Clinical evaluation of 2D versus 3D whole-body PET image quality using a dedicated BGO PET scanner

Purpose Three-dimensional positron emission tomography (3D PET) results in higher system sensitivity, with an associated increase in the detection of scatter and random coincidences. The objective of this work was to compare, from a clinical perspective, 3D and two-dimensional (2D) acquisitions in terms of whole-body (WB) PET image quality with a dedicated BGO PET system.Methods 2D and 3D WB emission acquisitions were carried out in 70 patients. Variable acquisition parameters in terms of time of emission acquisition per axial field of view (aFOV) and slice overlap between sequential aFOVs were used during the 3D acquisitions. 3D and 2D images were reconstructed using FORE+WLS and OSEM respectively. Scatter correction was performed by convolution subtraction and a model-based scatter correction in 2D and 3D respectively. All WB images were attenuation corrected using segmented transmission scans. Images were blindly assessed by three observers for the presence of artefacts, confidence in lesion detection and overall image quality using a scoring system.Results Statistically significant differences between 2D and 3D image quality were only obtained for 3D emission acquisitions of 3 min. No statistically significant differences were observed for image artefacts or lesion detectability scores. Image quality correlated significantly with patient weight for both modes of operation. Finally, no differences were seen in image artefact scores for the different axial slice overlaps considered, suggesting the use of five slice overlaps in 3D WB acquisitions.Conclusion 3D WB imaging using a dedicated BGO-based PET scanner offers similar image quality to that obtained in 2D considering similar overall times of acquisitions.     

Evaluating image reconstruction methods for tumor detection in 3-dimensional whole-body PET oncology imaging.

We compare 3 image reconstruction algorithms for use in 3-dimensional (3D) whole-body PET oncology imaging. We have previously shown that combining Fourier rebinning (FORE) with 2-dimensional (2D) statistical image reconstruction via the ordered-subsets expectation-maximization (OSEM) and attenuation-weighted OSEM (AWOSEM) algorithms demonstrates improvements in image signal-to-noise ratios compared with the commonly used analytic 3D reprojection (3DRP) or FORE+FBP (2D filtered backprojection) reconstruction methods. To assess the impact of these reconstruction methods on detecting and localizing small lesions, we performed a human observer study comparing the different reconstruction methods. The observer study used the same volumetric visualization software tool that is used in clinical practice, instead of a planar viewing mode as is generally used with the standard receiver operating characteristic (ROC) methodology. This change in the human evaluation strategy disallowed the use of a ROC analysis, so instead we compared the fraction of actual targets found and reported (fraction-found) and also investigated the use of an alternative free-response operating characteristic (AFROC) analysis. METHODS: We used a non-Monte Carlo technique to generate 50 statistically accurate realizations of 3D whole-body PET data based on an extended mathematic cardiac torso (MCAT) phantom and with noise levels typical of clinical scans performed on a PET scanner. To each realization, we added 7 randomly located 1-cm-diameter lesions (targets) whose contrasts were varied to sample the range of detectability. These targets were inserted in 3 organs of interest: lungs, liver, and soft tissues. The images were reconstructed with 3 reconstruction strategies (FORE+OSEM, FORE+AWOSEM, and FORE+FBP). Five human observers reported (localized and rated) 7 targets within each volume image. An observer's performance accuracy with each algorithm was measured, as a function of the lesion contrast and organ type, by the fraction of those targets reported and by the area below the AFROC curve. This AFROC curve plots the fraction of reported targets at each rating threshold against the fraction of cases with (> or =1) similarly rated false reports. RESULTS: Images reconstructed with FORE+AWOSEM yielded the best overall target detection as compared with FORE+FBP and FORE+OSEM, although these differences in detectability were region specific. The FORE+FBP and FORE+AWOSEM algorithms had similar performances for liver targets. The FORE+OSEM algorithm performed significantly worse at target detection, especially in the liver. We speculate that this is the result of using an incorrect statistical model for OSEM and that the incorporation of attenuation weighting in AWOSEM largely compensates for this model inaccuracy. These results were consistent for both the fraction of actual targets found and the AFROC analysis. CONCLUSION: We demonstrated the efficacy of performing observer detection studies using the same visualization tools as those used in clinical PET oncology imaging. These studies demonstrated that the FORE+AWOSEM algorithm led to the best overall detection and localization performance for 1-cm-diameter targets compared with the FORE+OSEM and FORE+FBP algorithms.     

Reducing loss of image quality because of the attenuation artifact in uncorrected PET whole-body images.

In whole-body PET, it is not unusual to shorten the study time by omitting the transmission scan and to ignore attenuation during reconstruction. If a transmission scan is available, many centers reconstruct the images with, but also without, attenuation correction. Although ignoring attenuation leads to an artifact in the reconstructed images, these images still provide valuable diagnostic information in oncologic applications. Several authors have reported that the attenuation artifact may actually increase the tumor-to-background ratio. In this study, we analyzed the causes of the artifact and proposed a new algorithm to reduce the adverse effects on visual image quality. METHODS: We analyzed the causes of the attenuation artifact mathematically and numerically, and we examined its effect on tumor-to-background ratio and on signal-to-noise ratio. In addition, we showed that the attenuation artifact may lead to loss of image detail in conventional maximum-likelihood expectation maximization (MLEM) reconstruction. A new maximum-likelihood algorithm allowing negative reconstruction values (NEG-ML) was derived to reduce this loss. RESULTS: The attenuation artifact consists of 2 components. The first component is the well-known scaling effect: The apparent activity is reduced because attenuation decreases the fraction of detected photons. The second component is a relatively smooth negative contribution that is added to attenuated regions surrounded by activity. The second component tends to increase the tumor-to-background ratio. However, a simulation experiment shows that this increase in signal may be entirely offset by an increase in noise. The negative contribution can interfere with the nonnegativity constraint of the MLEM algorithm, leading to loss of image detail in regions of high attenuation. The new NEG-ML algorithm avoids the problem by allowing negative pixel values. The algorithm is similar to MLEM in the suppression of the streak artifact but provides more anatomic information. In our department, it is in routine clinical use for reconstruction of PET whole-body images without attenuation correction. CONCLUSION: Ignoring attenuation may increase the tumor-to-background ratio, but this increase does not imply improved tumor detection. The NEG-ML algorithm reduces the adverse effect of the attenuation artifact on visual image quality.     

A lesion detection observer study comparing 2-dimensional versus fully 3-dimensional whole-body PET imaging protocols.

We compared the impact of 2-dimensional (2D) and fully 3-dimensional (3D) acquisition modes on the performance of human observers in detecting and localizing tumors in whole-body (18)F-FDG images. METHODS: We selected protocols based on noise equivalent count (NEC) rates derived from a series of 2D and fully 3D whole-body patient and phantom acquisitions on a dual-mode PET scanner. The fully 3D peak NEC value for a standard 70-kg patient was achieved for an injected dose of approximately 444 MBq (12 mCi) assuming a 90-min delay before acquisition, whereas the 2D peak value was never reached. The protocols were therefore set to those corresponding to a 444-MBq injected dose in fully 3D and 2D and a 740-MBq (20 mCi) injected dose in 2D that was considered as the maximum allowable dose. We used a non-Monte Carlo simulator to generate multiple realizations of whole-body PET data based on the geometry of the mathematic cardiac torso phantom (MCAT) with accurate noise properties. Two-dimensional and fully 3D acquisition times were set to 5 min per bed position. Spherical 1-cm-diameter lesions (targets) with random locations and contrasts were distributed in different organs. The simulated 2D datasets were reconstructed using attenuation-weighted ordered-subsets expectation maximization ((AW)OSEM) and the fully 3D datasets were reconstructed with FORE+(AW)OSEM (FORE = Fourier rebinning). Five human observers located and ranked the targets using a volumetric display of the whole-body PET data to replicate the clinical practice. An alternate free-response operating characteristic (AFROC) analysis of the human observer reports was performed for each protocol and each organ separately. RESULTS: The 2D protocol corresponding to 740-MBq injected dose allowed the overall best detection performance. It was followed by the fully 3D acquisition at the peak fully 3D NEC rate from a 444-MBq injected dose. A 2D acquisition corresponding to a 444-MBq injected dose was ranked last. Differences in detection performance were organ specific. CONCLUSION: This study showed that, for this patient size and scanner type, the fully 3D acquisition mode allowed better or equivalent detection performance than the 2D mode for an injected dose corresponding to the peak fully 3D NEC rate. The 2D acquisition protocol combined with a higher injected dose resulted in the highest detectabilities.     

Correction of head movement on PET studies: comparison of methods.

Head movement presents a continuing problem in PET studies. Head restraint minimizes movement but is unreliable, resulting in the need to develop alternative strategies. These include frame-by-frame (FBF) realignment or use of motion tracking (MT) during the scan to realign PET acquisition data. Here we present a comparative analysis of these 2 methods of motion correction. METHODS: Eight volunteers were examined at rest using (11)C-raclopride PET with the radioligand administered as a bolus followed by constant infusion to achieve steady state. Binding potential (BP) was estimated using the ratio method during 2 periods of the scan at steady state. Head movement was compensated by using coregistration between frames (FBF) and 3 methods using MT measurements of head position acquired with a commercially available optical tracking system. RESULTS: All methods of realignment improved test-retest reliability and noise characteristics of the raw data, with important consequences for the power to detect small changes in radiotracer binding, and the potential to reduce false-positive and false-negative results. MT methods were superior to FBF realignment using coregistration on some indices. CONCLUSION: Such methods have considerable potential to improve the reliability of PET data with important implications for the numbers of volunteers required to test hypotheses.     

18F-FDG PET for detecting myocardial viability: validation of 3D data acquisition.

(18)F-FDG PET is an important diagnostic tool for detecting myocardial viability in patients with coronary artery disease. In combination with perfusion scanning, (18)F-FDG PET allows differentiation between reversibly and irreversibly damaged myocardium and selection of patients likely to benefit from revascularization. Viability PET is usually performed in two-dimensional (2D) mode. Taking into account the rising number of three-dimensional (3D)-only scanners, a validation of 3D acquisition is required. METHODS: Twenty-one patients with coronary artery disease referred for (18)F-FDG PET underwent an imaging protocol of nongated 2D (2D-NG) and gated 2D (2D-G) acquisitions for 15 min each, followed by 3D gated acquisitions for 10 min (3D-10) and 5 min (3D-5), using an ECAT Exact HR+ scanner. Results were analyzed using a 20-segment polar map in terms of activity concentration (Bq/mL), viability (50% uptake threshold), regional activity distribution, visual assessment of viability based on a 3-point rating scale, and left ventricular ejection fraction. RESULTS: Activity concentration measured in each segment with 2D-G, 3D-10, and 3D-5 showed a good linear correlation with 2D-NG. Quantitative viability assessment with 3D-5 gave a sensitivity of 84% and a specificity of 98%, compared with 2D-NG. No differences in regional activity distribution and visual viability assessment were found between the various protocols. Left ventricular ejection fractions obtained with 3D-10 and 3D-5 showed a good linear correlation with those measured with 2D-G. CONCLUSION: An ECG-gated 3D imaging protocol gave results comparable to those of 2D acquisition with regard to absolute and regional myocardial activity distribution, left ventricular function, and visual viability assessment. Sensitivity for viability assessment with a 50% uptake threshold was significantly less with 3D, but specificity was maintained. This protocol delivers a clinical performance nearly equivalent to that of 2D acquisition.     

Investigation of the resolution and count recovery coefficients of a whole-body PET scanner (Shimadzu SET-2400W) in 2D and 3D mode image

We measured the resolution and count recovery coefficients (RC) of the SET-2400W whole-body PET scanner (Shimadzu Co., Japan) in the 2D and 3D clinical modes. METHOD: The 3D images were reconstructed by using the full 3D image reconstruction method (3-D reprojection algorithm: 3DRP) and the Fourier rebinning method (FORE). The 2D images were reconstructed with conventional filtered back-projection method (FBP). The measurements of resolution and recovery coefficient were according to JRIA (Japan Radioisotope Association) protocols. RESULTS: The transaxial resolutions of all methods were better than 7 mm FWHM at a radius of 10 cm with 1.25 cm-1 cutoff frequency. The average slice width of 2D FBP, 3DRP and FORE are 5.8 mm, 8.0 mm and 6.8 mm respectively at the center of transaxial field of view. The RC values were measured in a range from 10 mm to 27 mm at 6 cm from the center with the cylindrical and spherical hot area phantoms. In all methods, RC values at 27 mm diameter were nearly 1.0 in both type of hot area. RC values at 10 mm diameter in 2D FBP, 3DRP and FORE of cylindrical hot area were 0.69, 0.72, 0.73 and those of spherical hot area were 0.52, 0.51, 0.53 respectively. CONCLUSION: At the SET-2400W, resolution and recovery coefficient of 3D mode image under the clinical mode showed the value which did not differ from the 2D mode image.     

Low-dose whole-body CT using deep learning image reconstruction: image quality and lesion detection

Objectives:To evaluate image quality and lesion detection capabilities of low-dose (LD) portal venous phase whole-body computed tomography (CT) using deep learning image reconstruction (DLIR).Methods:The study cohort of 59 consecutive patients (mean age, 67.2 years) who underwent whole-body LD CT and a prior standard-dose (SD) CT reconstructed with hybrid iterative reconstruction (SD-IR) within one year for surveillance of malignancy were assessed. The LD CT images were reconstructed with hybrid iterative reconstruction of 40% (LD-IR) and DLIR (LD-DLIR). The radiologists independently evaluated image quality (5-point scale) and lesion detection. Attenuation values in Hounsfield units (HU) of the liver, pancreas, spleen, abdominal aorta, and portal vein; the background noise and signal-to-noise ratio (SNR) of the liver, pancreas, and spleen were calculated. Qualitative and quantitative parameters were compared between the SD-IR, LD-IR, and LD-DLIR images. The CT dose-index volumes (CTDI_vol) and dose-length product (DLP) were compared between SD and LD scans.Results:The image quality and lesion detection rate of the LD-DLIR was comparable to the SD-IR. The image quality was significantly better in SD-IR than in LD-IR (p < 0.017). The attenuation values of all anatomical structures were comparable between the SD-IR and LD-DLIR (p = 0.28–0.96). However, background noise was significantly lower in the LD-DLIR (p < 0.001) and resulted in improved SNRs (p < 0.001) compared to the SD-IR and LD-IR images. The mean CTDI_vol and DLP were significantly lower in the LD (2.9 mGy and 216.2 mGy•cm) than in the SD (13.5 mGy and 1011.6 mGy•cm) (p < 0.0001).Conclusion:LD CT images reconstructed with DLIR enable radiation dose reduction of >75% while maintaining image quality and lesion detection rate and superior SNR in comparison to SD-IR.Advances in knowledge:Deep learning image reconstruction algorithm enables around 80% reduction in radiation dose while maintaining the image quality and lesion detection compared to standard-dose whole-body CT.     

Clinical implications of different image reconstruction parameters for interpretation of whole-body PET studies in cancer patients.

The standardized uptake value (SUV) is the most commonly used parameter to quantify the intensity of radiotracer uptake in tumors. Previous studies suggested that measurements of (18)F-FDG accumulation in tissue might be affected by the image reconstruction method, but the clinical relevance of these findings has not been assessed. METHODS: Phantom studies were performed and clinical whole-body (18)F-FDG PET images of 85 cancer patients were analyzed. All images were reconstructed using either filtered backprojection (FBP) with measured attenuation correction (MAC) or iterative reconstruction (IR) with segmented attenuation correction (SAC). In a subset of 15 patients, images were reconstructed using all 4 combinations of IR+SAC, IR+MAC, FBP+SAC, and FBP+MAC. For phantom studies, a sphere containing (18)F-FDG was placed in a water-filled cylinder and the activity concentration of that sphere was measured in FBP and IR reconstructed images using all 4 combinations. Clinical studies were displayed simultaneously and identical regions of interest (ROIs, 50 pixels) were placed in liver, urinary bladder, and tumor tissue in both image sets. SUV max (maximal counts per pixel in ROI) and SUV avg (average counts per pixel) were measured. RESULTS: In phantom studies, measurements from FBP images underestimated the true activity concentration to a greater degree than those from IR images (20% vs. 5% underestimation). In patient studies, SUV derived from FBP images were consistently lower than those from IR images in both normal and tumor tissue: Tumor SUV max with IR+SAC was 9.6 +/- 4.5, with IR+MAC it was 7.7 +/- 3.5, with FBP+MAC it was 6.9 +/- 3.0, and with FBP+SAC it was 8.6 +/- 4.1 (all P < 0.01 vs. IR+SAC). Compared with IR+SAC, SUV from FBP+MAC images were 25%-30% lower. Similar discrepancies were noted for liver and bladder. Discrepancies between measurements became more apparent with increasing (18)F-FDG concentration in tissue. CONCLUSION: SUV measurements in whole-body PET studies are affected by the applied methods for both image reconstruction and attenuation correction. This should be considered when serial PET studies are done in cancer patients. Moreover, if SUV is used for tissue characterization, different cutoff values should be applied, depending on the chosen method for image reconstruction and attenuation correction.     

Development and evaluation of MRI based Bayesian image reconstruction methods for PET.

A maximum a posteriori algorithm, which incorporates correlated magnetic resonance images into the processing of positron emission tomography reconstruction with the aim of improving image quality was developed. The line site map from MRI a priori is made up of a modified Markov random field or Canny edge detector with Gaussian smoothing filter. It is used in the MAP algorithm by a weighted line site method. We evaluate and compare the performance of these reconstruction methods. The results show that the Bayesian methods produce reconstructed images with less noise and better spatial resolution than those produced by the maximum likelihood-expectation maximization method.     

PET instrumentation and reconstruction algorithms in whole-body applications.

The aim of this work is the presentation and comparison of state-of-the-art dedicated PET systems actually available on the market, in terms of physical performance and technical features. Particular attention has been given to evaluate the whole-body performance by sensitivity, spatial resolution, dead time, noise equivalent counting rate (NECR), and scatter fraction. PET/CT systems were also included as new proposals to improve diagnostic accuracy of PET, allowing effective anatomic integration to functional data. An overview of actually implemented reconstruction algorithms is also reported to fully understand all of the factors that contribute to image quality.     

Impact of body habitus on quantitative and qualitative image quality in whole-body FDG-PET

Obtaining consistent high image quality is desirable for clinical positron emission tomography (PET). Body morphology may impact image quality. The purpose of this study was to define the average and the range of body sizes in patients undergoing tumor PET studies in our center and to determine how the body habitus affects the statistical and visual quality of PET images. Height, weight, body surface area (BSA), and body mass index (BMI) were determined in 101 male and 101 female patients (group 1) referred for clinical PET. The summed total counts from three consecutive transaxial slices on non-attenuation-corrected (NAC) 2D fluorine-18 fluorodeoxyglucose (FDG) PET images, which included the largest liver section and no lesions, were determined and compared with body morphology and injected doses (ID) in a representative group of 30 male and 30 female patients (group 2) spanning a range of body morphologies. The visual quality of images was also evaluated using a scoring system by three readers. The average height, weight, and BSA were greater in male than in female patients, but the average BMI was not different between them in group 1. The largest value of weight or BMI was more than four times the smallest value in female patients. The total true counts were best correlated with ID/weight (mCi/kg) in group 2 ( r=0.929, P<0.0001). Intermediate to high total counts (930,000 or more) corresponded to ID/weight of 0.22 or higher. The average visual score was positively correlated with the total counts (rho=0.63, P<0.0001) and with ID/weight (rho=0.68, P<0.0001) on NAC images. The image quality in 22 (84.6%) of 26 patients with intermediate to high total counts was adequate to good, whereas that in 21 (61.8%) of 34 patients with lower total counts was suboptimal. A wide variety of body morphologies was observed in patients referred for clinical FDG-PET tumor studies in our center. The total counts and average image visual score were negatively correlated with weight. Counts in heavy patients were as low as one-fourth those in light patients. Adjusting injected FDG dose in each patient on the basis of body weight may be more appropriate to achieve consistent PET image quality than giving a fixed injected FDG dose.