If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Medical Physics, University of Trieste, Via Tiepolo 11, Trieste, ItalyAbdus Salam International Centre for Theoretical Physics (ICTP), Strada Costiera 11, Trieste, Italy
SUVmax and SULmax variation increases with PET scan duration.
•
SUVmax values are significantly higher in high BMI patients.
•
SULmax values are independent of patient’s BMI.
•
Noise is higher in high BMI patients and decreases when increasing acquisition time.
•
Use of SUL is recommended for improve lesion quantification accuracy.
Abstract
Purpose
To study how the quantitative parameters of 18F-FDG PET imaging change with the emission scan duration (ESD) and the body-mass-index (BMI) in phantom and patients on a time-of-flight (TOF)-PET/CT system.
Methods
The image-quality phantom with (b-NEMA-IQ, BMI = 29.2 kg/m2) and without (NEMA-IEC, BMI = 21.4 kg/m2) a ‘belt’ of water-bags was filled with 18F-FDG activities to obtain nominal standardized uptake values (SUV) of 19, 8 and 5.
Patients with BMI ≤ 25 kg/m2 (L-BMI) and BMI > 25 kg/m2 (H-BMI) were enrolled in this study. Phantom and patients underwent list-mode PET acquisition at 120 s/bed-position. Images reconstructed with clinical protocol and different ESD (120, 90, 75, 60, 45, 30 s) were analysed for comparison of maximum SUV (SUVmax), maximum standardized uptake value lean-body-mass corrected (SULmax) and noise.
Results
79 oncologic patients (45 L-BMI, 44 H-BMI) were analysed. From 90 s to 30 s, an increasing variation of SUVmax and SULmax with respect to the reference 120 s time was observed, from 18% to 60% and from 16% to 37% for phantom and patients, respectively. SUVmax values were significantly higher (+50%) in b-NEMA-IQ than NEMA-IQ phantom and in H-BMI (+33%) than L-BMI patients. No significant difference was found in SULmax for the two BMI categories in both phantom and patients.
CV values decreased when increasing ESD, being higher in H-BMI patients (0.13–0.25) and b-NEMA-IQ phantom (0.15–0.28) than in L-BMI patients (0.11–0.21) and NEMA-IQ phantom (0.11–0.20).
Conclusions
Reduction of ESD may severely impact on the variations of SUVmax and SULmax in 18F-FDG PET/CT imaging. This study confirms recommendations of using SUL for lesion uptake quantification, being unaffected by BMI variation.
18F-FDG PET/CT imaging is continuously increasing its importance for diagnosis, staging, radiotherapy planning, prognosis and treatment response for oncological patients.
The great advances in scintillator detectors, new photosensors and electronics coupled with time-of flight technology [
], determined a significant improvement in the accuracy of the quantitative information, enhancing lesion detectability and increasing the overall 18F-FDG PET/CT image quality.
The principal improvements that can be derived from the abovementioned technological advances are the reduction of the administered 18F-FDG activity and the emission scan duration [
Can the BMI-based dose regimen be used to reduce injection activity and to obtain a constant image quality in oncological patients by 18F-FDG total-body PET/CT imaging?.
The reduction of the administered activity aims to both reduce patient and staff radiation exposure as well as the economic burden. Furthermore, decreasing the acquisition time can ease the management of non-compliant patients, reducing the risk of patient movement.
Recent papers have been published on the possibility of a further decrease in the emission scan duration offered by digital technology [
From a PMT-based to a SiPM-based PET system: a study to define matched acquisition/reconstruction parameters and NEMA performance of the Biograph Vision 450.
A novel figure of merit to investigate 68Ga PET/CT image quality based on patient weight and lesion size using Q.Clear reconstruction algorithm: A phantom study.
From a PMT-based to a SiPM-based PET system: a study to define matched acquisition/reconstruction parameters and NEMA performance of the Biograph Vision 450.
] compared to the same generation of non-digital PET systems demonstrated to guarantee lesion detectability without compromising image quality in 18F-FDG PET/CT whole body oncologic imaging.
If exploring the fastest acquisition times can be attractive for the abovementioned reasons, particular attention should be paid when considering the quantitative issue in lesions evaluation. A few papers have outlined a significant increase in the quantitative parameters and noise of 18F-FDG PET/CT whole body oncologic imaging when the scan duration is excessively reduced, showing that this effect is particularly evident when the optimization process of the reconstruction parameters was not been carefully considered [
The further dependence of 18F-FDG PET/CT imaging quantitation on the patient’s body habitus represents a further issue that should be considered, as the presence of fat is known to impact on the standardized uptake value (SUV) [
Standardized uptake values of normal tissues at PET with 2-[Fluorine-18]-Fluoro-2-deoxy-D-glucose: variations with body weight and a method for correction.
This evidence claims the need for an in-depth analysis of the overall image quality with a special focus on the quantitative parameters used in PET clinical imaging when planning a reduction in the acquisition time and tailoring of the acquisition protocols on the body habitus for the clinical routine.
The aim of this work was to study how the quantitative parameters used in clinical imaging and related to 18F-FDG PET image quality change with the emission scan duration (ESD) and the patient’s body habitus. The study was performed in controlled conditions and in a retrospective multicentre study on a group of oncologic patients acquired on a PET/CT system with time of flight (TOF) technology.
2. Materials and methods
2.1 PET/CT systems
The study was performed on the Ingenuity TF 64 (Philips Healthcare, Cleveland, OH, USA) TOF-PET/CT systems at three PET centres in Italy: Istituto Oncologico Veneto (Padova, Italy, N.1), University Hospital Maggiore della Carità (Novara, Italy, N.2) and AO Ordine Mauriziano Hospital (Torino, Italy, N.3). The performance characteristics of this TOF-PET/CT system according to NEMA NU 2012 standard have already been described in detail [
], showing that these three systems demonstrated to have identical image quality performance. This retrospective multicentre observational study was approved by the medical ethics committee.
2.2 Phantom
The International Electrotechnical Commission (IEC) 61675-1 emission phantom (National Electrical Manufacturers Association image quality phantom, NEMA-IQ phantom) with FDG solution was used. To reproduce the clinical situation of activity outside the system field of view (FOV), the scatter phantom (Data Spectrum Corporation) was placed close at the end of the IEC phantom, strictly following NEMA NU-2 recommendations [
]. The NEMA-IQ and the b-NEMA-IQ phantoms body mass index (BMI) values were of 21.4 (normal BMI patient) and 29.2 (high-overweight BMI patient) kg/m2, respectively [
The spheres and the main compartment of the NEMA-IQ phantom were filled with 18F activity concentrations in order to provide three target-to-background ratios in three different experimental sessions or, equivalently, three theoretical standardized uptake values (SUV) in the NEMA-IQ phantom spheres. The theoretical SUV values obtained were 19, 8 and 5 respectively. The NEMA-IQ phantom and the NEMA-IQ phantom wrapped with the belt (b-NEMA-IQ) were sequentially acquired in list mode with an ESD of 120 s (one bed position) at different activity concentrations in the main compartment in the range 1.3–5.1 kBq/ml. The clinical protocol for whole-body examinations was used to reconstruct all the acquired images (99 equivalent iterations, TOF kernel width of 14.1 cm, full width at half maximum of the Gaussian filter equal to 4 mm, relaxation parameter equal to 1.0) on a 144 × 144 frame (4 mm isotropic voxel). In order to analyse the effect of acquisition times on quantitative parameters, different ESD were obtained by cutting the list-mode file after ESD seconds, namely 30, 45, 60, 75, 90 and 120 s.
The setup configuration, phantom preparation, image acquisition and reconstruction have already been described in detail [
The phantom session was performed in PET Centre N.2.
2.3 Patients’ population
Patients referred for oncologic clinical PET/CT were retrospectively selected in the three PET Centres, with the following inclusion criteria:
•
Patients who underwent a 18F-FDG PET/CT scan for clinical reasons in the three years prior to the study beginning;
•
Patients with a glucose level inferior to 198 mg/dL before 18F-FDG administration;
•
Patients acquired with the Ingenuity TF PET/CT system with an acquisition time of 120 s for each FOV;
•
Patients over 18 years of age at the time of the acquisition;
•
Patients with different BMI, spanning from normal (L) to overweight category (H), by using a threshold of 25 kg/m2;
•
Patients without FDG positive liver lesions.
The three PET Centres followed the same administration and acquisition scheme: after a fasting period of 4–6 h, patients received a weight-based bolus injection of 18F-FDG activity (3 MBq/kg) via intravenous infusion, in accordance with European Association of Nuclear Medicine guidelines for tumor imaging [
]. Approximately 60 min after injection (66 ± 9 min), patients underwent a list-mode PET/CT acquisition protocol on the TOF Ingenuity-TF 64 system. A standard low-dose CT scan (X-ray tube current modulation with a DoseRight index of 9/16, 50/80 reference mAs, a tube voltage of 120 kV, a spiral pitch factor of 0.83, collimation of 40 mm, FOV of 60 cm) was acquired from the vertex to the mid-thigh and used for attenuation correction, followed by an emission PET scan acquired in list mode at 120 s/bed position, during normal breathing.
Subsequently, PET list-mode data were reprocessed to produce additional imaging datasets at ESD = 30, 45, 60, 75 and 90 s/bed position. The protocol used to reconstruct all the PET clinical images was the same as the one used for phantom reconstruction (see Phantom section).
2.4 Image analysis
All reconstructed PET images, both phantom and patients, were analysed using LifeX v7.0.9 software [
The spheres/lesions that resulted visible on the phantom/patient reconstructed dataset were delineated by using the 3D iso-contour at 41% of the maximum pixel value in the volume of interest (VOI), according to EANM guidelines [
]. For each VOI, the maximum Standardized Uptake Value (SUVmax) and the maximum Standardized Uptake Value corrected for the lean body mass (SULmax) were evaluated as follows:
where:
-
Cmax-VOI is the maximum activity concentration (kBq/ml) measured in the VOI
-
Aadm is the total activity (MBq) present in the phantom/patient at the time of acquisition
-
W is the mass (kg) of the phantom/patient
-
LBM is the patient’s lean body mass (kg) calculated by using the formulas of Janmahasatian [
where LBMM and LBMF are the LBM for males and females, respectively.
LBM values for NEMA-IQ and b-NEMA-IQ phantoms were considered as the phantom mass (10 kg) without the belt, while the belt mass (5.5 kg) was considered as fat mass [
For an assessment of image noise, a spherical VOI of diameter 3 cm was placed in the main compartment of the NEMA-IQ and b-NEMA-IQ phantoms and on the right upper lobe of the patient’s liver. Image noise was defined as the coefficient of variation (CV):
where Cbkg and SDbkg are respectively the mean and the standard deviation of the activity concentration in the ‘background’ (liver for the patient, main compartment for the phantom) [
Each parameter was calculated for all the reconstructions.
Phantom acquisitions were performed in order to test the effect of size and ESD on quantitative parameters in a controlled situation. In case of presence of significant differences between phantom and patients’ data, the analysis on phantom parameters was conducted by selecting values that fall within the range of clinical quantitative parameters collected to make the comparison more reliable.
2.5 Statistical analysis
The normality test for all the parameters investigated in the study was performed by means of Kolmogorov-Smirnov test.
Data were presented with median and range or average and standard deviation in case of non-normal and normal distribution respectively. Absolute frequency and percentage were used for categorical variables.
In order to examine the stability of image parameters as a function of ESD values, SUVmax, SULmax and CV were represented by means of box plots and Bland-Altman graphs considering 120 s as the reference dataset. For the Bland-Altman analysis, the ratios between the quantitative parameter at each acquisition time with respect to the reference time (SUVmax_i/SUVmax_120, rSUVi and SULmax_i/SULmax_120, rSULi) were represented on a graph as a function of the mean of the two measures (mean(SUVmax_i;SUVmax_120), mSUVi; mean(SULmax_i; SULmax_120), mSULi); the limits of agreement were evaluated.
The pairwise paired Wilcoxon signed-rank test and the Mahn-Whitney test were used to highlight statistically significant differences for each metrics at different time points and BMI categories respectively. A p-value lower than 0.05 was considered statistically significant.
Analyses were performed with the software RStudio, version 4.0.3 (RStudio, Inc., Boston, MA) [
R core team R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
The Bland-Altman analysis showed that the variation of the quantitative parameters rSUVi and rSULi increased when the acquisition time decreases. In particular, the limits of agreement indicate variations from 18% to 60% when decreasing ESD from 90 s to 30 s (Table 1 and Fig. 1).
Table 1Range of mSUVi and mSULi and limits of agreement evaluated for the 30, 45, 60, 75 and 90 s with respect to the 120 s reference ESD in the phantom experiment.
Fig. 1Bland-Altman plots for rSUVi (left column) and rSULi (right column) for 90 s (a and b), 60 s (c and d) and 30 s (e and f) comparisons with respect to the 120 s reference ESD in the phantom experiment. Data from each nominal SUV value are represented in blue (SUV 5), yellow (SUV 8) and green (SUV 19). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
It is worth to note that mSULi ranges between 2.9 and 20.5, which is the nominal SUV range realized in the experiment (5–19) (Fig. 1d, 1e, 1f). On the contrary, the average of rSUVi is definitely greater (range: 3.0–31.3), than the expected (range: 5–19), showing a general overestimation of SUVmax (Fig. 1a, 1b, 1c).
Median and ranges of SUVmax, SULmax and CV values for the phantom analysis at different ESDs and for the two BMI categories are reported in Supplementary Table 1.
The results of the overall comparison between NEMA-IQ and b-NEMA-IQ phantom for the quantitative parameters evaluated for the outlined spheres are reported in Fig. 2 in terms of box plots.
Fig. 2SUVmax (a) and SULmax (b) values evaluated for the outlined spheres in the b-NEMA-IQ and NEMA-IQ phantoms. SUVmax (c) and SULmax (d) values evaluated for the outlined spheres for b-NEMA-IQ and NEMA-IQ phantoms and for each ESD.
Pooling together the quantitative values measured for all the outlined spheres for each activity concentration and ESDs, a different behaviour can be observed for SUVmax and SULmax values for the two phantom configurations. SUVmax values were significantly higher in the b-NEMA-IQ with respect to the NEMA-IQ phantom (p < 0.05, Fig. 2a), with an overall increase of 50% in former phantom. In addition, deviations between the measured SUVmax values and the theoretical ones of 40.3% (−56.6 to 158.5) and −12.5% (−66.8 to 82.5) were evaluated for b-NEMA-IQ and NEMA-IQ phantoms, respectively.
On the contrary, SULmax values were comparable for the two phantom configurations (p > 0.05, Fig. 2b).
The difference in SUVmax values between the phantom configurations becomes more significant (p < 0.001) when data are splitted for each acquisition times, as can be observed in Fig. 2c.
No difference was detected in SULmax values when splitting data into different acquisition times, as shown in Fig. 2d (p > 0.05).
The imaging dataset used for the CV analysis was the one corresponding to the background activity concentration of 4.5 ± 0.4 kBq/ml ([3.7 ÷ 5.1] range), being the closest to the liver clinical activity (4.6 ± 1.3 kBq/ml, [1.6 ÷ 12.6] see patients’ results paragraph).
The noise evaluated as the CV of the signal in the 3D VOI on the phantom background region shows a decreasing trend for the two phantom configurations (Fig. 3a). Moreover, the CV showed to be significantly higher in the b-NEMA-IQ when compared to the NEMA-IQ phantom (p < 0.001), as shown in Fig. 3b.
Fig. 3CV values for the two phantom configurations: mean and SD for the different ESD (a); box plot for phantom categories (b).
A total of 79 oncologic patients acquired between February 2019 and August 2020 were enrolled in this study. Patients’ demographic and clinical characteristics are reported in Table 2.
Table 2Characteristics of the patients enrolled in the study.
All the patients received a single weight-based bolus of 18F-FDG injected activity (206 ± 42 MBq, range: 126–314) via intravenous infusion.
Regarding BMI classification, 34 patients (43%) showed a BMI lower than 25 kg/m2, while 45 (57%) had a BMI higher than 25 kg/m2.
A total of 140 lesions were detected and analysed. Lesions were located in the head and neck region (23, 16.4%), thorax (68, 48.6%) and abdomen/pelvis region (49, 35.0%) respectively. A number between one to nine lesions were identified for each patient. Twenty-one patients were disease free.
Pooling together the quantitative values for all the outlined lesions in the patients imaging dataset regardless of BMI, the Bland-Altman analysis showed a reducing variation in the quantitative parameters with respect to the reference time (120 s) when the acquisition time is increased (Table 3). Bland-Altman plots are reported in Fig. 4 for rSUVi and rSULi respectively, for 30 s (a and d), 60 s (b and e) and 90 s (c and f) comparisons with respect to the reference acquisition time of 120 s.
Table 3Range of mSUVi and mSULi and limits of agreement evaluated for the lesions outlined in the PET whole body images for the 30, 45, 60, 75 and 90 s with respect to the 120 s reference ESD.
Fig. 4Bland-Altman plots for rSUVi (left column) and rSULi (right column) for 90 s (a and b), 60 s (c and d) and 30 s (e and f) comparisons with respect to the 120 s reference ESD for the lesions outlined in PET whole body images.
Both metrics showed an increased spread when reducing the acquisition time: for acquisition times of less than 60 s, statistically significant differences with respect to the reference 120 s were found for both SUVmax and SULmax (Table 4).
Table 4P-value for the paired Wilcoxon signed-rank test for comparing different ESD respect to 120 s.
The trend of both metrics as a function of different ESD values confirmed what was observed in the phantom experiment.
Median and ranges of SUVmax, SULmax and CV values for the patient’s analysis at different ESD and for the two categories of BMI are reported in Supplementary Table 2.
Box plots are obtained for the comparison between H-BMI and L-BMI patients when pooling together values of the quantitative parameters for all the lesions outlined in the imaging dataset across all the acquisition times (Fig. 5a and 5b).
Fig. 5SUVmax (a) and SULmax (b) values evaluated for the lesions outlined in the H-BMI and L-BMI patients, irrespective of the acquisition time. SUVmax (c) and SULmax (d) values evaluated for the outlined lesions for H-BMI and L-BMI patients and for each acquisition time.
As already observed for the phantom experiment, the results obtained indicate a different behaviour for SUVmax and SULmax. Regardless of the acquisition time, SUVmax values were higher for the lesions outlined in the H-BMI patients with respect to the L-BMI patients, and an overall increase of 33% is observed (Fig. 5a). On the contrary, the overall increase in SULmax values for the lesions outlined in the H-BMI patients is only marginal (Fig. 5b).
When considering the acquisition time, SUVmax values for the lesions outlined in H-BMI patients were significantly higher than those in L-BMI patients, for each acquisition time (p < 0.001) (Fig. 5c and Table 5). No significant difference was detected in SULmax values between the two BMI categories when considering different acquisition times (p > 0.05, Fig. 5d and Table 5).
Table 5P-value for the Mann Whitney signed rank test for SUVmax and SULmax values when comparing H-BMI and L-BMI group at different ESD.
The noise evaluated as the CV of the activity concentration in the 3D VOI on the tumour-free liver shows a decreasing trend with increasing the acquisition time with a statistically significant difference (p < 0.05 for all the ESDs compared to 120 sec). Furthermore, the CV showed to be significantly higher in H-BMI patients when compared to L-BMI ones for all ESD values, as shown in Fig. 6.
Fig. 6CV values for H-BMI and L-BMI patients for the explored acquisition times.
Recent technological advances in PET/CT lead to important improvements in the reduction of both the administered activity and the emission scan duration [
From a PMT-based to a SiPM-based PET system: a study to define matched acquisition/reconstruction parameters and NEMA performance of the Biograph Vision 450.
A novel figure of merit to investigate 68Ga PET/CT image quality based on patient weight and lesion size using Q.Clear reconstruction algorithm: A phantom study.
] in 18FDG whole body oncologic imaging. Recently published papers on this topic show the possibility of reducing the acquisition time by a factor from two [
From a PMT-based to a SiPM-based PET system: a study to define matched acquisition/reconstruction parameters and NEMA performance of the Biograph Vision 450.
A novel figure of merit to investigate 68Ga PET/CT image quality based on patient weight and lesion size using Q.Clear reconstruction algorithm: A phantom study.
] with digital systems with respect to the same generation of non-digital PET systems, guaranteeing lesion detectability without compromising image quality. However, the attractive possibility of fastest acquisition times should be carefully considered; few papers outlined a significant overestimation of the quantitative parameters as well as in the noise of 18F-FDG PET/CT whole body oncologic imaging when the scan duration is excessively reduced [
This paper was aimed to investigate how quantitative parameters used in clinical imaging and physical figures of merit related to 18F-FDG PET image quality change with the emission scan duration and the patient’s body habitus, both in a phantom experiment and in a multicentre retrospective study on a group of oncologic patients acquired on a PET/CT system with TOF technology. Standardized uptake values, standardized uptake values normalized to lean body mass of detected lesions and the coefficient of variation over a uniform uptake region were evaluated to study the dependence of quantitation and noise variation on the acquisition time and patient’s habitus.
The main result of this study is the evidence of an increase in the variation of both SUVmax and SULmax values when the acquisition time is reduced from the reference 120 s acquisition time, both in the phantom experiment and in the clinical study. The limits of agreement in Bland-Altman analysis show an increase from 18% to 60% and from 16% to 37% when comparing 90 and 30 s to 120 s respectively for phantom (Table 1 and Fig. 1) and patients (Table 3 and Fig. 4). These results indicate a considerable erroneous estimate of the quantitative parameter at low ESD values. Similar results were obtained by van Sluis [
] for the spread of SUVmax values of the lesions outlined in 18FDG PET/CT oncologic imaging reconstructed with the clinical protocol when the acquisition time was reduced from the reference time of 180 s: the 30 s and 10 s acquisition time dataset show a spread in lesion SUVmax values up to 14% and 27%, respectively. In addition, no increase of SUVmax values was observed for the EARL compliant reconstructed dataset.
] studied the variation in the mean lesion to background ratio in a group of cancer patients acquired on a digital TOF PET/CT system by reducing the standard 90 s acquisition time to 30 s and 20 s and reported an average decrease of 23%.
It may be considered to reduce the acquisition time until the overall deviation of SUVmax from the actual value is less than 30%, as this value is proposed as a significant variation in tumour activity by the framework for PET Response Criteria in Solid Tumours [
]. Based on the abovementioned criterion, the results of this study indicate that a reduction of the acquisition time is feasible up to 45 s.
In the present study, no increase in SUVmax and SULmax values was observed in either phantom or patients when decreasing ESD values in the range explored. Van Sluis [
] reported significant differences in lesion SUVmax between 180 s images and the 30- and 10-second ones only when using the clinical protocol, while no difference was reported when using the EARL compliant reconstruction protocol. The only difference between the EARL compliant reconstruction protocol suggested for the Philips Ingenuity-TF system and the clinical reconstruction protocol applied in this study is the full width of the Gaussian filter which is set to 5 mm in the former protocol and 4 mm in the latter one. It is reasonable to believe that the clinical reconstruction protocol used is almost optimized.
Another important result obtained in this study is the significant difference found in the SUVmax values of the lesions in high BMI with respect to low BMI patients. It has been known for decades [
] that 18F-FDG uptake in fat is very low and the SUV in fat-free tissues in larger patients may be overestimated compared to the thin patient. This condition supports the statement that corrected SUV for the lean body mass is a weight-independent index for 18F-FDG uptake [
Standardized uptake values of normal tissues at PET with 2-[Fluorine-18]-Fluoro-2-deoxy-D-glucose: variations with body weight and a method for correction.
] by analysing the effect of obesity on SUV and SUL values in a group of patients with normal and high BMI, reporting SUVmax values in the blood pool and liver of high BMI patients from 36 to 41% higher than in low BMI patients. As 18F-FDG reaches tumours via the bloodstream, it is reasonable to expect that tumour SUV values in larger patients may be higher than in thin patients. Our results are in close agreement with the findings of Sarikaya [
Standardized uptake values of normal tissues at PET with 2-[Fluorine-18]-Fluoro-2-deoxy-D-glucose: variations with body weight and a method for correction.
], having found an overall increase of 33% in lesion SUVmax in high BMI patients with respect to low BMI patients (Fig. 5). Furthermore, both studies confirmed that SULmax values were not affected by the patient’s BMI.
The additional value of our study is provided by the phantom experiment, that evaluated the SUVmax data for the spheres mimicking lesions in the clinical setting and confirmed in controlled conditions the results obtained for the lesions in the 18F-FDG whole body imaging: SUVmax values obtained for the spheres in the b-NEMA-IEC phantom showed an overall increase of about 50% with respect to those of the NEMA-IEC one. Moreover, the phantom study in which the theoretical SUV values of the spheres are known, allowed to evaluate in detail the SUVmax overestimation at low, medium and high sphere-to-background ratio values, +56%, +79% and +39%, respectively. As observed in the clinical setting, also SULmax values demonstrated to be unaffected by the phantom configuration (Fig. 1, Fig. 2).
These findings once again bring the attention to the limitation of SUV, nevertheless this parameter is still widely used for PET image quantitation. This consideration is particularly important when comparing PET studies of a patient whose weight may have significantly changed during follow up and indicates SUL as a better alternative to SUV, as already recommended by international guidelines on tumour imaging [
]. This consideration becomes particularly relevant when considering a reduction in the acquisition time as part of the optimization procedure in clinical trials.
The noise evaluated in PET clinical studies showed a decreasing trend with increasing the acquisition time and generally higher values in H-BMI (0.13–0.25) than in L-BMI patients (0.11–0.21). The same conclusion was found also in the phantom experiment when comparing the noise evaluated at different acquisition times and for the two phantom configurations, i.e (0.15–0.28) and (0.11–0.20) for the b-NEMA-IQ and the NEMA-IQ phantoms, respectively. Usually, a CV inferior to 15% is considered an acceptable noise level for clinical image interpretation as suggested in the EARL procedure [
], enabling 18F-FDG PET image comparison for quality assessments. Our data show that on average this requirement is met when the acquisition time is greater than 60 s for L-BMI patients, while only the 120 s acquisition time meets this requirement for H-BMI patients. The results of the phantom experiment are in agreement with the clinical data. A similar analysis on the NEMA-IEC phantom was performed by Gnesin [
] who studied the CV in PET images as a function of the acquisition time, administered activity and reconstruction parameters for several digital and non-digital PET/CT systems. When considering the most similar configuration (acquisition time 120″, pixel size 2.73 mm, FWHM = 5 mm, 3.5 MBq/kg) to the acquisition/reconstruction setup used in the present study (acquisition time 120″, pixel size 4 mm, FWHM 4 mm, 3 MBq/kg), a coefficient of variation of about 11% is observed, very similar to the median values of 11% and 10% found for the NEMA-IEC and the L-BMI patients.
] studied the signal-to-noise ratio (i.e. the inverse of the CV) on the liver of 18F-FDG whole body imaging of oncological patients acquired on a total-body PET/CT system as a function of BMI. The authors reported values of 16.0 and 18.1 for 120″ acquisition time on overweight and normal BMI patients. When translating the CV into signal-to-noise ratio, we found values of 9.1–10.0 for normal BMI patients and NEMA-IEC phantom, and 7.7–6.7 for overweight patients and b-NEMA-IEC phantoms, respectively. Our values are definitely lower than the corresponding ones reported by Xiao, but the disagreement can be explained when considering that the 18F-FDG administration scheme was 3.7 MBq/kg and the PET/CT system used was the whole-body uEXPLORER PET/CT that is known to have a sensitivity about 40-fold of that for a conventional PET/CT system: these two factors act in a synergic way to increase the signal-to-noise ratio value.
A few limitations of the study should be acknowledged.
Only definitely visible lesions in patient whole-body imaging, were considered in this study. A future development could include both phantom experiment and patient imaging with spheres/lesions characterized by low SUV values (i.e. <5) that would explore the behaviour of SUV/SUL variation in the region of borderline visibility.
The study considered SUVmax and SULmax metrics, as suggested by EANM guidelines [
] and being the most common figures of merit used by the nuclear medicine community. SUVpeak, total lesion glycolysis (TLG) and the metabolic tumour volume (MTV) should be enquired as well, the last ones of increasing interest for therapy response monitoring and prognostic assessment [
Finally, the number of patients should be increased to confirm these findings.
5. Conclusion
Reduction of the acquisition time may severely impact on the variations of SUVmax and SULmax values of lesions outlined in 18FDG PET/CT imaging. Moreover, SULmax values demonstrated to be unaffected by BMI variation; this finding suggests the use of SULmax rather than SUVmax for the quantification of lesion’s uptake.
SUVmax values are affected also by the BMI of the patient, being higher in high BMI patients.
This evidence claims the need of a thorough analysis of the overall image quality with a special focus on the quantitative parameters used in PET clinical imaging when planning a reduction in the acquisition time in the clinical routine. This issue must be carefully evaluated against the differences that patients’ BMI can introduce in quantitative parameters, especially when using the standard SUV.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Can the BMI-based dose regimen be used to reduce injection activity and to obtain a constant image quality in oncological patients by 18F-FDG total-body PET/CT imaging?.
From a PMT-based to a SiPM-based PET system: a study to define matched acquisition/reconstruction parameters and NEMA performance of the Biograph Vision 450.
A novel figure of merit to investigate 68Ga PET/CT image quality based on patient weight and lesion size using Q.Clear reconstruction algorithm: A phantom study.
Standardized uptake values of normal tissues at PET with 2-[Fluorine-18]-Fluoro-2-deoxy-D-glucose: variations with body weight and a method for correction.
R core team R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
00061-3&id=gr1.jpg){kind=link}
00061-3&id=gr2.jpg){kind=link}
00061-3&id=gr3.jpg){kind=link}
00061-3&id=gr4.jpg){kind=link}
00061-3&id=gr5.jpg){kind=link}
00061-3&id=gr6.jpg){kind=link}