Search

Aim: The ever-growing diversity and complexity of PET/CT systems make it increasingly difficult to define common acquisition protocols for quantifiable and comparable PET measurements in clinical routine and, above all, in multicenter clinical trials. We aimed to analyze the dependency of quantitative PET on technical variability, in order to improve acquisition and reconstruction protocols. The results formed the base for selected clinical acquisitions protocols, tailored to particular PET/CT systems. Materials and Methods: Image noise was examined in 422 PET/CT datasets from 18 PETCT systems participating in a Swiss multicenter phantom study. 215 of these measurements contained also hot spheres for recovery curve (RC) analysis. The study protocol combined different acquisition durations and low- and high-resolution image reconstructions with filtered back projection (FBP), ordered subset expectation maximization (OSEM), and vendor specific point spread function (PSF) based reconstruction. Data was analyzed with regard to exposure, defined as the product of background activity concentration and acquisition time. This produced results comparable in relation to the number of available decays per volume, independently from the actual activity concentration. Quality assurance (QA) limits for RCs were taken from guidelines issued by the Federal Office of Public Health in Switzerland. Additionally, RC shapes were quantitatively analyzed. Results: Passing the given QA limits depended highly on adequate exposure by keeping image noise levels low. Also, the minimal exposures required to fulfill the given QA limits differed between PET/CT systems, acquisition protocols and reconstruction algorithms. Despite higher image noise levels, quantifiability of FBP data was affected less by low count numbers than OSEM or PSF data. Insufficient count numbers led to faulty and incomparable image quantification,and consequently also erratic RCs. From the gathered data, we were able to propose limits for minimal exposure for ten different PET/CT devices and for common clinical protocols, suitable for whole body and organ-specific high-resolution acquisitions. Conclusions: In order to reproducibly generate comparable and quantitative data, count numbers in PET/CT acquisitions must be specifically adapted to the clinically necessary image resolution, reconstruction method and the PET/CT system used. Therefore, optimal PET/CT acquisition times vary with the required exposure and with the expected activity uptake in target organs. This in turn argues against fixed acquisition times and mandates site and organ specific acquisitions protocols. Adapting exposure to a site’s distinct technical factors appears to be a relevant element of PET/CT standardization in the context of multi-center trials.

Context: PET quality assurance (QA) is becoming increasingly important as well as more complex, following the technological and radiopharmaceutical progress in PET imaging. Consequently, PET quality control (QC) has to handle an increasing heterogeneity of devices and diversity of clinical applications. However, actual PET QC focuses on equipment and good practice, while QC of individual datasets stays costly in terms of time and resources, and thus often lacks adequate diligence. Aim: Our aim is to promote automated PET QC of individual datasets and demonstrate its importance and feasibility as integral part of PET imaging. First objective was to develop a software prototype for automated PET QC on diverse data sets. Second objective was to demonstrate feasibility and benefits of such software. In particular, the automated QC should enable an in depth evaluation of data from a nationwide PET survey organized by the Swiss Society of Nuclear Medicine (SGNM-SSMN). Materials and Methods: The software implements a multi-paradigm approach based on an in-house software development tool. Among a range of modern design patterns, the software implements an embedded rule-based system to support decisions and reconfigurations of itself at runtime. A total of 453 datasets originating from 18 different PET/CT systems at 14 Swiss sites was available for QC. According to a detailed study protocol, each site was requested to provide a range of various image reconstructions from multiple acquisitions of two differing phantoms. The software was applied in a single run to all datasets, producing an individual quality report for each dataset and a database making all results available for further evaluation. Results: The automated QC correctly classified the diverse datasets according to acquisition protocol, reconstruction protocol and type of phantom used. Corresponding protocol adherence verification and quantitative evaluations were performed. The system rejected 7% of the data sets because of serious violations of protocol. Another 57% of the datasets were categorized as of restricted use, because of a variety of minor violations of protocol or a mismatch between quantitative specifications and actual data. Only 36% of the data sets fully passed QC. The resulting database proved to be essential for further evaluations as published elsewhere. Conclusion: The results indisputably prove the need, the feasibility and the benefits of automated PET QC. More intelligent software is a prerequisite for rigorous PET QC of individual datasets, especially given the increasing diversity of PET imaging applications.

Aim: To gain an overview about the variability of PET measurements with diverse PET/CT-systems used in Switzerland, and to provide an inventory on acquisition protocol adherence in a multi-centre setting. Materials and methods: A total of 28 PET/CT sites in Switzerland was invited to participate in a PET phantom survey. Based on obligatory semiannual phantom measurements as prescribed by the Federal Office of Public Health in Switzerland, a detailed study protocol defined PET acquisitions of two different durations with a homogeneous cylinder phantom (activity concentration (AC) 10 kBq/ml) and a fillable “hot spheres” (18F) phantom (3.5 kBq background AC; sphere to background ratio 5:1). At least 7 image reconstructions for each acquisition (28 per study site) were requested. Reconstructions included sets of standard (4mm, isotropic) and high resolution (2mm, isotropic) reconstructions, using filtered back-projection (FBP), ordered subset expectation maximization (OSEM) and vendor specific point spread function (PSF) based reconstructions. Data of all sites underwent automatic quality control (QC) utilizing an in-house developed multi-paradigm software and were analyzed with regard to comparability of measured AC. Results: Phantom data was provided by 14 of 28 PET sites (5/5 university hospitals, 7/13 cantonal hospitals, 2/9 private hospitals). From 18 PET devices, 459/504 expected datasets (92%) were received and underwent QC. Of these, 6 datasets were not readable, and 31 datasets were not accepted (e.g. no attenuation correction), resulting in 422 evaluated datasets. 257/422 (61%) datasets were of restricted use for quantification because of protocol violations (e.g. strongly anisotropic voxels, filter parameters or AC out of proposed range, faulty exposure and axial or radial offset of phantom spheres), and/or actual background AC was not provided in 160/422 datasets (38 %). We found high variations in prepared actual or presupposed phantom AC and in recovery of AC, latter depending on reconstruction methods and reconstruction parameters (e.g. isotropy of voxels). Device-specific acquisition parameters - in terms of exposure per reconstructed voxel size - had a major influence on inter-site comparability of quantitative PET measurements. Underexposed PET acquisitions lead to faulty quantification of AC. Conclusions: As expected from comparable studies, the nationwide Swiss phantom survey also found considerable variation in PET AC quantification. In the context of multi-centre trials, protocol compliance may not be assumed. Thus, a throughout QC of all datasets appears to be mandatory to assure data quality and comparability. In addition, quantitative PET data may only be used after check of adequate image exposure.

PURPOSE A PET/CT system's imaging capabilities are best described by its point spread function (PSF) in the spatial domain or equivalently by its modulation transfer function (MTF) in the spatial frequency domain. Knowing PSFs or MTFs is a prerequisite for many numerical methods attempting to improve resolution and to reduce the partial volume effect. In PET/CT, the observed PSF is a convolution of the system's intrinsic imaging capabilities including image reconstruction (PSF0) and the positron range function (PRF) of the imaged β(+) emitting isotope. A PRF describes the non-Gaussian distribution of β(+) annihilation events around a hypothetical point source. The main aim was to introduce a new method for determining a PET/CT system's intrinsic MTF (MTF0) from phantom measurements of hot spheres independently of the β(+) emitting isotope used for image acquisition. Secondary aim was to examine non-Gaussian and nonlinear MTFs of a modern iterative reconstruction algorithm. METHODS PET/CT images of seven phantom spheres with volumes ranging from 0.25 to 16 ml and filled either with (18)F or with (68)Ga were acquired and reconstructed using filtered back projection (FBP). MTFs were modeled with linear splines. The spline fit iteratively minimized the mean squared error between the acquired PET/CT image and a convolution of the thereof derived PSF with a numerical representation of the imaged hot phantom sphere. For determining MTF0, the numerical sphere representations were convolved with a PRF, simulating a fill with either (18)F or (68)Ga. The MTFs determined by this so-called MTF fit method were compared with MTFs derived from point source measurements and also compared with MTFs derived with a previously published PSF fit method. The MTF fit method was additionally applied to images reconstructed by a vendor iterative algorithm with PSF recovery (Siemens TrueX). RESULTS The MTF fit method was able to determine (18)F and (68)Ga dependent MTFs and MTF0 from FBP reconstructed images. Root-mean-square deviation between fit determined MTFs and point source determined MTFs ranged from 0.023 to 0.039. MTFs from Siemens TrueX reconstructions varied with size of the imaged sphere. CONCLUSIONS MTF0 can be determined regardless of the imaged isotope, when using existing PRF models for the MTF fit method presented. The method proves that modern iterative PET/CT reconstruction algorithms have nonlinear imaging properties. This behaviour is not accessible by point source measurements. MTFs resulting from these clinically applied algorithms need to be estimated from objects of similar geometry to those intended for clinical imaging.

PURPOSE Positron emission tomography (PET)∕computed tomography (CT) measurements on small lesions are impaired by the partial volume effect, which is intrinsically tied to the point spread function of the actual imaging system, including the reconstruction algorithms. The variability resulting from different point spread functions hinders the assessment of quantitative measurements in clinical routine and especially degrades comparability within multicenter trials. To improve quantitative comparability there is a need for methods to match different PET∕CT systems through elimination of this systemic variability. Consequently, a new method was developed and tested that transforms the image of an object as produced by one tomograph to another image of the same object as it would have been seen by a different tomograph. The proposed new method, termed Transconvolution, compensates for differing imaging properties of different tomographs and particularly aims at quantitative comparability of PET∕CT in the context of multicenter trials. METHODS To solve the problem of image normalization, the theory of Transconvolution was mathematically established together with new methods to handle point spread functions of different PET∕CT systems. Knowing the point spread functions of two different imaging systems allows determining a Transconvolution function to convert one image into the other. This function is calculated by convolving one point spread function with the inverse of the other point spread function which, when adhering to certain boundary conditions such as the use of linear acquisition and image reconstruction methods, is a numerically accessible operation. For reliable measurement of such point spread functions characterizing different PET∕CT systems, a dedicated solid-state phantom incorporating (68)Ge∕(68)Ga filled spheres was developed. To iteratively determine and represent such point spread functions, exponential density functions in combination with a Gaussian distribution were introduced. Furthermore, simulation of a virtual PET system provided a standard imaging system with clearly defined properties to which the real PET systems were to be matched. A Hann window served as the modulation transfer function for the virtual PET. The Hann's apodization properties suppressed high spatial frequencies above a certain critical frequency, thereby fulfilling the above-mentioned boundary conditions. The determined point spread functions were subsequently used by the novel Transconvolution algorithm to match different PET∕CT systems onto the virtual PET system. Finally, the theoretically elaborated Transconvolution method was validated transforming phantom images acquired on two different PET systems to nearly identical data sets, as they would be imaged by the virtual PET system. RESULTS The proposed Transconvolution method matched different PET∕CT-systems for an improved and reproducible determination of a normalized activity concentration. The highest difference in measured activity concentration between the two different PET systems of 18.2% was found in spheres of 2 ml volume. Transconvolution reduced this difference down to 1.6%. In addition to reestablishing comparability the new method with its parameterization of point spread functions allowed a full characterization of imaging properties of the examined tomographs. CONCLUSIONS By matching different tomographs to a virtual standardized imaging system, Transconvolution opens a new comprehensive method for cross calibration in quantitative PET imaging. The use of a virtual PET system restores comparability between data sets from different PET systems by exerting a common, reproducible, and defined partial volume effect.