List-mode Image Reconstruction Research Articles

Performance comparison of a dedicated total breast PET system with a clinical whole-body PET system: a simulation study

This paper presents a novel PET geometry for breast cancer imaging. The scanner consists of a ‘stadium’ (a rectangle with two semi-circles on opposite sides) shaped ring, along with anterior and posterior panels to provide high sensitivity and high spatial resolution for an imaging field-of-view (FOV) that include both breasts, mediastinum and axilla. We simulated this total-breast PET system using GATE and reconstructed the coincidence events using a GPU-based list-mode image reconstruction implementing maximum likelihood expectation-maximization (ML-EM) algorithm. The rear-panel is made up of a single layer of LSO crystals (3.2 × 3.2 × 20 mm3 each), while the ‘stadium’-shaped elongated ring and the anterior panel are made with dual-layered LSO crystals (1.6 × 1.6 × 6 mm3 each). The energy resolution and coincidence resolving time of all detectors are assumed to be 12% and 250 ps full-width-at-half-maximum, respectively. Various sized simulated lesions (4, 5, 6 mm) having 4:1, 5:1, and 6:1 lesion-to-background radioactivity concentration ratios, mimicking different biological uptakes, were strategically located throughout a volumetric torso phantom. We compared system sensitivity and lesion detectability of the dedicated total-breast PET system to a state-of-the-art clinical whole-body PET scanner. The mean sensitivity of the total-breast PET system is 3.21 times greater than that of a whole-body PET scanner in the breast regions. The total-breast PET system also provides better contrast-recovery coefficients for lesions of all sizes and lesion-to-background ratios in the breast when compared to a reference clinical whole-body PET scanner. Receiver operating characteristics (ROC) study shows the area under the ROC curve is 0.948 and 0.924 for the total-breast system and the whole-body PET scanner, respectively, in the detection of 4 mm diameter lesions with 4:1 lesion-to-background ratio. This study demonstrates our novel geometry can provide an imaging FOV larger than conventional PEM systems to simultaneously image both breasts, chest wall and axillae with significantly improved lesion detectability in the breasts when compared to a whole-body PET scanner.

Rapid workflow of mMR PET list-mode data processing using CUDA.

The purpose of this work is to provide a software solution which would enable very fast list-mode data processing offered by the massively parallel graphics processor units (GPU) while dealing efficiently with the common bottleneck of data transfers between hard disk storage, host (CPU) memory and device (GPU) memory. This software is dedicated to the PETLINK™ list-mode data format which is used by the Siemens’ hybrid Biograph Molecular MR (mMR) scanner. This software is developed as part of our platform for mMR 3D and 4D PET image reconstruction and will be freely accessible to the community. The aim is to perform real time processing (i.e. without perceivable delay) of the list-mode data during data transfers from hard disk to the CPU memory. The output of the processing includes: (1) The count-rate data (head curve: prompts, randoms and singles per second). (2) Plot of the variation of the centre of mass due to kinetics and motion (crude motion detection and quantification). (3) Projection movies (sagittal and coronal) for visual inspection for motion and data quality. (4) Crystal fan-sums for any time frame (used for randoms noise reduction). (5) Singles rate for each detector bucket (used for dead-time correction during normalisation). (6) Static and dynamic sinograms of span-1 and span-11. This workflow enables list-mode data processing synchronously with data transfer from disk to CPU memory. It opens a way for fast creation of multiple bootstrap realisations and multiple image reconstructions for a single dataset providing valuable insight into distributions of any statistic used in the development of robust image bio-markers. Also, the proposed workflow is advantageous for list-mode image reconstruction.

Reconstruction-Incorporated Respiratory Motion Correction in Clinical Simultaneous PET/MR Imaging for Oncology Applications.

Simultaneous PET and MR imaging is a promising new technique allowing the fusion of functional (PET) and anatomic/functional (MR) information. In the thoracic-abdominal regions, respiratory motion is a major challenge leading to reduced quantitative and qualitative image accuracy. Correction methodologies include the use of gated frames that lead to low signal-to-noise ratio considering the associated low statistics. More advanced correction approaches, previously developed for PET/CT imaging, consist of either registering all the reconstructed gated frames to the reference frame or incorporating motion parameters into the iterative reconstruction process to produce a single motion-compensated PET image. The goal of this work was to compare these two—previously implemented in PET/CT—correction approaches within the context of PET/MR motion correction for oncology applications using clinical 4-dimensional PET/MR acquisitions. Two different correction approaches were evaluated comparing the incorporation of elastic transformations extracted from 4-dimensional MR imaging datasets during PET list-mode image reconstruction to a postreconstruction image-based approach. Eleven patient datasets acquired on a PET/MR system were used. T1-weighted 4D MR images were registered to the end-expiration image using a nonrigid B-spline registration algorithm to derive deformation matrices accounting for respiratory motion. The derived matrices were subsequently incorporated within a PET image reconstruction of the original emission list-mode data (reconstruction space [RS] method). The corrected images were compared with those produced by applying the deformation matrices in the image space (IS method) followed by summing the realigned gated frames, as well as with uncorrected motion-averaged images. Both correction techniques led to significant improvement in accounting for respiratory motion artifacts when compared with uncorrected motion-averaged images. These improvements included signal-to-noise ratio (mean increase of 28.0% and 24.2% for the RS and IS methods, respectively), lesion size (reduction of 60.4% and 47.9%, respectively), lesion contrast (increase of 70.1% and 57.2%, respectively), and lesion position (changes of 60.9% and 46.7%, respectively). Our results demonstrate significant respiratory motion compensation using both methods, with superior results from a 4D PET RS approach.

PET Energy-Based Scatter Estimation in the Presence of Randoms, and Image Reconstruction With Energy-Dependent Scatter and Randoms Corrections

In this paper, we address the problem of energy-based scatter estimation in PET in the presence of randoms. We refine a previous proposed model for comprehensive use of the energy information in PET [Phys. Med. Biol., vol. 51, pp. 2919-2937, 2006] by introducing a model for the random coincidences. This model is used to estimate the scatter components in randoms from delayed coincidence data, and then of the nonrandom coincidence scatters from the prompt coincidence data. By performing these estimations on a spatial grid covering the data space, we obtain scatter and randoms correction terms that depend both on position and energy. These terms are incorporated into the list-mode image reconstruction algorithm, thus allowing for an individual treatment of each event according to its position and energy information. We test this approach by using a recently developed image quality evaluation procedure that measures the detectability of signals at unknown locations. The procedure combines a three-dimensional image scanning technique with a novel nonparametric free-response data analysis method that provides a signal detectability metric by using an exponential transformation of the free-response operating characteristic (EFROC). The method allows for scaling of the detectability index to any given image size, a property that makes it advantageous for application to phantom experiments with multiple signals, or large background volumes that can be scanned for false signals. Such experiments can provide substantial signal detectability information with only a few images, as we show here. Image reconstructions using traditional, no energy-dependent, and energy-dependent corrections are compared for different time-of-flight (TOF) resolutions, and number of counts. A comparison of PET time-of-flight (TOF) performance versus no-TOF is also presented.

GPU-Based PET Image Reconstruction Using an Accurate Geometrical System Model

In positron emission tomography (PET), 3D iterative image reconstruction methods have a huge computational burden. In this paper, we developed a list-mode image reconstruction method using graphics processing units (GPUs). Efficiency of acceleration for GPU implementation largely depends on the method chosen, where a reduced number of conditional statements and a reduced memory size are required. On the other hand, accurate system models are required to improve the quality of reconstructed images. Various accurate system models for conventional CPU implementation have been proposed, but these models basically require many conditional statements and huge memory size. Therefore, we developed a new system model which matches GPU implementation better. In this model, the detector response functions, which vary depending on each line of response (LOR), are pre-computed in CPUs and modeled by sixth-order polynomial functions in order to reduce the memory size occupied in GPUs. Each element of a system matrix is obtained on-the-fly in GPUs by calculating the distance between an LOR and a voxel. Therefore the developed system model enables efficient GPU implementation of the accurate system modeling with a reduced number of conditional statements and a reduced memory size. We applied the developed method to a small OpenPET prototype, in which 4-layered depth-of-interaction (DOI) detectors were used. For image reconstruction, we used the dynamic row-action maximum likelihood algorithm (DRAMA). Compared with a conventional model for GPU implementation, in which DRFs are given as a Gaussian function of fixed width, we saw no remarkable difference for DOI data, but for non-DOI data the proposed model outperformed the conventional at the peripheral region of the field-of-view. The proposed model had almost the same calculation time as the conventional model did. For further acceleration, we tried parallel GPU implementation, and we obtained 3.8-fold acceleration by using 4 GPUs.

Calculation of the Sensitivity Image in List-Mode Reconstruction for PET

List-mode image reconstruction is attracting renewed interest in emission tomography as the number of possible lines of response (LORs) in modern scanners becomes more than the number of detected events in one data set. One advantage of list-mode processing is that it eliminates the storage of empty sinogram bins. However, a single back projection of all LORs is still necessary for calculating the sensitivity image. Since the detection sensitivity is dependent on the object attenuation and detector efficiency, the sensitivity image must be computed for each study. Exact computation of the sensitivity image can be a daunting task for modern scanners with a huge number of LORs. Thus, some fast approximate calculation is desirable. Here we use theoretical results that we have derived to analyze a Monte Carlo based backprojection approach for calculating the sensitivity image. We derive expressions for the total number of LORs required in the backprojection and the sampling distribution. Computer simulations show that the nonuniform sampling schemes can reduce reconstruction artifact compared to the uniform sampling approach with the same computation cost

List-mode image reconstruction for real-time PET imaging

The problem of tomographic image reconstruction from measured projection data can be reduced to solving a set of linear equations represented in matrix form. In this paper, we show how image reconstruction in positron emission tomography (PET) can be achieved potentially in real-time using a regularized pseudo-inverse of the system matrix. An update of the current image estimate accounting for the next registered event can be obtained using a single column of the regularized pseudo-inverse matrix, thus allowing for instant visualization of the radioactivity distribution while the subject is still being scanned. The computed image estimate is the minimum-norm least-squares solution of the regularized inverse problem. The speed and performance of the algorithm were characterized and compared to those of a well-known statistical iterative algorithm. The proposed algorithm was validated with experimental list-mode PET data.

Statistical dynamic image reconstruction in state-of-the-art high-resolution PET

Modern high-resolution PET is now more than ever in need of scrutiny into the nature and limitations of the imaging modality itself as well as image reconstruction techniques. In this work, we have reviewed, analysed and addressed the following three considerations within the particular context of state-of-the-art dynamic PET imaging: (i) the typical average numbers of events per line-of-response (LOR) are now (much) less than unity, (ii) due to the physical and biological decay of the activity distribution, one requires robust and efficient reconstruction algorithms applicable to a wide range of statistics and (iii) the computational considerations in dynamic imaging are much enhanced (i.e., more frames to be stored and reconstructed). Within the framework of statistical image reconstruction, we have argued theoretically and shown experimentally that the sinogram non-negativity constraint (when using the delayed-coincidence and/or scatter-subtraction techniques) is especially expected to result in an overestimation bias. Subsequently, two schemes are considered: (a) subtraction techniques in which an imagenon-negativity constraint has been imposed and (b) implementation of random and scatter estimates inside the reconstruction algorithms, thus enabling direct processing of Poisson-distributed prompts. Both techniques are able to remove the aforementioned bias, while the latter, being better conditioned theoretically, is able to exhibit superior noise characteristics. We have also elaborated upon and verified the applicability of the accelerated list-mode image reconstruction method as a powerful solution for accurate, robust and efficient dynamic reconstructions of high-resolution data (as well as a number of additional benefits in the context of state-of-the-art PET).

Statistical list-mode image reconstruction for the high resolution research tomograph

We have investigated statistical list-mode reconstruction applicable to a depth-encoding high resolution research tomograph. An image non-negativity constraint has been employed in the reconstructions and is shown to effectively remove the overestimation bias introduced by the sinogram non-negativity constraint. We have furthermore implemented a convergent subsetized (CS) list-mode reconstruction algorithm, based on previous work (Hsiao et al 2002 Conf. Rec. SPIE Med. Imaging 4684 10–19; Hsiao et al 2002 Conf. Rec. IEEE Int. Symp. Biomed. Imaging 409–12) on convergent histogram OSEM reconstruction. We have demonstrated that the first step of the convergent algorithm is exactly equivalent (unlike the histogram-mode case) to the regular subsetized list-mode EM algorithm, while the second and final step takes the form of additive updates in image space. We have shown that in terms of contrast, noise as well as FWHM width behaviour, the CS algorithm is robust and does not result in limit cycles. A hybrid algorithm based on the ordinary and the convergent algorithms is also proposed, and is shown to combine the advantages of the two algorithms (i.e. it is able to reach a higher image quality in fewer iterations while maintaining the convergent behaviour), making the hybrid approach a good alternative to the ordinary subsetized list-mode EM algorithm.