Generalized algorithms for direct reconstruction of parametric images from dynamic PET data

Direct reconstruction of cardiac PET kinetic parametric images using a preconditioned conjugate gradient approach

PURPOSE

Our research goal is to develop an algorithm to reconstruct cardiac positron emission tomography (PET) kinetic parametric images directly from sinograms and compare its performance with the conventional indirect approach.

METHODS

Time activity curves of a NCAT phantom were computed according to a one-tissue compartmental kinetic model with realistic kinetic parameters. The sinograms at each time frame were simulated using the activity distribution for the time frame. The authors reconstructed the parametric images directly from the sinograms by optimizing a cost function, which included the Poisson log-likelihood and a spatial regularization terms, using the preconditioned conjugate gradient (PCG) algorithm with the proposed preconditioner. The proposed preconditioner is a diagonal matrix whose diagonal entries are the ratio of the parameter and the sensitivity of the radioactivity associated with parameter. The authors compared the reconstructed parametric images using the direct approach with those reconstructed using the conventional indirect approach.

RESULTS

At the same bias, the direct approach yielded significant relative reduction in standard deviation by 12%-29% and 32%-70% for 50 × 10(6) and 10 × 10(6) detected coincidences counts, respectively. Also, the PCG method effectively reached a constant value after only 10 iterations (with numerical convergence achieved after 40-50 iterations), while more than 500 iterations were needed for CG.

CONCLUSIONS

The authors have developed a novel approach based on the PCG algorithm to directly reconstruct cardiac PET parametric images from sinograms, and yield better estimation of kinetic parameters than the conventional indirect approach, i.e., curve fitting of reconstructed images. The PCG method increases the convergence rate of reconstruction significantly as compared to the conventional CG method.

Patlak image estimation from dual time-point list-mode PET data

We investigate using dual time-point PET data to perform Patlak modeling. This approach can be used for whole body dynamic PET studies in which we compute voxel-wise estimates of Patlak parameters using two frames of data for each bed position. Our approach directly uses list-mode arrival times for each event to estimate the Patlak parametric image. We use a penalized likelihood method in which the penalty function uses spatially variant weighting to ensure a count independent local impulse response. We evaluate performance of the method in comparison to fractional changes in SUV values (%DSUV) between the two frames using Cramer Rao analysis and Monte Carlo simulation. Receiver operating characteristic (ROC) curves are used to compare performance in differentiating tumors relative to background based on the dynamic data sets. Using area under the ROC curve as a performance metric, we show superior performance of Patlak relative to %DSUV over a range of dynamic data sets and parameters. These results suggest that Patlak analysis may be appropriate for analysis of dual time-point whole body PET data and could lead to superior detection of tumors relative to %DSUV metrics.

Quality and precision of parametric images created from PET sinogram data by direct reconstruction: proof of concept

We have previously implemented the direct reconstruction of dense kinetic model parameter images ("parametric images") from sinogram data, and compared it to conventional image domain kinetic parameter estimation methods . Although it has been shown that the direct reconstruction algorithm estimates the kinetic model parameters with lower root mean squared error than the conventional image domain techniques, some theoretical obstacles remain. These obstacles include the difficulty of evaluating the accuracy and precision of the estimated parameters. In image domain techniques, the reconstructed time activity curve (TAC) and the model predicted TAC are compared, and the goodness-of-fit is evaluated as a measure of the accuracy and precision of the estimated parameters. This approach cannot be applied to the direct reconstruction technique as there are no reconstructed TACs. In this paper, we propose ways of evaluating the precision and goodness-of-fit of the kinetic model parameters estimated by the direct reconstruction algorithm. Specifically, precision of the estimates requires the calculation of variance images for each parameter, and goodness-of-fit is addressed by reconstructing the difference between the measured and the fitted sinograms. We demonstrate that backprojecting the difference from sinogram space to image space creates error images that can be examined for goodness-of-fit and model selection purposes. The presence of nonrandom structures in the error images may indicate an inadequacy of the kinetic model that has been incorporated into the direct reconstruction algorithm. We introduce three types of goodness-of-fit images. We propose and demonstrate a number-of-runs image as a means of quantifying the adequacy or deficiency of the model. We further propose and demonstrate images of the F statistic and the change in the Akaike Information Criterion as devices for identifying the statistical advantage of one model over another at each voxel. As direct reconstruction to parametric images proliferates, it will be essential for imagers to adopt methods such as those proposed herein to assess the accuracy and precision of their parametric images.

Dynamic positron emission tomography image restoration via a kinetics-induced bilateral filter

Dynamic positron emission tomography (PET) imaging is a powerful tool that provides useful quantitative information on physiological and biochemical processes. However, low signal-to-noise ratio in short dynamic frames makes accurate kinetic parameter estimation from noisy voxel-wise time activity curves (TAC) a challenging task. To address this problem, several spatial filters have been investigated to reduce the noise of each frame with noticeable gains. These filters include the Gaussian filter, bilateral filter, and wavelet-based filter. These filters usually consider only the local properties of each frame without exploring potential kinetic information from entire frames. Thus, in this work, to improve PET parametric imaging accuracy, we present a kinetics-induced bilateral filter (KIBF) to reduce the noise of dynamic image frames by incorporating the similarity between the voxel-wise TACs using the framework of bilateral filter. The aim of the proposed KIBF algorithm is to reduce the noise in homogeneous areas while preserving the distinct kinetics of regions of interest. Experimental results on digital brain phantom and in vivo rat study with typical ¹⁸F-FDG kinetics have shown that the present KIBF algorithm can achieve notable gains over other existing algorithms in terms of quantitative accuracy measures and visual inspection.

Preclinical evaluation of parametric image reconstruction of [18F]FMISO PET: correlation with ex vivo immunohistochemistry

Compared to indirect methods, direct parametric image reconstruction (PIR) has the advantage of high quality and low statistical errors. However, it is not yet clear if this improvement in quality is beneficial for physiological quantification. This study aimed to evaluate direct PIR for the quantification of tumor hypoxia using the hypoxic fraction (HF) assessed from immunohistological data as a physiological reference. Sixteen mice with xenografted human squamous cell carcinomas were scanned with dynamic [18F]FMISO PET. Afterward, tumors were sliced and stained with H&E and the hypoxia marker pimonidazole. The hypoxic signal was segmented using k-means clustering and HF was specified as the ratio of the hypoxic area over the viable tumor area. The parametric Patlak slope images were obtained by indirect voxel-wise modeling on reconstructed images using filtered back projection and ordered-subset expectation maximization (OSEM) and by direct PIR (e.g., parametric-OSEM, POSEM). The mean and maximum Patlak slopes of the tumor area were investigated and compared with HF. POSEM resulted in generally higher correlations between slope and HF among the investigated methods. A strategy for the delineation of the hypoxic tumor volume based on thresholding parametric images at half maximum of the slope is recommended based on the results of this study.

Direct estimation of kinetic parametric images for dynamic PET

Dynamic positron emission tomography (PET) can monitor spatiotemporal distribution of radiotracer in vivo. The spatiotemporal information can be used to estimate parametric images of radiotracer kinetics that are of physiological and biochemical interests. Direct estimation of parametric images from raw projection data allows accurate noise modeling and has been shown to offer better image quality than conventional indirect methods, which reconstruct a sequence of PET images first and then perform tracer kinetic modeling pixel-by-pixel. Direct reconstruction of parametric images has gained increasing interests with the advances in computing hardware. Many direct reconstruction algorithms have been developed for different kinetic models. In this paper we review the recent progress in the development of direct reconstruction algorithms for parametric image estimation. Algorithms for linear and nonlinear kinetic models are described and their properties are discussed.

Tumor quantification in clinical positron emission tomography

Positron emission tomography (PET) is used extensively in clinical oncology for tumor detection, staging and therapy response assessment. Quantitative measurements of tumor uptake, usually in the form of standardized uptake values (SUVs), have enhanced or replaced qualitative interpretation. In this paper we review the current status of tumor quantification methods and their applications to clinical oncology. Factors that impede quantitative assessment and limit its accuracy and reproducibility are summarized, with special emphasis on SUV analysis. We describe current efforts to improve the accuracy of tumor uptake measurements, characterize overall metabolic tumor burden and heterogeneity of tumor uptake, and account for the effects of image noise. We also summarize recent developments in PET instrumentation and image reconstruction and their impact on tumor quantification. Finally, we offer our assessment of the current development needs in PET tumor quantification, including practical techniques for fully quantitative, pharmacokinetic measurements.

Scanning linear estimation: improvements over region of interest (ROI) methods

In tomographic medical imaging, a signal activity is typically estimated by summing voxels from a reconstructed image. We introduce an alternative estimation scheme that operates on the raw projection data and offers a substantial improvement, as measured by the ensemble mean-square error (EMSE), when compared to using voxel values from a maximum-likelihood expectation-maximization (MLEM) reconstruction. The scanning-linear (SL) estimator operates on the raw projection data and is derived as a special case of maximum-likelihood estimation with a series of approximations to make the calculation tractable. The approximated likelihood accounts for background randomness, measurement noise and variability in the parameters to be estimated. When signal size and location are known, the SL estimate of signal activity is unbiased, i.e. the average estimate equals the true value. By contrast, unpredictable bias arising from the null functions of the imaging system affect standard algorithms that operate on reconstructed data. The SL method is demonstrated for two different tasks: (1) simultaneously estimating a signal's size, location and activity; (2) for a fixed signal size and location, estimating activity. Noisy projection data are realistically simulated using measured calibration data from the multi-module multi-resolution small-animal SPECT imaging system. For both tasks, the same set of images is reconstructed using the MLEM algorithm (80 iterations), and the average and maximum values within the region of interest (ROI) are calculated for comparison. This comparison shows dramatic improvements in EMSE for the SL estimates. To show that the bias in ROI estimates affects not only absolute values but also relative differences, such as those used to monitor the response to therapy, the activity estimation task is repeated for three different signal sizes.

Patient-adaptive lesion metabolism analysis by dynamic PET images

Dynamic PET imaging provides important spatial-temporal information for metabolism analysis of organs and tissues, and generates a great reference for clinical diagnosis and pharmacokinetic analysis. Due to poor statistical properties of the measurement data in low count dynamic PET acquisition and disturbances from surrounding tissues, identifying small lesions inside the human body is still a challenging issue. The uncertainties in estimating the arterial input function will also limit the accuracy and reliability of the metabolism analysis of lesions. Furthermore, the sizes of the patients and the motions during PET acquisition will yield mismatch against general purpose reconstruction system matrix, this will also affect the quantitative accuracy of metabolism analyses of lesions. In this paper, we present a dynamic PET metabolism analysis framework by defining a patient adaptive system matrix to improve the lesion metabolism analysis. Both patient size information and potential small lesions are incorporated by simulations of phantoms of different sizes and individual point source responses. The new framework improves the quantitative accuracy of lesion metabolism analysis, and makes the lesion identification more precisely. The requirement of accurate input functions is also reduced. Experiments are conducted on Monte Carlo simulated data set for quantitative analysis and validation, and on real patient scans for assessment of clinical potential.

Parametric imaging via kinetics-induced filter for dynamic positron emission tomography

Due to the noisy measurement of the voxel-wise time activity curve (TAC), parametric imaging for dynamic positron emission tomography (PET) is a challenging task. To address this problem, some spatial filters, such as Gaussian filter, bilateral filter, wavelet-based filter, and so on, are often performed to reduce the noise of each frame. However, these filters usually just consider local properties of each frame without exploring the kinetic information. In this paper, aiming to improve the quantitative accuracy of parametric imaging, we present a kinetics-induced filter to lower the noise of dynamic PET images by incorporating the kinetic information. The present kinetics-induced filter is designed via the similarity between voxel-wise TACs under the framework of bilateral filter. Experimental results with a simulation study demonstrate that the present kinetics-induced filter can achieve noticeable gains than other existing methods for parametric images in terms of quantitative accuracy measures.

Direct 4-D PET list mode parametric reconstruction with a novel EM algorithm

The production of images of kinetic parameters is often the ultimate goal of positron emission tomography (PET) imaging. The indirect method of PET parametric imaging, also called the frame-based method (FM), is performed by fitting the time-activity curve (TAC) for each voxel with an appropriate compartment model after image reconstruction. The indirect method is simple and easily implemented, however, it usually leads to some loss of accuracy or precision, due to the use of two separate steps. This paper presents a direct 4-D method for producing 3-D images of kinetic parameters from list mode PET data. In this application, the TAC for each voxel is described by a one-tissue compartment model (1T). Extending previous EM algorithms, a new spatiotemporal complete data space was introduced to optimize the maximum likelihood function. This leads to a straightforward closed-form parametric image update equation. This method was implemented by extending the current list mode platform MOLAR to produce a parametric algorithm PMOLAR-1T. Using an ordered subset approach, qualitative and quantitative evaluations were performed using 2-D (x, t) and 4-D (x, y, z, t) simulated list mode data based on brain receptor tracers and also with a human brain study. Comparisons with the indirect method showed that the proposed direct method can lead to accurate estimation of the parametric image values with reduced variance, especially at low count levels. In the 2-D test, the direct method showed similar bias to the frame-based method but with variance reduction of 23%-60%. In the 4-D test, bias values of both methods were no more than 4% and the direct method had lower variability (coefficient of variation reduction of 0%-64% compared to the frame-based method) at the normal count level. The direct method had a larger reduction in variability (27%-81%) and lower bias (1%-5% for 4-D and 1%-19% for FM) at low count levels. The results in the human brain study are similar with PMOLAR-1T showing lower noise than FM.

3.5D dynamic PET image reconstruction incorporating kinetics-based clusters

Standard 3D dynamic positron emission tomographic (PET) imaging consists of independent image reconstructions of individual frames followed by application of appropriate kinetic model to the time activity curves at the voxel or region-of-interest (ROI). The emerging field of 4D PET reconstruction, by contrast, seeks to move beyond this scheme and incorporate information from multiple frames within the image reconstruction task. Here we propose a novel reconstruction framework aiming to enhance quantitative accuracy of parametric images via introduction of priors based on voxel kinetics, as generated via clustering of preliminary reconstructed dynamic images to define clustered neighborhoods of voxels with similar kinetics. This is then followed by straightforward maximum a posteriori (MAP) 3D PET reconstruction as applied to individual frames; and as such the method is labeled '3.5D' image reconstruction. The use of cluster-based priors has the advantage of further enhancing quantitative performance in dynamic PET imaging, because: (a) there are typically more voxels in clusters than in conventional local neighborhoods, and (b) neighboring voxels with distinct kinetics are less likely to be clustered together. Using realistic simulated (11)C-raclopride dynamic PET data, the quantitative performance of the proposed method was investigated. Parametric distribution-volume (DV) and DV ratio (DVR) images were estimated from dynamic image reconstructions using (a) maximum-likelihood expectation maximization (MLEM), and MAP reconstructions using (b) the quadratic prior (QP-MAP), (c) the Green prior (GP-MAP) and (d, e) two proposed cluster-based priors (CP-U-MAP and CP-W-MAP), followed by graphical modeling, and were qualitatively and quantitatively compared for 11 ROIs. Overall, the proposed dynamic PET reconstruction methodology resulted in substantial visual as well as quantitative accuracy improvements (in terms of noise versus bias performance) for parametric DV and DVR images. The method was also tested on a 90 min (11)C-raclopride patient study performed on the high-resolution research tomography. The proposed method was shown to outperform the conventional method in visual as well as quantitative accuracy improvements (in terms of noise versus regional DVR value performance).

An optimization transfer algorithm for nonlinear parametric image reconstruction from dynamic PET data

Direct reconstruction of kinetic parameters from raw projection data is a challenging task in molecular imaging using dynamic positron emission tomography (PET). This paper presents a new optimization transfer algorithm for penalized likelihood direct reconstruction of nonlinear parametric images that is easy to use and has a fast convergence rate. Each iteration of the proposed algorithm can be implemented in three simple steps: a frame-by-frame maximum likelihood expectation-maximization (EM)-like image update, a frame-by-frame image smoothing, and a pixel-by-pixel time activity curve fitting. Computer simulation shows that the direct algorithm can achieve a better bias-variance performance than the indirect reconstruction algorithm. The convergence rate of the new algorithm is substantially faster than our previous algorithm that is based on a separable paraboloidal surrogate function. The proposed algorithm has been applied to real 4-D PET data.

Direct 4D parametric imaging for linearized models of reversibly binding PET tracers using generalized AB-EM reconstruction

Due to high noise levels in the voxel kinetics, development of reliable parametric imaging algorithms remains one of most active areas in dynamic brain PET imaging, which in the vast majority of cases involves receptor/transporter studies with reversibly binding tracers. As such, the focus of this work has been to develop a novel direct 4D parametric image reconstruction scheme for such tracers. Based on a relative equilibrium (RE) graphical analysis formulation (Zhou et al 2009b Neuroimage 44 661-70), we developed a closed-form 4D EM algorithm to directly reconstruct distribution volume (DV) parametric images within a plasma input model, as well as DV ratio (DVR) images within a reference tissue model scheme (wherein an initial reconstruction was used to estimate the reference tissue time-activity curves). A particular challenge with the direct 4D EM formulation is that the intercept parameters in graphical (linearized) analysis of reversible tracers (e.g. Logan or RE analysis) are commonly negative (unlike for irreversible tracers, e.g. using Patlak analysis). Subsequently, we focused our attention on the AB-EM algorithm, derived by Byrne (1998, Inverse Problems 14 1455-67) to allow inclusion of prior information about the lower (A) and upper (B) bounds for image values. We then generalized this algorithm to the 4D EM framework, thus allowing negative intercept parameters. Furthermore, our 4D AB-EM algorithm incorporated and emphasized the use of spatially varying lower bounds to achieve enhanced performance. As validation, the means of parameters estimated from 55 human (11)C-raclopride dynamic PET studies were used for extensive simulations using a mathematical brain phantom. Images were reconstructed using conventional indirect as well as proposed direct parametric imaging methods. Noise versus bias quantitative measurements were performed in various regions of the brain. Direct 4D EM reconstruction resulted in notable qualitative and quantitative accuracy improvements (over 35% noise reduction, with matched bias, in both plasma and reference-tissue input models). Similar improvements were also observed in the coefficient of variation of the estimated DV and DVR values even for relatively low uptake cortical regions, suggesting the enhanced ability for robust parameter estimation. The method was also tested on a 90 min (11)C-raclopride patient study performed on the high-resolution research tomograph wherein the proposed method was shown across a variety of regions to outperform the conventional method in the sense that for a given DVR value, improved noise levels were observed.

Hepatic blood perfusion measured by 3-minute dynamic 18F-FDG PET in pigs

There is an unmet clinical need for an imaging method for quantification of hepatic blood perfusion. The purpose of the present study was to develop and validate a PET method using blood-to-cell clearance (K(1)) of (18)F-FDG, 3-O-(11)C-methylglucose ((11)C-MG), or 2-(18)F-fluoro-2-deoxy-D-galactose ((18)F-FDGal) as a measure of hepatic blood perfusion without the need for portal venous blood samples. We aimed to make the method as simple as possible with the prospect of future application to clinical studies. For this purpose, we examined the possibility of using a 3-min data acquisition and a model-derived dual input calculated from measurements of radioactivity concentrations in a peripheral artery.

METHODS

Pigs (40 kg) underwent dynamic PET of the liver with (18)F-FDG, (11)C-MG, or (18)F-FDGal with simultaneous measurements of time-activity curves in blood sampled from a femoral artery and the portal vein (PV); blood flow rates were measured in the hepatic artery (HA) and PV by transit-time flow meters. Two input functions were compared: A measured dual input and a model-derived dual input, the latter with the PV time-activity curve estimated from the measured arterial time-activity curve and a previously validated 1-parametric PV model. (K(1)) was estimated for each tracer by fitting compartmental models to the data, comparing 3-min and 60-min data acquisitions and the 2 dual-input time-activity curves.

RESULTS

Agreement between (K(1)) estimated using the measured and the model-derived dual input was good for all 3 tracers. For (18)F-FDG and (11)C-MG, (K(1)) (3-min data acquisition, model-derived dual input, and 1-tissue compartmental model) correlated to the measured blood perfusion (P = 0.01 and P = 0.07, respectively). For (18)F-FDGal, the correlation was not significant.

CONCLUSION

A simplified method for quantification of hepatic blood perfusion using 3-min dynamic (18)F-FDG PET or (11)C-MG PET with blood sampling from only a peripheral artery was developed. Parametric (K(1)) images were constructed and showed homogeneous blood perfusion in these normal livers.

Motion compensation for fully 4D PET reconstruction using PET superset data

Fully 4D PET image reconstruction is receiving increasing research interest due to its ability to significantly reduce spatiotemporal noise in dynamic PET imaging. However, thus far in the literature, the important issue of correcting for subject head motion has not been considered. Specifically, as a direct consequence of using temporally extensive basis functions, a single instance of movement propagates to impair the reconstruction of multiple time frames, even if no further movement occurs in those frames. Existing 3D motion compensation strategies have not yet been adapted to 4D reconstruction, and as such the benefits of 4D algorithms have not yet been reaped in a clinical setting where head movement undoubtedly occurs. This work addresses this need, developing a motion compensation method suitable for fully 4D reconstruction methods which exploits an optical tracking system to measure the head motion along with PET superset data to store the motion compensated data. List-mode events are histogrammed as PET superset data according to the measured motion, and a specially devised normalization scheme for motion compensated reconstruction from the superset data is required. This work proceeds to propose the corresponding time-dependent normalization modifications which are required for a major class of fully 4D image reconstruction algorithms (those which use linear combinations of temporal basis functions). Using realistically simulated as well as real high-resolution PET data from the HRRT, we demonstrate both the detrimental impact of subject head motion in fully 4D PET reconstruction and the efficacy of our proposed modifications to 4D algorithms. Benefits are shown both for the individual PET image frames as well as for parametric images of tracer uptake and volume of distribution for (18)F-FDG obtained from Patlak analysis.

Acceleration of the direct reconstruction of linear parametric images using nested algorithms

Parametric imaging using dynamic positron emission tomography (PET) provides important information for biological research and clinical diagnosis. Indirect and direct methods have been developed for reconstructing linear parametric images from dynamic PET data. Indirect methods are relatively simple and easy to implement because the image reconstruction and kinetic modeling are performed in two separate steps. Direct methods estimate parametric images directly from raw PET data and are statistically more efficient. However, the convergence rate of direct algorithms can be slow due to the coupling between the reconstruction and kinetic modeling. Here we present two fast gradient-type algorithms for direct reconstruction of linear parametric images. The new algorithms decouple the reconstruction and linear parametric modeling at each iteration by employing the principle of optimization transfer. Convergence speed is accelerated by running more sub-iterations of linear parametric estimation because the computation cost of the linear parametric modeling is much less than that of the image reconstruction. Computer simulation studies demonstrated that the new algorithms converge much faster than the traditional expectation maximization (EM) and the preconditioned conjugate gradient algorithms for dynamic PET.

Initial Evaluation of Direct 4D Parametric Reconstruction with Human PET Data

Previously, we presented a direct EM method for producing kinetic parameter images from list mode PET data, where the time-activity curve for each voxel is described by a one-tissue compartment model (1T). The initial evaluations were performed with simulations, without motion, randoms, or scatter effects included. By extension of our previous frame-based physics correction methods, a practical direct 4D parametric reconstruction algorithm is now proposed and implemented for human data. Initial evaluations were performed using 3 human subjects with the serotonin transporter tracer [(11)C]AFM. Comparisons with the 2-step approach (frame-based reconstruction followed by voxel-by-voxel parameter estimation) provided encouraging initial results. Regional analysis showed that the 2-step and 4D methods have similar K(1) and V(T) values, but with a consistent difference. Visual analysis showed some noise reduction in 4D. These initial results suggest that direct 4D parametric reconstruction can be performed with real data, and offers the potential for improved accuracy and precision over the 2-step frame method.

DIRECT RECONSTRUCTION OF DYNAMIC PET PARAMETRIC IMAGES USING SPARSE SPECTRAL REPRESENTATION

To generate parametric images for dynamic PET, direct reconstruction from projection data is statistically more efficient than conventional indirect methods that perform image reconstruction and kinetic modeling in two separate steps. Existing direct reconstruction methods often use nonlinear compartmental models, which require the knowledge of model order. This paper presents a direct reconstruction approach using a linear spectral representation and does not require model order assumption. A Laplacian prior is used to ensure sparsity in the spectral representation. The resultant maximum a posteriori (MAP) formulation is solved by an expectation maximization shrinkage algorithm. A bias correction step is developed to improve the MAP estimate. Computer simulations show that the proposed method achieves better bias-variance tradeoff than a conventional indirect method for estimating parametric images from dynamic PET data.