured SUV values for the spheres and background were used to calculate a predicted sphere activity-by-volume (\( {A}_v^{\mathrm{measured}} \)) using Eq. 1 and corrected for 1-CRC loss of counts due to system resolution limits:
$$ {A}_v^{\mathrm{measured}}= SUV\bullet \frac{A}{w}\bullet \left[1+\left(1- CRC\right)\right]. $$
(3)
The \( {A}_v^{\mathrm{measured}} \) was compared with the actual activity-by-volume calculated for the fillable spheres 35.4 kBq cm− 3 and for the background 11.0 kBq cm− 3.
Study analysis
Each examination was analyzed for interscanner SUV reproducibility for 9 regions of normal [I-123] mIBG tissue uptake, namely left/right parotid glands, left/right submandibular glands, left ventricle of the heart, liver, left/right adrenal glands, and the bladder. Additionally, neoplastic tissue, when present in the examination data, was quantified by using SUVs and explored over time. Regions of interest were drawn over the SPECT images and overlaid on the CT for anatomic confirmation. An analysis of the intrapatient mean SUV variability was calculated by measuring normal liver uptake for patients scanned on both scanners. Intrapatient variability was measured for 28 patients with two or more studies at different time points: two studies (n = 6), three studies (n = 14), four studies (n = 5), five studies (n = 1), six studies (n = 1), and ten studies (n = 1).
Descriptive statistics were used to describe the data. A two-tailed Mann-Whitney U-test (significance level equal to 0.05) was used to compare SUVs measured in the normal tissues. Statistical calculations were performed using PRISM (GraphPad, La Jolla, CA).
Patient demographics
A total of 111 examinations obtained over 12 months (April 2015 through March 2016) were analyzed for this study, with 43 unique patients imaged (22 boys, 60 studies; 21 girls, 51 studies; number of studies per patient, 1–10). The patients’ median age was 3.0 years (range 0.8 to 17 years); median weight, 16.1 kg (range 9.4 to 90.9); median height, 98.1 cm (range 72 to 174 cm); median BSA, 0.7 m2 (range 0.4 to 2.1 m2); and median radiopharmaceutical dose of [I-123] mIBG, 162.8 MBq (range 107.3 to 429.2 MBq) [4.4 mCi (range 2.9 to 11.6 mCi)]. Three patients received radiopharmaceutical doses above 370 MBq (10 mCi) after consultation with the nuclear medicine physician because of their age, weight, and BSA concentration at the time of examination.