Bias reduction for low-statistics PET: maximum likelihood reconstruction with a modified Poisson distribution

Impact of framing scheme optimization and smoking status on binding potential analysis in dynamic PET with [11C]ABP688

BACKGROUND

For positron emission tomography (PET) ligands, such as [¹¹C]ABP688, to be able to provide more evidence about the glutamatergic hypothesis in schizophrenia (SZ), quantification bias during dynamic PET studies and its propagation into the estimated values of non-displaceable binding potential (BP_ND) must be addressed. This would enable more accurate quantification during bolus + infusion (BI) neuroreceptor studies and further our understanding of neurological diseases. Previous studies have shown BP_ND-related biases can often occur due to overestimated cerebellum activity (reference region). This work investigates whether an alternative framing scheme can minimize quantification biases propagated into BP_ND, whether confounders, such as smoking status, need to be controlled for during the study, and what the consequences for the data interpretation following analysis are. A group of healthy controls (HC) and a group of SZ patients (balanced and unbalanced number of smokers) were investigated with [¹¹C]ABP688 and a BI protocol. Possible differences in BP_ND quantification as a function of smoking status were tested with constant 5 min ('Const 5 min') and constant true counts ('Const Trues') framing schemes. In order to find biomarkers for SZ, the differences in smoking effects were compared between groups. The normalized BP_ND and the balanced number of smokers and non-smokers for both framing schemes were evaluated.

RESULTS

When applying F-tests to the 'Const 5 min' framing scheme, effect sizes (η²p) and brain regions which showed significant effects fluctuated considerably with F = 50.106 ± 54.948 (9.389 to 112.607), P-values 0.005 to < 0.001 and η²p = 0.514 ± 0.282 (0.238 to 0.801). Conversely, when the 'Const Trues' framing scheme was applied, the results showed much smaller fluctuations with F = 78.038 ± 8.975 (86.450 to 68.590), P < 0.001 for all conditions and η²p = 0.730 ± 0.017 (0.742 to 0.710), and regions with significant effects were more robustly reproduced. Further, differences, which would indicate false positive identifications between HC and SZ groups in five brain regions when using the 'Const 5 min' framing scheme, were not observed with the 'Const Trues' framing.

CONCLUSIONS

Based on an [¹¹C]ABP688 PET study in SZ patients, the results show that non-consistent BP_ND outcomes can be propagated by the framing scheme and that potential bias can be minimized using 'Const Trues' framing.

Deep learning-based image reconstruction for TOF PET with DIRECT data partitioning format

Conventional positron emission tomography (PET) image reconstruction is achieved by the statistical iterative method. Deep learning provides another opportunity for speeding up the image reconstruction process. However, conventional deep learning-based image reconstruction requires a fully connected network for learning the Radon transform. The use of fully connected networks greatly complicated the network and increased hardware cost. In this study, we proposed a novel deep learning-based image reconstruction method by utilizing the DIRECT data partitioning method. The U-net structure with only convolutional layers was used in our approach. Patch-based model training and testing were used to achieve 3D reconstructions within current hardware limitations. Time-of-flight (TOF)-histoimages were first generated from the listmode data to replace conventional sinograms. Different projection angles were used as different channels in the input. A total of 15 patient data were used in this study. For each patient, the dynamic whole-body scanning protocol was used to expand the training dataset and a total of 372 separate scans were included. The leave-one-patient-out validation method was used. Two separate studies were carried out. In the first study, the measured TOF-histoimages were directly used for model training and testing, to study the performance of the method in real-world applications. In the second study, TOF-histoimages were simulated from already reconstructed images to exclude the scatters, randoms, attenuation-activity mismatch effects. This study was used to evaluate the optimal performance when all other corrections are ideal. Volumes of interests were placed in the liver and lesion region to study image noise and lesion quantitations. The reconstructed images using the proposed deep learning method showed similar image quality when compared with the conventional expectation-maximization approach. A minimal difference was observed when the simulated TOF-histoimages were used as model input and testing, suggesting the deep learning model can indeed learn the reconstruction process. Some quantitative difference was observed when the measured TOF-histoimages were used. The two studies suggested that the major difference is caused by inaccurate corrections performed by the network itself, which indicated that physics-based corrections are still required for better quantitative performance. In conclusion, we have proposed a novel deep learning-based image reconstruction method for TOF PET. With the help of the DIRECT data partitioning method, no fully connected layers were used and 3D image reconstruction can be directly achieved within the limits of the current hardware.

Bias evaluation and reduction in 3D OP-OSEM reconstruction in dynamic equilibrium PET studies with 11C-labeled for binding potential analysis

Iterative image reconstruction is widely used in positron emission tomography. However, it is known to contribute to quantitation bias and is particularly pronounced during dynamic studies with ¹¹C-labeled radiotracers where count rates become low towards the end of the acquisition. As the strength of the quantitation bias depends on the counts in the reconstructed frame, it can differ from frame to frame of the acquisition. This is especially relevant in the case of neuro-receptor studies with simultaneous PET/MR when a bolus-infusion protocol is applied to allow the comparison of pre- and post-task effects. Here, count dependent changes in quantitation bias may interfere with task changes. We evaluated the impact of different framing schemes on quantitation bias and its propagation into binding potential (BP) using a phantom decay study with ¹¹C and 3D OP-OSEM. Further, we propose a framing scheme that keeps the true counts per frame constant over the acquisition time as constant framing schemes and conventional increasing framing schemes are unlikely to achieve stable bias values during the acquisition time range. For a constant framing scheme with 5 minutes frames, the BP bias was 7.13±2.01% (10.8% to 3.8%) compared to 5.63±2.85% (7.8% to 4.0%) for conventional increasing framing schemes. Using the proposed constant true counts framing scheme, a stabilization of the BP bias was achieved at 2.56±3.92% (3.5% to 1.7%). The change in BP bias was further studied by evaluating the linear slope during the acquisition time interval. The lowest slope values were observed in the constant true counts framing scheme. The constant true counts framing scheme was effective for BP bias stabilization at relevant activity and time ranges. The mean BP bias under these conditions was 2.56±3.92%, which represents the lower limit for the detection of changes in BP during equilibrium and is especially important in the case of cognitive tasks where the expected changes are low.

Local-Mean Preserving Post-Processing Step for Non-Negativity Enforcement in PET Imaging: Application to 90Y-PET

In a low-statistics PET imaging context, the positive bias in regions of low activity is a burning issue. To overcome this problem, algorithms without the built-in non-negativity constraint may be used. They allow negative voxels in the image to reduce, or even to cancel the bias. However, such algorithms increase the variance and are difficult to interpret since the resulting images contain negative activities, which do not hold a physical meaning when dealing with radioactive concentration. In this paper, a post-processing approach is proposed to remove these negative values while preserving the local mean activities. Its original idea is to transfer the value of each voxel with negative activity to its direct neighbors under the constraint of preserving the local means of the image. In that respect, the proposed approach is formalized as a linear programming problem with a specific symmetric structure, which makes it solvable in a very efficient way by a dual-simplex-like iterative algorithm. The relevance of the proposed approach is discussed on simulated and on experimental data. Acquired data from an yttrium-90 phantom show that on images produced by a non-constrained algorithm, a much lower variance in the cold area is obtained after the post-processing step, at the price of a slightly increased bias. More specifically, when compared with the classical OSEM algorithm, images are improved, both in terms of bias and of variance.

TOF-PET Image Reconstruction With Multiple Timing Kernels Applied on Cherenkov Radiation in BGO

Today Time-of-Flight (TOF), in PET scanners, assumes a single, well-defined timing resolution for all events. However, recent BGO-Cherenkov detectors, combining prompt Cherenkov emission and the typical BGO scintillation, can sort events into multiple timing kernels, best described by the Gaussian mixture models. The number of Cherenkov photons detected per event impacts directly the detector time resolution and signal rise time, which can later be used to improve the coincidence timing resolution. This work presents a simulation toolkit which applies multiple timing spreads on the coincident events and an image reconstruction that incorporates this information. A full cylindrical BGO-Cherenkov PET model was compared, in terms of contrast recovery and contrast-to-noise ratio, against an LYSO model with a time resolution of 213 ps. Two reconstruction approaches for the mixture kernels were tested: 1) mixture Gaussian and 2) decomposed simple Gaussian kernels. The decomposed model used the exact mixture component applied during the simulation. Images reconstructed using mixture kernels provided similar mean value and less noise than the decomposed. However, typically, more iterations were needed. Similarly, the LYSO model, with a single TOF kernel, converged faster than the BGO-Cherenkov with multiple kernels. The results indicate that the model complexity slows down convergence. However, due to the higher sensitivity, the contrast-to-noise ratio was 26.4% better for the BGO model.

Feasibility and limitations of quantitative SPECT for 223Ra

The aim of this paper is to investigate the feasibility and limitations of activity-concentration estimation for ²²³Ra using SPECT. Phantom measurements are performed using spheres (volumes 5.5 mL to 26.4 mL, concentrations 1.6 kBq mL^-1 to 4.5 kBq mL^-1). Furthermore, SPECT projections are simulated using the SIMIND Monte Carlo program for two geometries, one similar to the physical phantom and the other being an anthropomorphic phantom with added lesions (volumes 34 mL to 100 mL, concentrations 0.5 kBq mL^-1 to 4 kBq mL^-1). Medium-energy and high-energy collimators, 60 projections with 55 s per projection and a 20% energy window at 82 keV are employed. For the Monte Carlo simulated images, Poisson-distributed noise is added in ten noise realizations. Reconstruction is performed (OS-EM, 40 iterations, 6 subsets) employing compensation for attenuation, scatter, and collimator-detector response. The estimated concentrations in the anthropomorphic phantom are also corrected using recovery coefficients. Errors for the largest sphere in the physical phantom range from -25% to -34% for the medium-energy collimator and larger deviations for smaller spheres. Corresponding results for the high-energy collimator are -15% to -31%. The corresponding Monte Carlo simulations show standard deviations of a few percentage points. For the anthropomorphic phantom, before application of recovery coefficients the bias ranges from -16% to -46% (medium-energy collimator) and -10% to -28% (high-energy collimator), with standard deviations of 2% to 14% and 1% to 16%. After the application of recovery coefficients, the biases range from -3% to -35% (medium energy collimator) and from 0% to -18%. The errors decrease with increasing concentrations. Activity-concentration estimation of ²²³Ra with SPECT is feasible, but problems with repeatability need to be further studied.

Comparison of the Biograph Vision and Biograph mCT for quantitative 90Y PET/CT imaging for radioembolisation

BACKGROUND

New digital PET scanners with improved time of flight timing and extended axial field of view such as the Siemens Biograph Vision have come on the market and are expected to replace current generation photomultiplier tube (PMT)-based systems such as the Siemens Biograph mCT. These replacements warrant a direct comparison between the systems, so that a smooth transition in clinical practice and research is guaranteed, especially when quantitative values are used for dosimetry-based treatment guidance. The new generation digital PET scanners offer increased sensitivity. This could particularly benefit 90Y imaging, which tends to be very noisy owing to the small positron branching ratio and high random fraction of 90Y. This study aims to determine the ideal reconstruction settings for the digital Vision for quantitative 90Y imaging and to evaluate the image quality and quantification of the digital Vision in comparison with its predecessor, the PMT-based mCT, for 90Y imaging in radioembolisation procedures.

METHODS

The NEMA image quality phantom was scanned to determine the ideal reconstruction settings for the Vision. In addition, an anthropomorphic phantom was scanned with both the Vision and the mCT, mimicking a radioembolisation patient with lung, liver, tumour, and extrahepatic deposition inserts. Image quantification of the anthropomorphic phantom was assessed by the lung shunt fraction, the tumour to non-tumour ratio, the parenchymal dose, and the contrast to noise ratio of extrahepatic depositions.

RESULTS

For the Vision, a reconstruction with 3 iterations, 5 subsets, and no post-reconstruction filter is recommended for quantitative 90Y imaging, based on the convergence of the recovery coefficient. Comparing both systems showed that the noise level of the Vision is significantly lower than that of the mCT (background variability of 14% for the Vision and 25% for the mCT at 2.5·103 MBq for the 37 mm sphere size). For quantitative 90Y measures, such as needed in radioembolisation, both systems perform similarly.

CONCLUSIONS

We recommend to reconstruct 90Y images acquired on the Vision with 3 iterations, 5 subsets, and no post-reconstruction filter for quantitative imaging. The Vision provides a reduced noise level, but similar quantitative accuracy as compared with its predecessor the mCT.

Comments on the NEMA NU 4-2008 Standard on Performance Measurement of Small Animal Positron Emission Tomographs

The National Electrical Manufacturers Association’s (NEMA) NU 4-2008 standard specifies methodology for evaluating the performance of small-animal PET scanners. The standard’s goal is to enable comparison of different PET scanners over a wide range of technologies and geometries used. In this work, we discuss if the NEMA standard meets these goals and we point out potential flaws and improvements to the standard.For the evaluation of spatial resolution, the NEMA standard mandates the use of filtered backprojection reconstruction. This reconstruction method can introduce star-like artifacts for detectors with an anisotropic spatial resolution, usually caused by parallax error. These artifacts can then cause a strong dependence of the resulting spatial resolution on the size of the projection window in image space, whose size is not fully specified in the NEMA standard. If the PET ring has detectors which are perpendicular to a Cartesian axis, then the resolution along this axis will typically improve with larger projection windows.We show that the standard’s equations for the estimation of the random rate for PET systems with intrinsic radioactivity are circular and not satisfiable. However, a modified version can still be used to determine an approximation of the random rates under the assumption of negligible random rates for small activities and a constant scatter fraction. We compare the resulting estimated random rates to random rates obtained using a delayed coincidence window and two methods based on the singles rates. While these methods give similar estimates, the estimation method based on the NEMA equations overestimates the random rates.In the NEMA standard’s protocol for the evaluation of the sensitivity, the standard specifies to axially step a point source through the scanner and to take a different scan for each source position. Later, in the data analysis section, the standard does not specify clearly how the different scans have to be incorporated into the analysis, which can lead to unclear interpretations of publicized results.The standard’s definition of the recovery coefficients in the image quality phantom includes the maximum activity in a region of interest, which causes a positive correlation of noise and recovery coefficients. This leads to an unintended trade-off between desired uniformity, which is negatively correlated with variance (i.e., noise), and recovery.With this work, we want to start a discussion on possible improvements in a next version of the NEMA NU-4 standard.

PET Reconstruction With Non-Negativity Constraint in Projection Space: Optimization Through Hypo-Convergence

Standard positron emission tomography (PET) reconstruction techniques are based on maximum-likelihood (ML) optimization methods, such as the maximum-likelihood expectation-maximization (MLEM) algorithm and its variations. Most methodologies rely on a positivity constraint on the activity distribution image. Although this constraint is meaningful from a physical point of view, it can be a source of bias for low-count/high-background PET, which can compromise accurate quantification. Existing methods that allow for negative values in the estimated image usually utilize a modified log-likelihood, and therefore break the data statistics. In this paper, we propose to incorporate the positivity constraint on the projections only, by approximating the (penalized) log-likelihood function by an adequate sequence of objective functions that are easily maximized without constraint. This sequence is constructed such that there is hypo-convergence (a type of convergence that allows the convergence of the maximizers under some conditions) to the original log-likelihood, hence allowing us to achieve maximization with positivity constraint on the projections using simple settings. A complete proof of convergence under weak assumptions is given. We provide results of experiments on simulated data where we compare our methodology with the alternative direction method of multipliers (ADMM) method, showing that our algorithm converges to a maximizer, which stays in the desired feasibility set, with faster convergence than ADMM. We also show that this approach reduces the bias, as compared with MLEM images, in necrotic tumors-which are characterized by cold regions surrounded by hot structures-while reconstructing similar activity values in hot regions.

Negative binomial maximum likelihood expectation maximization (NB-MLEM) algorithm for reconstruction of pre-corrected PET data

PURPOSE

Positron emission tomography (PET) image reconstruction is usually performed using maximum likelihood (ML) iterative reconstruction methods, under the assumption of Poisson distributed data. Pre-correcting raw measured counts, this assumption is no longer realistic. The goal of this work is to develop a reconstruction algorithm based on the Negative Binomial (NB) distribution, which can generalize over the Poisson distribution in case of over-dispersion of raw data, that may occur if sinogram pre-correction is used.

METHODS

The mathematical derivation of a Negative Binomial Maximum Likelihood Expectation-Maximization (NB-MLEM) algorithm is presented. A simulation study to compare the performance of the proposed NB-MLEM algorithm with respect to a Poisson-based MLEM (P-MLEM) method was performed, in reconstructing PET data. The proposed NB-MLEM reconstruction was tested on a real phantom and human brain data.

RESULTS

For the property of NB distribution, it is a generalization of the conventional P-MLEM: for not over dispersed data, the proposed NB-MLEM algorithm behaves like the conventional P-MLEM; for over-dispersed PET data, the additional evaluation of the dispersion parameter after each reconstruction iteration leads to a more accurate final image with respect to P-MLEM.

CONCLUSIONS

A novel approach for PET image reconstruction from pre-corrected data has been developed, which exhibits a statistical behavior that deviates from the Poisson distribution. Simulation study and preliminary tests on real data showed how the NB-MLEM algorithm, being able to explain the over-dispersion of pre-corrected data, can outperform other algorithms that assume no over-dispersion of pre-corrected data, while still not accounting for the presence of negative data, such as P-MLEM.

Dynamic whole-body PET imaging: principles, potentials and applications

PURPOSE

In this article, we discuss dynamic whole-body (DWB) positron emission tomography (PET) as an imaging tool with significant clinical potential, in relation to conventional standard uptake value (SUV) imaging.

BACKGROUND

DWB PET involves dynamic data acquisition over an extended axial range, capturing tracer kinetic information that is not available with conventional static acquisition protocols. The method can be performed within reasonable clinical imaging times, and enables generation of multiple types of PET images with complementary information in a single imaging session. Importantly, DWB PET can be used to produce multi-parametric images of (i) Patlak slope (influx rate) and (ii) intercept (referred to sometimes as "distribution volume"), while also providing (iii) a conventional 'SUV-equivalent' image for certain protocols.

RESULTS

We provide an overview of ongoing efforts (primarily focused on FDG PET) and discuss potential clinically relevant applications.

CONCLUSION

Overall, the framework of DWB imaging [applicable to both PET/CT(computed tomography) and PET/MRI (magnetic resonance imaging)] generates quantitative measures that may add significant value to conventional SUV image-derived measures, with limited pitfalls as we also discuss in this work.

A PET reconstruction formulation that enforces non-negativity in projection space for bias reduction in Y-90 imaging

Most existing PET image reconstruction methods impose a nonnegativity constraint in the image domain that is natural physically, but can lead to biased reconstructions. This bias is particularly problematic for Y-90 PET because of the low probability positron production and high random coincidence fraction. This paper investigates a new PET reconstruction formulation that enforces nonnegativity of the projections instead of the voxel values. This formulation allows some negative voxel values, thereby potentially reducing bias. Unlike the previously reported NEG-ML approach that modifies the Poisson log-likelihood to allow negative values, the new formulation retains the classical Poisson statistical model. To relax the non-negativity constraint embedded in the standard methods for PET reconstruction, we used an alternating direction method of multipliers (ADMM). Because choice of ADMM parameters can greatly influence convergence rate, we applied an automatic parameter selection method to improve the convergence speed. We investigated the methods using lung to liver slices of XCAT phantom. We simulated low true coincidence count-rates with high random fractions corresponding to the typical values from patient imaging in Y-90 microsphere radioembolization. We compared our new methods with standard reconstruction algorithms and NEG-ML and a regularized version thereof. Both our new method and NEG-ML allow more accurate quantification in all volumes of interest while yielding lower noise than the standard method. The performance of NEG-ML can degrade when its user-defined parameter is tuned poorly, while the proposed algorithm is robust to any count level without requiring parameter tuning.

Respiratory motion compensation for simultaneous PET/MR based on highly undersampled MR data

PURPOSE

Positron emission tomography (PET) of the thorax region is impaired by respiratory patient motion. To account for motion, the authors propose a new method for PET/magnetic resonance (MR) respiratory motion compensation (MoCo), which uses highly undersampled MR data with acquisition times as short as 1 min/bed.

METHODS

The proposed PET/MR MoCo method (4D jMoCo PET) uses radial MR data to estimate the respiratory patient motion employing MR joint motion estimation and image reconstruction with temporal median filtering. Resulting motion vector fields are incorporated into the system matrix of the PET reconstruction. The proposed approach is evaluated for the thorax region utilizing a PET/MR simulation with 1 min MR acquisition time and simultaneous PET/MR measurements of six patients with MR acquisition times of 1 and 5 min and radial undersampling factors of 11.2 and 2.2, respectively. Reconstruction results are compared to 3D PET, 4D gated PET and a standard MoCo method (4D sMoCo PET), which performs iterative image reconstruction and motion estimation sequentially. Quantitative analysis comprises the parameters SUV_mean, SUV_max, full width at half-maximum/lesion volume, contrast and signal-to-noise ratio.

RESULTS

For simulated PET data, our quantitative analysis shows that the proposed 4D jMoCo PET approach with temporal filtering achieves the best quantification accuracy of all tested reconstruction methods with a mean absolute deviation of 2.3% when compared to the ground truth. For measured PET patient data, the mean absolute deviation of 4D jMoCo PET using a 1 min MR acquisition for motion estimation is 2.1% relative to the 5 min MR acquisition. This demonstrates a robust behavior even in case of strong undersampling at MR acquisition times as short as 1 min. In contrast, 4D sMoCo PET shows considerable reduction of quantification accuracy for the 1 min MR acquisition time. Relative to 3D PET, the proposed 4D jMoCo PET approach with temporal filtering yields an average increase of SUV_mean, SUV_max, and contrast of 29.9% and 13.8% for simulated and measured PET data, respectively.

CONCLUSIONS

Employing artifact-robust motion estimation enables PET/MR respiratory MoCo with MR acquisition times as short as 1 min/bed improving PET image quality and quantification accuracy.

A gate evaluation of the sources of error in quantitative90 Y PET

PURPOSE

Accurate reconstruction of the dose delivered by ⁹⁰ Y microspheres using a postembolization PET scan would permit the establishment of more accurate dose-response relationships for treatment of hepatocellular carcinoma with ⁹⁰ Y. However, the quality of the PET data obtained is compromised by several factors, including poor count statistics and a very high random fraction. This work uses Monte Carlo simulations to investigate what impact factors other than low count statistics have on the quantification of⁹⁰ Y PET.

METHODS

PET acquisitions of two phantoms-a NEMA PET phantom and the NEMA IEC PET body phantom-containing either ⁹⁰ Y or ¹⁸ F were simulated using gate. Simulated projections were created with subsets of the simulation data allowing the contributions of random, scatter, and LSO background to be independently evaluated. The simulated projections were reconstructed using the commercial software for the simulated scanner, and the quantitative accuracy of the reconstruction and the contrast recovery of the reconstructed images were evaluated.

RESULTS

The quantitative accuracy of the ⁹⁰ Y reconstructions were not strongly influenced by the high random fraction present in the projection data, and the activity concentration was recovered to within 5% of the known value. The contrast recovery measured for simulated ⁹⁰ Y data was slightly poorer than that for simulated ¹⁸ F data with similar count statistics. However, the degradation was not strongly linked to any particular factor. Using a more restricted energy range to reduce the random fraction in the projections had no significant effect.

CONCLUSIONS

Simulations of ⁹⁰ Y PET confirm that quantitative ⁹⁰ Y is achievable with the same approach as that used for ¹⁸ F, and that there is likely very little margin for improvement by attempting to model aspects unique to ⁹⁰ Y, such as the much higher random fraction or the presence of bremsstrahlung in the singles data.

Optimization of brain PET imaging for a multicentre trial: the French CATI experience

Background

CATI is a French initiative launched in 2010 to handle the neuroimaging of a large cohort of subjects recruited for an Alzheimer’s research program called MEMENTO. This paper presents our test protocol and results obtained for the 22 PET centres (overall 13 different scanners) involved in the MEMENTO cohort. We determined acquisition parameters using phantom experiments prior to patient studies, with the aim of optimizing PET quantitative values to the highest possible per site, while reducing, if possible, variability across centres.

Methods

Jaszczak’s and 3D-Hoffman’s phantom measurements were used to assess image spatial resolution (ISR), recovery coefficients (RC) in hot and cold spheres, and signal-to-noise ratio (SNR). For each centre, the optimal reconstruction parameters were chosen as those maximizing ISR and RC without a noticeable decrease in SNR. Point-spread-function (PSF) modelling reconstructions were discarded. The three figures of merit extracted from the images reconstructed with optimized parameters and routine schemes were compared, as were volumes of interest ratios extracted from Hoffman acquisitions. The net effect of the 3D-OSEM reconstruction parameter optimization was investigated on a subset of 18 scanners without PSF modelling reconstruction.

Results

Compared to the routine parameters of the 22 PET centres, average RC in the two smallest hot and cold spheres and average ISR remained stable or were improved with the optimized reconstruction, at the expense of slight SNR degradation, while the dispersion of values was reduced.

For the subset of scanners without PSF modelling, the mean RC of the smallest hot sphere obtained with the optimized reconstruction was significantly higher than with routine reconstruction. The putamen and caudate-to-white matter ratios measured on 3D-Hoffman acquisitions of all centres were also significantly improved by the optimization, while the variance was reduced.

Conclusions

This study provides guidelines for optimizing quantitative results for multicentric PET neuroimaging trials.

A method for comparing intra-tumoural radioactivity uptake heterogeneity in preclinical positron emission tomography studies

Background

Non-uniformity influences the interpretation of nuclear medicine based images and consequently their use in treatment planning and monitoring. However, no standardised method for evaluating and ranking heterogeneity exists. Here, we have developed a general algorithm that provides a ranking and a visualisation of the heterogeneity in small animal positron emission tomography (PET) images.

Methods

The code of the algorithm was written using the Matrix Laboratory software (MATLAB). Parameters known to influence the heterogeneity (distances between deviating peaks, gradients and size compensations) were incorporated into the algorithm. All data matrices were mathematically constructed in the same format with the aim of maintaining overview and control. Histograms visualising the spread and frequency of contributions to the heterogeneity were also generated. The construction of the algorithm was tested using mathematically generated matrices and by varying post-processing parameters. It was subsequently applied in comparisons of radiotracer uptake in preclinical images in human head and neck carcinoma and endothelial and ovarian carcinoma xenografts.

Results

Using the developed algorithm, entire tissue volumes could be assessed and gradients could be handled in an indirect manner. Similar-sized volumes could be compared without modifying the algorithm. Analyses of the distribution of different tracers gave results that were generally in accordance with single plane preclinical images, indicating that it could appropriately handle comparisons of targeting vs. non-targeting tracers and also for different target levels. Altering the reconstruction algorithm, pixel size, tumour ROI volumes and lower cut-off limits affected the calculated heterogeneity factors in expected directions but did not reverse conclusions about which tumour was more or less heterogeneous.

Conclusions

The algorithm constructed is an objective and potentially user-friendly tool for one-to-one comparisons of heterogeneity in whole similar-sized tumour volumes in PET imaging.