For Filtered Back Projection Research Articles

Evaluating the Efficacy of Deep Learning Reconstruction in Reducing Radiation Dose for Computer-Aided Volumetry for Liver Tumor: A Phantom Study.

The purpose of this study was to compare radiation dose reduction capability for accurate liver tumor measurements of a computer-aided volumetry (CADv) software for filtered back projection (FBP), hybrid-type iterative reconstruction (IR), mode-based iterative reconstruction (MBIR), and deep learning reconstruction (DLR) at a phantom study. A commercially available anthropomorphic abdominal phantom was scanned five times with a 320-detector row CT at 600 mA, 400 mA, 200 mA, and 100 mA and reconstructed by four methods. Signal-to-noise ratios (SNRs) of all lesions within the arterial and portal-venous phase inserts were calculated, and SNR of the lesion phantom was compared with that of all reconstruction methods by means of Tukey's honestly significant difference (HSD) test. Then, tumor volume (V) of each nodule was automatically measured using commercially available CADv software. To compare dose reduction capability for each reconstruction method at both phases, mean differences between measured V and standard references were compared by Tukey's honestly significant difference test among the four different reconstruction methods on CT obtained at each of the four tube currents. With each of the tube currents, SNRs for MBIR and DLR were significantly higher than those for FBP and hybrid-type IR (p < 0.05). At the arterial phase, the mean difference in V for the CT protocol obtained at 600 or 100 mA and reconstructed with DLR was significantly smaller than that for others (p < 0.05). At the portal-venous phase, the mean differences in V for the CT protocol obtained at 100 mA and reconstructed with hybrid-type IR, MBIR, and DLR were significantly smaller than that for FBP (p < 0.05). Findings of our phantom study show that reconstruction method had influence on CADv merits for abdominal CT with not only standard but also reduced dose examinations and that DLR can potentially yield better image quality and CADv measurements than FBP, hybrid-type IR, or MBIR in this setting.

Performance evaluation of image reconstruction algorithms for a megavoltage computed tomography system on a helical tomotherapy unit

Objective. To evaluate the impact of image reconstruction algorithm selection, as well as imaging mode and the reconstruction interval, on image quality metrics for megavoltage computed tomography (MVCT) image acquisition for use in image-guided (IGRT) and adaptive radiotherapy (ART) on a next-generation helical tomotherapy system. Approach. A CT image quality phantom was scanned across all available acquisition modes for filtered back projection (FBP) and both iterative reconstruction (IR) algorithms available on the system. Image quality metrics including noise, uniformity, contrast, spatial resolution, and mean CT number were compared. Analysis of DICOM data was performed using ImageJ software and Python code. ANOVA single factor and Tukey’s honestly significant difference post-hoc tests were utilized for statistical analysis. Main Results. Application of both IR algorithms noticeably improved noise and image contrast when compared to the FBP algorithm available on all previous-generation helical tomotherapy systems. Use of the FBP algorithm improved image uniformity and spatial resolution in the axial plane, though values for the IR algorithms were well within tolerances recommended for IGRT and/or MVCT-based ART implementation by the American Association of Physicists in Medicine (AAPM). Additionally, longitudinal resolution showed little dependence on the reconstruction algorithm, while a negligible variation in mean CT number was observed regardless of the reconstruction algorithm or acquisition parameters. Statistical analysis confirmed the significance of these results. Significance. An overall improvement in image quality for metrics most important to IGRT and ART—mainly image noise and contrast—was evident in the application of IR when compared to FBP. Furthermore, since other imaging parameters remain identical regardless of the reconstruction algorithm, this improved image quality does not come at the expense of additional patient dose or an increased scan acquisition time for otherwise identical parameters. These improvements are expected to enhance fidelity in IGRT and ART implementation.

Accelerating compressed sensing reconstruction of subsampled radial k-space data using geometrically-derived density compensation

Objective. To accelerate compressed sensing (CS) reconstruction of subsampled radial k-space data using a geometrically-derived density compensation function (gDCF) without significant loss in image quality. Approach. We developed a theoretical framework to calculate a gDCF based on Nyquist distance along the radial and circumferential directions of a discrete polar coordinate system. Our gDCF was compared against standard DCF (e.g. ramp filter) and another commonly used DCF (modified Shepp–Logan (SL) filter). The resulting image quality produced by each DCF was quantified using normalized root-mean-square-error (NRMSE), blur metric (1 = blurriest; 0 = sharpest), and structural similarity index (SSIM; 1 = perfect match; 0 = no match) compared with the reference. For filtered backprojection (FBP) of phantom data obtained at the Nyquist sampling rate, Cartesian k-space sampling was used as the reference. For CS reconstruction of subsampled cardiac magnetic resonance imaging datasets (real-time cardiac cine data with 11 projections per frame, n = 20 patients; cardiac perfusion data with 30 projections per frame, n = 19 patients), CS reconstruction without DCF was used as the reference. Main results. The NRMSE, SSIM, and blur metrics of the phantom data were good for all DCFs, confirming that our gDCF produces uniform densities at the upper limit (Nyquist). For CS reconstruction of subsampled real-time cine and cardiac perfusion datasets, the image quality metrics (SSIM, NRMSE) were significantly (p < 0.05) higher for our gDCF than other DCFs, and the reconstruction time was significantly (p < 0.05) faster for our gDCF (reference) than no DCF (11.9%–52.9% slower), standard DCF (23.9%–57.6% slower), and modified SL filter (13.5%–34.8% slower). Significance. The proposed gDCF accelerates CS reconstruction of subsampled radial k-space data without significant loss in image quality compared with no DCF as the reference.

Impact of noise reduction schemes on quantitative accuracy of CT numbers.

In previous works, it has been demonstrated that for filtered backprojection (FBP) reconstruction-based computed tomography (CT) images, the measured CT numbers are biased and the bias level decreases with increasing radiation dose. Low-dose scans typically include noise reduction schemes to reduce noise level. The purpose of this work was to investigate the potential impact of different noise reduction schemes on the CT number bias. Three different filtration methods: Gaussian, adaptive trimmed mean (ATM), and anisotropic diffusion (AD) were implemented to reduce noise. All filters were independently applied in three different domains: raw counts, log-processed sinogram, or reconstructed image domain. A quality assurance phantom was scanned on a benchtop CT cone beam CT system, at dose levels ranging from 0.6 to 4.0mGy. The conventional FBP reconstructions were performed to reconstruct CT images for the study of CT number biases. The CT number bias of different material inserts in the phantom was then measured. To further study the overall impact of CT number bias together with the potential consequences of noise reduction schemes on both the spatial resolution and noise characteristics, the task-based detectability of a high-contrast and high spatial resolution imaging task was used as an example to assess the performance of each noise reduction scheme. To qualitatively assess the impact of these noise reduction schemes on image, an anthropomorphic head phantom was alsoscanned on the benchtop CT system and processed with the above noise reduction schemes to generate images for demonstration. Our results demonstrated the following major findings: (a) CT number bias can be significantly reduced when the noise reduction schemes are implemented in the raw counts domain; CT number bias cannot be reduced when these noise reduction schemes are implemented either in the reconstructed image domain or in the log-processed sinogram domain. (b) The extent of CT number bias reduction is dependent on both the material composition and noise reduction parameters. (c) The overall impact of the noise reduction schemes can be studied using the task-based detectability analysis framework and this framework can be used to select the appropriate parameters in each noise reduction scheme to optimize the performance for a given imaging task. Noise reduction schemes can be used to considerably reduce CT number bias when they are implemented in the raw counts domain; however, their application cannot be arbitrarily extended to either the log-processed sinogram data domain or image domain. Trade-offs between bias reduction and overall image quality must be studied for an optimal performance of a given imaging task.

Technical Note: The use of DirectDensityTM and dual-energy CT in the radiation oncology clinic.

PurposeTwo new tools available in Radiation Oncology clinics are Dual‐energy CT (DECT) and Siemens’ DirectDensity™ (DD) reconstruction algorithm, which allows scans of any kV setting to use the same calibration. This study demonstrates why DD scans should not be used in combination with DECT and quantifies the magnitude of potential errors in image quality and dose.MethodsA CatPhan 504 phantom was scanned with a dual‐pass DECT and reconstructed with many different kernels, including several DD kernels. The HU values of various inserts were measured. The RANDO ® man phantom was also scanned. Bone was contoured and then histograms of the bone HU values were analyzed for Filtered‐Backprojection (FBP) and DD reconstructions of the 80 and 140 kV scans, as well as several virtual, monoenergetic reconstructions generated from FBP and DD reconstructions. “Standard” dose distributions were calculated on several reconstructions of both phantoms for comparison.ResultsThe DD kernel overcorrected the high‐Z material inserts relative to bone, giving an excessively low relative electron density (RED). A unique artifact was observed in the high density inserts of the CatPhan in the monoenergetic scans when utilizing a DD kernel, due to the overcorrection in the DD scan of the material, especially at lower kV.ConclusionsWhile DD and DECT perform as expected when used independently, errors from their combined use were demonstrated. Dose errors from misuse of the DD kernel with DECT post‐processing were as large as 2.5%. The DECT post‐processing was without value because the HU differences between low and high energy were removed by the DD kernel. When using DD and DECT, we recommend the use of a DD reconstruction of the high energy scan for the dose calculation, and use of a FBP filter for the low and high energy scans for the DECT post‐processing.

A data-efficient method for local noise power spectrum (NPS) estimation in FDK-reconstructed 3D cone-beam CT.

For computed tomography (CT) systems in which noise is nonstationary, a local noise power spectrum (NPS) is often needed to characterize its noise property. We have previously developed a data-efficient radial NPS method to estimate the two-dimensional (2D) local NPS for filtered back projection (FBP)-reconstructed fan-beam CT utilizing the polar separability of CT NPS. In this work, we extend this method to estimate three-dimensional (3D) local NPS for feldkamp-davis-kress (FDK)-reconstructed cone-beam CT (CBCT) volumes. Starting from the 2D polar separability, we analyze the CBCT geometry and FDK image reconstruction process to derive the 3D expression of the polar separability for CBCT local NPS. With the polar separability, the 3D local NPS of CBCT can be decomposed into a 2D radial NPS shape function and a one-dimensional (1D) angular amplitude function with certain geometrical transforms. The 2D radial NPS shape function is a global function characterizing the noise correlation structure, while the 1D angular amplitude function is a local function reflecting the varying local noise amplitudes. The 3D radial local NPS method is constructed from the polar separability. We evaluate the accuracy of the 3D radial local NPS method using simulated and real CBCT data by comparing the radial local NPS estimates to a reference local NPS in terms of normalized mean squared error (NMSE) and a task-based performance metric (lesion detectability). In both simulated and physical CBCT examples, a very small NMSE (<5%) was achieved by the radial local NPS method from as few as two scans, while for the traditional local NPS method, about 20 scans were needed to reach this accuracy. The results also showed that the detectability-based system performances computed using the local NPS estimated with the NPS method developed in this work from two scans closely reflected the actual system performance. The polar separability greatly reduces the data dimensionality of the 3D CBCT local NPS. The radial local NPS method developed based on this property is shown to be capable of estimating the 3D local NPS from only two CBCT scans with acceptable accuracy. The minimum data requirement indicates the potential utility of local NPS in CBCT applications even for clinical situations.

Task-driven optimization of CT tube current modulation and regularization in model-based iterative reconstruction

Tube current modulation (TCM) is routinely adopted on diagnostic CT scanners for dose reduction. Conventional TCM strategies are generally designed for filtered-backprojection (FBP) reconstruction to satisfy simple image quality requirements based on noise. This work investigates TCM designs for model-based iterative reconstruction (MBIR) to achieve optimal imaging performance as determined by a task-based image quality metric. Additionally, regularization is an important aspect of MBIR that is jointly optimized with TCM, and includes both the regularization strength that controls overall smoothness as well as directional weights that permits control of the isotropy/anisotropy of the local noise and resolution properties. Initial investigations focus on a known imaging task at a single location in the image volume. The framework adopts Fourier and analytical approximations for fast estimation of the local noise power spectrum (NPS) and modulation transfer function (MTF)—each carrying dependencies on TCM and regularization. For the single location optimization, the local detectability index (d′) of the specific task was directly adopted as the objective function. A covariance matrix adaptation evolution strategy (CMA-ES) algorithm was employed to identify the optimal combination of imaging parameters. Evaluations of both conventional and task-driven approaches were performed in an abdomen phantom for a mid-frequency discrimination task in the kidney. Among the conventional strategies, the TCM pattern optimal for FBP using a minimum variance criterion yielded a worse task-based performance compared to an unmodulated strategy when applied to MBIR. Moreover, task-driven TCM designs for MBIR were found to have the opposite behavior from conventional designs for FBP, with greater fluence assigned to the less attenuating views of the abdomen and less fluence to the more attenuating lateral views. Such TCM patterns exaggerate the intrinsic anisotropy of the MTF and NPS as a result of the data weighting in MBIR. Directional penalty design was found to reinforce the same trend. The task-driven approaches outperform conventional approaches, with the maximum improvement in d′ of 13% given by the joint optimization of TCM and regularization. This work demonstrates that the TCM optimal for MBIR is distinct from conventional strategies proposed for FBP reconstruction and strategies optimal for FBP are suboptimal and may even reduce performance when applied to MBIR. The task-driven imaging framework offers a promising approach for optimizing acquisition and reconstruction for MBIR that can improve imaging performance and/or dose utilization beyond conventional imaging strategies.

Joint Optimization of Fluence Field Modulation and Regularization in Task-Driven Computed Tomography.

This work presents a task-driven joint optimization of fluence field modulation (FFM) and regularization in quadratic penalized-likelihood (PL) reconstruction. Conventional FFM strategies proposed for filtered-backprojection (FBP) are evaluated in the context of PL reconstruction for comparison. We present a task-driven framework that leverages prior knowledge of the patient anatomy and imaging task to identify FFM and regularization. We adopted a maxi-min objective that ensures a minimum level of detectability index (d') across sample locations in the image volume. The FFM designs were parameterized by 2D Gaussian basis functions to reduce dimensionality of the optimization and basis function coefficients were estimated using the covariance matrix adaptation evolutionary strategy (CMA-ES) algorithm. The FFM was jointly optimized with both space-invariant and spatially-varying regularization strength (β) - the former via an exhaustive search through discrete values and the latter using an alternating optimization where β was exhaustively optimized locally and interpolated to form a spatially-varying map. The optimal FFM inverts as β increases, demonstrating the importance of a joint optimization. For the task and object investigated, the optimal FFM assigns more fluence through less attenuating views, counter to conventional FFM schemes proposed for FBP. The maxi-min objective homogenizes detectability throughout the image and achieves a higher minimum detectability than conventional FFM strategies. The task-driven FFM designs found in this work are counter to conventional patterns for FBP and yield better performance in terms of the maxi-min objective, suggesting opportunities for improved image quality and/or dose reduction when model-based reconstructions are applied in conjunction with FFM.

Iterative Reconstruction as a Method for Optimisation of Computed Tomography Procedures

SummaryBackgroundComputed tomography (CT) is still commonly regarded as a method that causes a high radiation exposure. For that reason, producers intensively try to find new solutions for dose reduction while maintaining a high diagnostic value of images. One of the recent strategies focuses on CT image reconstruction. Iterative reconstruction (IR) is an alternative for filtered back projection (FBP) that is commonly used today.The aim of the article is to demonstrate and compare the effects of two IR algorithms on dose value and image details.Material/MethodsInvestigations were performed on two 128 multi-detector (MDCT) CT scanners: – iCT (Philips Healthcare with iDose4); – Definitions AS+ (Siemens Medical Solutions with SAFIRE system).The measurements involved: – image quality indicators for the CATPHAN 600 phantom; – dosimetric indicators of exposure (DLP i CTDIvol).ResultsThe signal-to-noise ratios (SNR) in the images reconstructed with IR and FBP were analysed, and the SNR(IR)/SNR(FBP) ratios were calculated and correlated with CTDIvol values. The effects of IR and FBP algorithms on low-contrast resolution were also compared in relation to CTDIvol values. The smallest diameter of supra-slice objects in the Catphan phantom were taken into consideration.Both iterative algorithms definitely improved the visibility of low-contrast objects in comparison to a standard algorithm (FBP) with similar exposure parameters.These algorithms allow an 80% reduction of the CTDIvol value while maintaining an acceptable visibility of low-contrast objects. However, the results obtained with each of the studied iterative algorithms differ.

Development and validation of a segmentation-free polyenergetic algorithm for dynamic perfusion computed tomography

Dynamic perfusion imaging can provide the morphologic details of the scanned organs as well as the dynamic information of blood perfusion. However, due to the polyenergetic property of the x-ray spectra, beam hardening effect results in undesirable artifacts and inaccurate CT values. To address this problem, this study proposes a segmentation-free polyenergetic dynamic perfusion imaging algorithm (pDP) to provide superior perfusion imaging. Dynamic perfusion usually is composed of two phases, i.e., a precontrast phase and a postcontrast phase. In the precontrast phase, the attenuation properties of diverse base materials (e.g., in a thorax perfusion exam, base materials can include lung, fat, breast, soft tissue, bone, and metal implants) can be incorporated to reconstruct artifact-free precontrast images. If patient motions are negligible or can be corrected by registration, the precontrast images can then be employed as a priori information to derive linearized iodine projections from the postcontrast images. With the linearized iodine projections, iodine perfusion maps can be reconstructed directly without the influence of various influential factors, such as iodine location, patient size, x-ray spectrum, and background tissue type. A series of simulations were conducted on a dynamic iodine calibration phantom and a dynamic anthropomorphic thorax phantom to validate the proposed algorithm. The simulations with the dynamic iodine calibration phantom showed that the proposed algorithm could effectively eliminate the beam hardening effect and enable quantitative iodine map reconstruction across various influential factors. The error range of the iodine concentration factors ([Formula: see text]) was reduced from [Formula: see text] for filtered back-projection (FBP) to [Formula: see text] for pDP. The quantitative results of the simulations with the dynamic anthropomorphic thorax phantom indicated that the maximum error of iodine concentrations can be reduced from [Formula: see text] for FBP to less than [Formula: see text] for pDP, which suggested that the proposed algorithm could not only effectively eliminate beam hardening artifacts but also significantly reduce the influence of the metal artifacts and accurately reconstruct the iodine map regardless of the influential factors. A segmentation-free polyenergetic dynamic perfusion imaging algorithm was proposed and validated via simulations. This method can accurately reconstruct artifact-free iodine maps for quantitative analyses.

Optimization of the filter parameters in 99mTc myocardial perfusion SPECT studies: the formulation of flowchart

Myocardial perfusion single photon emission computed tomography (SPECT) is typically subject to a variation in image quality due to the use of different acquisition protocols, image reconstruction parameters and image display settings by each institution. One of the principal image reconstruction parameters is the Butterworth filter cut-off frequency, a parameter strongly affecting the quality of myocardial images. The objective of this study was to formulate a flowchart for the determination of the optimal parameters of the Butterworth filter for filtered back projection (FBP), ordered subset expectation maximization (OSEM) and collimator-detector response compensation OSEM (CDR-OSEM) methods using the evaluation system of the myocardial image based on technical grounds phantom. SPECT studies were acquired for seven simulated defects where the average counts of the normal myocardial components of 45° left anterior oblique projections were approximately 10-120counts/pixel. These SPECT images were then reconstructed by FBP, OSEM and CDR-OSEM methods. Visual and quantitative assessment of short axis images were performed for the defect and normal parts. Finally, we formulated a flowchart indicating the optimal image processing procedure for SPECT images. Correlation between normal myocardial counts and the optimal cut-off frequency could be represented as a regression expression, which had high or medium coefficient of determination. We formulated the flowchart in order to optimize the image reconstruction parameters based on a comprehensive assessment, which enabled us to perform objectively processing. Furthermore, the usefulness of image reconstruction using the flowchart was demonstrated by a clinical case.

Task-Driven Tube Current Modulation and Regularization Design in Computed Tomography with Penalized-Likelihood Reconstruction.

This work applies task-driven optimization to design CT tube current modulation and directional regularization in penalized-likelihood (PL) reconstruction. The relative performance of modulation schemes commonly adopted for filtered-backprojection (FBP) reconstruction were also evaluated for PL in comparison. We adopt a task-driven imaging framework that utilizes a patient-specific anatomical model and information of the imaging task to optimize imaging performance in terms of detectability index (d'). This framework leverages a theoretical model based on implicit function theorem and Fourier approximations to predict local spatial resolution and noise characteristics of PL reconstruction as a function of the imaging parameters to be optimized. Tube current modulation was parameterized as a linear combination of Gaussian basis functions, and regularization was based on the design of (directional) pairwise penalty weights for the 8 in-plane neighboring voxels. Detectability was optimized using a covariance matrix adaptation evolutionary strategy algorithm. Task-driven designs were compared to conventional tube current modulation strategies for a Gaussian detection task in an abdomen phantom. The task-driven design yielded the best performance, improving d' by ~20% over an unmodulated acquisition. Contrary to FBP, PL reconstruction using automatic exposure control and modulation based on minimum variance (in FBP) performed worse than the unmodulated case, decreasing d' by 16% and 9%, respectively. This work shows that conventional tube current modulation schemes suitable for FBP can be suboptimal for PL reconstruction. Thus, the proposed task-driven optimization provides additional opportunities for improved imaging performance and dose reduction beyond that achievable with conventional acquisition and reconstruction.

CT reconstruction techniques for improved accuracy of lung CT airway measurement.

To determine the impact of constrained reconstruction techniques on quantitative CT (qCT) of the lung parenchyma and airways for low x-ray radiation dose. Measurement of small airways with qCT remains a challenge, especially for low x-ray dose protocols. Images of the COPDGene quality assurance phantom (CTP698, The Phantom Laboratory, Salem, NY) were obtained using a GE discovery CT750 HD scanner for helical scans at x-ray radiation dose-equivalents ranging from 1 to 4.12 mSv (12-100 mA s current-time product). Other parameters were 40 mm collimation, 0.984 pitch, 0.5 s rotation, and 0.625 mm thickness. The phantom was sandwiched between 7.5 cm thick water attenuating phantoms for a total length of 20 cm to better simulate the scatter conditions of patient scans. Image data sets were reconstructed using STANDARD (STD), DETAIL, BONE, and EDGE algorithms for filtered back projection (FBP), 100% adaptive statistical iterative reconstruction (ASIR), and Veo reconstructions. Reduced (half) display field of view (DFOV) was used to increase sampling across airway phantom structures. Inner diameter (ID), wall area percent (WA%), and wall thickness (WT) measurements of eight airway mimicking tubes in the phantom, including a 2.5 mm ID (42.6 WA%, 0.4 mm WT), 3 mm ID (49.0 WA%, 0.6 mm WT), and 6 mm ID (49.0 WA%, 1.2 mm WT) were performed with Airway Inspector (Surgical Planning Laboratory, Brigham and Women's Hospital, Boston, MA) using the phase congruency edge detection method. The average of individual measures at five central slices of the phantom was taken to reduce measurement error. WA% measures were greatly overestimated while IDs were underestimated for the smaller airways, especially for reconstructions at full DFOV (36 cm) using the STD kernel, due to poor sampling and spatial resolution (0.7 mm pixel size). Despite low radiation dose, the ID of the 6 mm ID airway was consistently measured accurately for all methods other than STD FBP. Veo reconstructions showed slight improvement over STD FBP reconstructions (4%-9% increase in accuracy). The most improved ID and WA% measures were for the smaller airways, especially for low dose scans reconstructed at half DFOV (18 cm) with the EDGE algorithm in combination with 100% ASIR to mitigate noise. Using the BONE + ASIR at half BONE technique, measures improved by a factor of 2 over STD FBP even at a quarter of the x-ray dose. The flexibility of ASIR in combination with higher frequency algorithms, such as BONE, provided the greatest accuracy for conventional and low x-ray dose relative to FBP. Veo provided more modest improvement in qCT measures, likely due to its compatibility only with the smoother STD kernel.

Missing head and color banding in low-count SPECT reconstructions.

Due to low counts in an 111In single photon emission computed tomography (SPECT) scan, a large part of the head was missing in the reconstructed images on Philips Extended Brilliance Workspace (EBW) and IntelliSpace Portal (ISP) workstations. This problem occurred for the ordered subsets expectation maximization (OSEM) algorithm with and without resolution recovery (Astonish), but not for filtered backprojection (FBP) or maximum likelihood expectation maximization (MLEM). There were also underflow problems because the images are stored as integers resulting in a loss of intensity resolution and color banding.Philips EBW2.0 and ISP5.02 workstations upscale low-count images, but the result is not always optimal, for example, in the case of low counts in one part and more counts in another part of an image. On these workstations, the missing head artefact problem could be resolved by applying a Hann pre-filter (with a cutoff at the Nyquist frequency, which only influences the filtering) in the reconstruction process. Upscaling of the projection data prior to reconstruction did not recover the head in the images, neither did limiting the reconstructed volume to the low-count part of interest.Underflow problems were partially solved by the new version 2.0 of the Philips EBW and ISP stations, although situations could arise where underflow still poses a problem. A solution for the underflow problems is to upscale the raw projection data before reconstruction. While this results in a pure upscaling of the FBP reconstruction, the effect in iterative statistical reconstruction is not only upscaling of the intensities because the assumption of Poisson statistics of the data is violated. However, the influence of this last matter seems limited.Reconstruction of studies with low counts in relevant areas should be performed with care. Reconstruction artefacts and scaling issues can easily arise.Electronic supplementary materialThe online version of this article (doi:10.1186/2197-7364-1-10) contains supplementary material, which is available to authorized users.

Appropriate parameters of the ordered-subset expectation maximization algorithm on measurement of myocardial blood flow and oxygen consumption with 11C-acetate PET

The purpose of this study was to evaluate the appropriate parameters of a filter and of subsets (S) and iterations (I) of the ordered-subset expectation maximization (OSEM) algorithm in 11C-acetate PET. A Hanning filter (HF) and a Gaussian filter (GF) were selected for filtered back-projection (FBP) and the OSEM algorithm, respectively. After evaluation of the optimal HF size, the GF size was optimized using healthy volunteers (HV). Myocardial blood flow (MBF) and oxygen consumption (k(mono)) values were calculated by combining 4S, 16S, or 28S with 2I, 4I, 6I, or 8I of the OSEM (MBF(OSEM) and k(monoOSEM), respectively) in eight HV and eight coronary artery disease (CAD) patients. These MBF(OSEM) and k(monoOSEM) values were compared with those obtained using FBP (MBF(FBP) and k(monoFBP), respectively). Optimal HF and GF (10.0GF) sizes for the FBP and OSEM algorithms, respectively, were 10.0 mm full-width resolution at half-maximum. MBF(OSEM) was changed by modifying the parameters of the OSEM algorithm. The best correlations were between MBF(FBP) and MBF(OSEM), with 28S6I and 10.0GF for HV patients and 28S8I for CAD patients. However, the MBF(OSEM) with 28S8I was significantly different from MBF(FBP) at the global myocardium in HV. The k(monoOSEM) with 28S6I was not significantly different from k(monoFBP) in HV or CAD patients. Appropriate parameters are 28S6I with a 10.0GF on the MBF(OSEM) and k(monoOSEM) measurement using 11C-acetate. Diagnostic performance will improve using noiseless, artifact-reduction images, and accurate quantitative values that are provided by the OSEM algorithm with the appropriate parameters.

PET Demonstrator for a Human Brain Scanner Based on Monolithic Detector Blocks

We have implemented and evaluated a positron emission tomography (PET) demonstrator using two monolithic detector blocks operating in coincidence with dedicated application-specific integrated circuit (ASIC) readout. Each detector is composed of a monolithic lutetium yttrium orthosilicate (LYSO) scintillator coupled to a pair of Hamamatsu S8550-02 APD arrays. The front-end electronics of this demonstrator is based on the VATA240 ASIC readout, which sums the charge provided by each row and column of the APD array. The ASIC has been characterized obtaining a noise per row or column less than 2000 electrons rms with the APD at its inputs and a good linear response in the range from 5 fC to 30 fC. We have acquired energy spectra of 22Na and 137Cs radioactive sources, achieving energy resolutions between 13.2% and 14.1% full width at half maximum (FWHM) at 511 keV. We have estimated the interaction position over the surface of the monolithic blocks using Neural Networks (NN) position determining algorithms, obtaining spatial resolutions at the detector level down to 2.1 mm FWHM. By using this detector technology and electronics we have achieved images of point sources with spatial resolutions as good as 2.1 mm FWHM for filtered back projection (FBP) reconstructions methods with single slice rebinning (SSRB). Based on the results obtained with this demonstrator, we are developing a PET insert for existing magnetic resonance imaging (MRI) equipment, to be installed in a collaborating hospital and used for clinical PET-MRI of the human brain.

Optimization of ordered subset expectation maximization reconstruction for reducing urinary bladder artifacts in single-photon emission computed tomography imaging

Bladder artifact during bone single-photon emission computed tomography (SPECT) is a common source of error. The extent and severity of bladder artifacts have been described for filtered back projection (FBP) reconstruction. Ordered subset expectation maximization (OSEM) may help to address this problem of bladder artifacts, which render up to 20% of the SPECT images unreadable. The objective of this study was to evaluate the relationship of the bladder to acetabulum ratio in guiding the choice of the number of iterations and subsets used for OSEM reconstruction, for reducing bladder artifacts found on FBP reconstruction. One hundred five patients with various indications for bone scans were selected and planar and SPECT images were acquired. The SPECT images were reconstructed with both FBP and OSEM using four different combinations of iterations and subsets. The images were given to three experienced nuclear physicians who were blinded to the diagnosis and type of reconstruction used. They then labeled images from the best to the worst after which the data were analyzed. The bladder to acetabulum ratio for each image was determined which was then correlated with the different iterations and subsets used. The study demonstrated that reconstruction using OSEM led to better lesion detectability compared to FBP in 87.62% of cases. It further demonstrated that the iterations and subsets used for reconstruction of an image correlate with the bladder to acetabulum ratio. Four iterations and 8 subsets yielded the best results in 48.5% of the images, whilst 2 iterations and 8 subsets yielded the best results in 33.8%. The number of reconstructed images which yielded the best results with 2 iterations and 8 subsets was the same as or more than those with 4 iterations and 8 subsets when the bladder/acetabulum ratio (A/B) was between 0.2 and 0.39. A ratio below 0.2 or above 0.39 supports the usage of 4 iterations and 8 subsets over 2 iterations and 8 subsets. We conclude that bladder to acetabulum ratio can be used to select the optimum number of iterations and subsets for reconstruction of bone SPECT for accurate characterization of lesions. This study also confirms that reconstruction with OSEM (vs. FBP) leads to better lesion detectability and characterization.

The image quality of FBP and MLEM reconstruction

We have developed an image quality theory for filtered back-projection (FBP) and maximum likelihood expectation maximization (MLEM) based on quality measures like signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), signal and noise power spectra and the detective quantum efficiency (DQE) concept. Analytic expressions are derived for unattenuated SPECT reconstruction with ideal collimation to obtain the fundamental performance parameters of different reconstruction algorithms. This theory is verified by measurements of signal and noise power on simulated phantoms. We demonstrate that the noise power of reconstructed images is proportional to the diameter of the object given as a number of voxels. This is analytically proven for FBP and clarified by assessment of the convergence properties for MLEM. The latter technique is shown to be superior to FBP in terms of a noise level at least two times lower. The free choice of MLEM reconstruction parameters and correction for physical effects in the image acquisition enables quantitative evaluation of SPECT and PET images.

Spatial resolution of the Inveon small-animal PET scanner for the entire field of view

In high throughput small-animal positron emission tomography (PET) studies, it can be advantageous to scan more than one animal at the same time. To optimally position these animals in the scanner, both sensitivity and spatial resolution have to be known across the entire field of view (FOV). The Siemens Inveon small-animal PET scanner has been characterized according to NEMA NU 4-2008 standards, but the spatial resolution has only been determined in a small portion of the FOV and only for filtered backprojection (FBP). In this study, the spatial resolution in the entire FOV was determined for the Inveon using FBP, ordered subset expectation maximization in 2 dimensions (OSEM2D) and maximum a posteriori (MAP) reconstructions. OSEM2D produced the best spatial resolution in the axial center of the FOV, but MAP outperformed the two other reconstruction algorithms in terms of uniformity of the resolution.