ABSTRACT
Objective:To explore the influence of time of flight (TOF) and point spread function (PSF) on PET/CT image quality and standardized uptake value (SUV) based on a phantom study.Methods:PET/CT imaging were performed using Jaszczak phantom and International Electrotechnical Commission (IEC) body phantom respectively, and conventional clinical imaging protocol (3 min/bed) was used for three consecutive imaging. The ordered subsets expectation maximization (OSEM), OSEM+ TOF, OSEM+ PSF and OSEM+ TOF+ PSF were used to reconstruct the raw data respectively. The differences of image resolution, image uniformity, contrast, signal to noise ratio (SNR) and SUV among different reconstruction algorithms were compared by using one-way analysis of variance and the least significant difference (LSD) t test. The correlations between SNR, mean SUV (SUV mean), maximum SUV (SUV max) and the sphere diameter were analyzed by Pearson correlation analysis. Results:The minimum image resolution for the hot column was 6.4 mm by OSEM and OSEM+ PSF reconstruction, while it was 4.8 mm by OSEM+ TOF and OSEM+ TOF+ PSF reconstruction. The contrast was significantly better by OSEM+ TOF ((78.56±1.21)%) and OSEM+ TOF+ PSF ((78.85±1.17)%) reconstruction than that by OSEM reconstruction ((73.44±1.47)%; F=61.068, t values: 9.503, 10.018, both P<0.001). The maximum and minimum nonuniformity percentages were significantly better for images reconstructed with OSEM+ PSF than those reconstructed with OSEM ( F values: 10.286, 27.630, t values: -2.599, 7.698, both P<0.05). In IEC phantom imaging, the SNR and SUV max of hot sphere were higher by OSEM+ PSF and OSEM+ TOF+ PSF reconstruction than those by OSEM reconstruction (SNR: (98.99±34.76)%, (98.29±28.66)%, (73.64±26.05)%; F=5.712, t values: 3.209, 3.412, both P<0.05; SUV max: 8.96±2.72, 9.28±2.17, 8.01±2.21; F=3.931, t values: 2.154, 2.863, both P<0.05), but there was no significant difference between OSEM+ TOF and OSEM reconstruction ( t values: 0.297, 0.272, both P>0.05). Among the four reconstruction methods, the SNR and SUV mean increased with the increase of the sphere diameter ( r values: 0.913-0.963, all P<0.05), but SUV max in images reconstructed with OSEM+ PSF and OSEM+ TOF+ PSF had no significant correlations with the sphere diameter ( r values: 0.496, 0.614, both P>0.05). Conclusions:Under specific acquisition and reconstruction conditions, TOF mainly improves image resolution and contrast, while PSF mainly improves image uniformity and SNR. The combination of the two method can obtain better image quality and significantly improve the SUV of hot lesions.
ABSTRACT
Objective To explore the quantitative determination of standardized uptake value(SUV)of lumbar vertebra by 99Tcm-methylene diphosphonate(MDP)bone imaging with SPECT/CT.Methods A total of 90 patients(16 males,74 females,age:(60.0±11.2)years)who underwent 99Tcm-MDP bone imaging and local lumbar SPECT/CT imaging from May 2017 to November 2017 were retrospectively analyzed.CT attenuation correction,scatter correction and 3-dimensional ordered subset expectation maximization(3D-OSEM)reconstruction were used.The radioactive counts of lumbar vertebra LL-L4 in patients with high MDP-uptake(including degenerative changes and bone metastases)and patients with non-high MDP-uptake(including osteoporosis and non-osteoporosis)were measured.Basing on that,the radioactivity of unit volume in lumbar and lesion was calculated and converted to SUV.The difference of SUV for lumbar and lesions from different types of patients was analyzed,and the correlation between lumbar SUV and varies of parameters were also analyzed.Data were analyzed by two-sample t test and Pearson correlation analysis.Results The mean SUV(SUVmean)and maximum SUV(SUVmax)of L1-L4 in 54 patients with non-high MDP-uptake were 4.64±1.03 and 7.09±1.61 respectively.They were positively correlated with bone mineral content,CT value and body weight(r values:0.433-0.700,all P<0.05),and negatively correlated with age(r values;-0.514 and -0.456,both P<0.001).The SUVmean and SUVmax of patients with osteoporosis were 3.81±0.76 and 5.86±1.34 respectively,which were significantly lower than those of non-osteoporosis(5.13±0.84 and7.82±1.30;t values:-5.892 and-5.240,both P<0.001).The SUVmean and SUVmax of 37 metastases were 16.40±6.07 and 24.25±11.35 respectively in 16 patients,which were significantly higher than those in patients with degenerative changes(20 patients with 27 lesions;9.30±3.16 and 13.48±5.21;t values:6.074 and 5.083,both P<0.001).Conclusion Based on the specific reconstruction algorithm and correction method,the SUV of lumbar vertebra on 99Tcm-MDP SPECT/CT bone imaging can be obtained,and the bone mineral metabolism can be reflected.
ABSTRACT
Objective To investigate the differences of acquisition protocols from continuous bed motion (CBM) and step-and-shoot (SS) modes and to observe their effects on image quality and standard uptake value (SUV) in 18 F-fluorodeoxyglucose (FDG) PET/CT.Methods A total of 30 patients (13 males,17 females;40-71 years) who underwent 18F-FDG PET/CT from June 2017 to September 2017 were selected.Simulated acquisition protocols for a specific range (upper margin of the skull to the lower edge of sciatic bone) were established with CBM and SS modes.The differences between 2 modes for actual length requiring for a specific acquisition range and the differences in CT radiation dose were compared.Real PET/ CT scans were performed using CBM and SS modes consecutively,and the differences in image quality and SUV were compared.Paired t test andx2 test were used to analyze the data.Results For the specific acquisition range,the average acquisition length of CBM was reduced by 6.65％ ((87.11 ± 3.78) vs (93.32 ±6.02) cm;t=-7.737,P＜0.001) and the CT radiation dose was reduced by 6.88％ ((812±170) vs (872±192) mGy · cm;t=-6.432,P＜0.001) for each patient compared with the results of SS.There were no significant differences in maximum SUV (SUVmax) and mean SUV (SUVmean) between SS and CBM in normal tissues including liver,bone and waist muscles (t values:from-1.895 to 0.132,all P＞0.05).The SUVmax of leg muscles at the end of the image was significantly higher in SS than that in CBM (1.24±0.53 vs 1.06±0.42;t=3.450,P＜0.01).There were no statistically significant differences in SUVmax and SUV between SS and CBM in 40 FDG high uptake lesions (t values:0.420 and-0.260,both P＞0.05).There were 73.33％ (22/30) patients had images with overall high quality during SS and the percentage was 80.00％ (24/30) during CBM (x2 =0.373,P＞0.05).The percentage of patients with images of high quality at the end was 16.67％(5/30) during SS,which was significantly less than that during CBM (63.33％,19/30;x2 =13.611,P＜0.001).Conclusions For the specific acquisition range,CBM can reduce unnecessary CT scan range and radiation dose compared with SS.There is no significant difference in image quality and SUV from normal tissue and lesion except for the end of the image.
ABSTRACT
Objective To evaluate the impact of 3-dimensional ordered subset expectation maximization (3D-OSEM),2-dimensional ordered subset expectation maximization (2D-OSEM),filtered-back projection (FBP),and the number of OSEM iterations and subsets on quantitative results in SPECT/CT.Methods A Jaszczak cylindrical phantom and NEMA/IEC body phantom were performed in routine SPECT/CT using CT attenuation correction,scatter correction and 3D-OSEM reconstructed image.The system volume sensitivity (cpm/kBq) was acquired from the reconstructed image of Jaszczak phantom,and the absolute activity concentration (kBq/ml) of the hot sphere from IEC phantom was calculated and compared with true activity concentration,then quantitative error was calculated.With changing the reconstruction algorithm (2D-OSEM,FBP) and OSEM reconstruction parameters (number of iterations and subsets) to reconstruct the IEC phantom,the differences of quantitative results from different reconstruction algorithms and OSEM iterations and subsets were compared.Results When reconstructing images with 3D-OSEM,2D-OSEM and FBP,the quantitative error of each sphere increased with the decrease of the spherical volume (r=-0.831,-0.831,-0.826,all P＜0.05).The quantitative error of different volume spheres among three reconstruction algorithms was statistically significant (F=8.850,P＜0.05).Among them,the result of 3D-OSEM was less than that of 2D-OS EM (P＜0.05),and of 2D-OSEM was less than of FBP (P＜0.05).The quantitative error of different volume spheres decreased with the increasing of number of iterations (r=-0.721,-0.681,-0.691,-0.711,-0.845,-0.893,all P＜0.05) and decreased with the increase of number of subsets (r=-0.670,-0.694,-0.717,-0.852,-0.956,-0.998,all P＜0.05).Conclusion The quantitative accuracy of 3D-OSEM reconstruction based on CT attenuation correction and scattering correction is significantly better than that of 2D-OSEM and FBP.Choosing the appropriate OSEM reconstruction parameters is helpful to improve the quantification accuracy in SPECT/CT.
ABSTRACT
Objective To investigate the methodology for absolute quantification with SPECT/CT using specified reconstruction algorithms and correction methods,and to determine the influence factors for quantitative accuracy.Methods The routine SPECT/CT imaging were performed with Jaszczak cylindrical phantom and International Electrotechnical Conmission (IEC) body phantom using a given acquisition and reconstruction parameter.The system volume sensitivities(SvoI)were acquired from the reconstructed images of Jaszczak phantom study.The absolute radioactivity of the hot spheres from IEC phantom study was calculated,and then the quantitative accuracies were checked.Differences of absolute quantitative value on quantitative accuracy were compared between 3 dimensional (3D)-ordered-subsets expectation maximization (OSEM) and 2 dimensional (2D)-OSEM,CT attenuation correction (AC) and no attenuation correction (NOAC),scatter correction (SC) and no scatter correction (NOSC),partial volume effect (PVE) correction and no PVE correction.Two-sample t test,one-way analysis of variance and Pearson correlation analysis were used.Results When 3D-OSEM,CTAC,SC and PVE correction were used,the average difference between absolute quantitative value and true value (374 kBq/ml) was-2.49％ in 6 hot spheres,and there was no obvious correlation between the quantitative error and the spheres volume (r=-0.795,P＞0.05).In the absence of PVE correction,the quantitative error of each sphere increased with the decrease of sphere volume (r=-0.852,P＜0.05).Differences of the quantitative value and the true value for 6 hot spheres among different reconstruction/correction methods were statistically significant (F =826.631,P＜0.05).The difference was smaller in 3D-OSEM+CTAC+NOSC than that in 2D-OSEM+CTAC+NOSC:(52.83±13.79) vs (61.33 ± 15.00) k Bq/ml (P＜0.05).The similar results were found when comparing differences between 3DOSEM+CTAC+SC and 3D-OSEM+CTAC+NOSC,also between 3D-OSEM+CTAC+NOSC and 3D-OSEM+ NOAC+NOSC:(9.33±5.47) vs (52.83±13.79) kBq/ml,(52.83±13.79) vs (307.66±9.24) kBq/ml (both P＜0.05).Conclusions An accurate absolute quantitative value could be obtained with 3D-OSEM reconstruction algorithm,CTAC,SC and PVE correction in SPECT/CT imaging.
ABSTRACT
Purpose The partial volume effect is a main factor to impact the image quality and the accuracy of the quantitative analysis of nuclear medicine image.The aim of this study is to evaluate the effects of partial volume effect on quantitative result in SPECT/CT.Materials and Methods A Jaszczak cylindrical phantom and IEC body phantom were performed with routine SPECT/CT imaging using CT attenuation correction,scatter correction and three-dimensional ordered subset expectation maximization (3D-OSEM) reconstruct image.The system volume sensitivity (cpm/kBq) were acquired from the reconstructed image of Jaszczak phantom study.The absolute activity concentration (kBq /ml) of the hot sphere was calculated according to the IEC phantom study that the intensity of radioactivity had been measured in the dose calibrator,then checking computations the quantitative accuracy of all.The difference of quantitative accuracies from the different volume hot sphere was compared tusing a given acquisition and reconstruction parameter.The effects of volume of interest (VOI),target-to-background ratio (TBR) and acquisition time on quantitative accuracy were evaluated.Results CT attenuation correction and scatter correction,the difference of absolute activity and true activity value was obviously correlation with the sphere size by using 3D-OSEM.The quantitative errors were decreased along with the increased of sphere size (r=-0.844,P＜0.05).The quantitative accuracy were significantly affected by the VOI size in the 1.15-11.49 ml sphere.It was obviously lower in the limbic of sphere than that in the center of sphere.And the quantitative errors were decreased along with the decreased of VOI size (r=0.999,0.992,0.994 and 0.767,P＜0.05).However,the quantitative error was no obviously correlation with the VOI size (r=0.348,0.478,P＞0.05) in the 26.52 ml and 0.52 ml sphere.The average quantitative error was lower in high TBR imaging (44.19％) than that in low TBR imaging (-46.18％),and it was lower in 20 s/frame imaging (-44.33％) than that in 10 s/frame imaging (-46.04％).Conclusion The quantitative accurate could be obviously affected by the partial volume effect in SPECT/CT imaging,exceptionally in the small lesions and the limbic of lesions.