Calculated Attenuation Correction Research Articles

Accurate Transmission-Less Attenuation Correction Method for Amyloid-β Brain PET Using Deep Neural Network

The lack of physically measured attenuation maps (μ-maps) for attenuation and scatter correction is an important technical challenge in brain-dedicated stand-alone positron emission tomography (PET) scanners. The accuracy of the calculated attenuation correction is limited by the nonuniformity of tissue composition due to pathologic conditions and the complex structure of facial bones. The aim of this study is to develop an accurate transmission-less attenuation correction method for amyloid-β (Aβ) brain PET studies. We investigated the validity of a deep convolutional neural network trained to produce a CT-derived μ-map (μ-CT) from simultaneously reconstructed activity and attenuation maps using the MLAA (maximum likelihood reconstruction of activity and attenuation) algorithm for Aβ brain PET. The performance of three different structures of U-net models (2D, 2.5D, and 3D) were compared. The U-net models generated less noisy and more uniform μ-maps than MLAA μ-maps. Among the three different U-net models, the patch-based 3D U-net model reduced noise and cross-talk artifacts more effectively. The Dice similarity coefficients between the μ-map generated using 3D U-net and μ-CT in bone and air segments were 0.83 and 0.67. All three U-net models showed better voxel-wise correlation of the μ-maps compared to MLAA. The patch-based 3D U-net model was the best. While the uptake value of MLAA yielded a high percentage error of 20% or more, the uptake value of 3D U-nets yielded the lowest percentage error within 5%. The proposed deep learning approach that requires no transmission data, anatomic image, or atlas/template for PET attenuation correction remarkably enhanced the quantitative accuracy of the simultaneously estimated MLAA μ-maps from Aβ brain PET.

Measurement and analysis of patient attenuation correction factor during radioiodine therapy.

The calculated dose rate from the radioiodine therapy patient should normally include a factor accounting for the attenuation and scatter of patient body tissues. The attenuation factor is currently neglected, and not applied in operational radiation protection. Realistic estimation of radiation dose rate levels from radioiodine therapy patients when properly performed will reduce operational cost and optimise institutional radiation protection practice. In this work, the existence of a patient body tissue attenuation factor is verified by comparing the dose rates measured from the radioiodine capsules immediately before administration with those measured from the patient immediately after administration. The correlation between the factors suspected to influence the patient body tissue attenuation and the measured dose rates from the patient normalised per unit activity is statistically analysed. The calculated attenuation correction factor based on authors' measurements was (0.55 ± 0.17). The measured dose rate per unit of radioactivity from the patient showed a negative correlation with their body mass index.

Estimation of patient attenuation factor for iodine-131 based on direct dose rate measurements from radioiodine therapy patients.

The aim of the study was to measure the actual dose at 1 m from the patients per unit activity with the aim of providing a more accurate prediction of the dose levels around radioiodine patients in the hospital, as well as to compare our results with the literature. In this work the demonstration of a patient body tissue attenuation factor is verified by comparing the dose rates measured from the patients with those measured from the unshielded radioiodine capsules immediately after administration of the radioactivity. The normalized dose rate per unit activity is therefore proposed as an operational quantity that can be used to predict exposure rates to staff and patients' relatives. The average dose rate measured from our patient per unit activity was 38.4±11.8 μSv/h/GBq. The calculated attenuation correction factor based on our measurements was 0.55±0.17. The calculated dose rate from a radioiodine therapy patient should normally include a factor accounting for patient body tissue attenuation and scatter. The attenuation factor is currently neglected and not applied in operational radiation protection. Realistic estimation of radiation dose levels from radioiodine therapy patients when properly performed will reduce the operational cost and optimize institutional radiation protection practice. It is recommended to include patient attenuation factors in risk assessment exercises - in particular, when accurate estimates of total effective doses to exposed individuals are required when direct measurements are not possible. The information provided about patient attenuation might benefit radiation protection specialists and regulators.

Feasibility of Template-Guided Attenuation Correction in Cat Brain PET Imaging

Attenuation correction (AC) is important in quantitative positron emission tomography (PET) imaging of medium-sized animals such as the cat. However, additional time for transmission (TX) scanning and tracer uptake is required in PET studies with animal-dedicated PET scanners because post-injection TX scanning is not available in these systems. The aim of this study was to validate a template-guided AC (TGAC) method that does not require TX PET data for AC in cat 2-deoxy-2-[F-(18)fluoro-D-glucose (FDG) brain PET imaging. PET scans were acquired using a microPET Focus 120 scanner. TX data were obtained using a (68)Ge point source before the injection of FDG. To generate the attention map (mu-map) template for the TGAC, a target image of emission (EM) PET was selected, and spatial normalization parameters of individual EM data onto the target were reapplied to the corresponding mu-maps. The inverse transformations of the mu-map template into the individual spaces were performed, and the transformed template was forward projected to generate the AC factor. The TGAC method was compared with measured AC (MAC) and calculated AC (CAC) methods using region of interest (ROI) and SPM analyses. The ROI analysis showed that the activity of the TGAC EM PET images strongly correlated with those of the MAC data (y = 0.98x + 0.01, R(2) = 0.96). In addition, no significant difference was observed in the SPM analysis. By contrast, the CAC showed a significantly higher uptake in the deep gray regions compared to the MAC (corrected P < 0.05). The ROI correlation with MAC was worse than with the TGAC (R(2) = 0.84). In SPM analysis for the voxel-wise group comparisons between before and after the induction of deafness, only the TGAC showed equivalent results with the MAC. The TGAC was reliable in cat FDG brain PET studies in terms of compatibility with the MAC method. The TGAC might be a useful option for increasing study throughput and decreasing the probability of subject movement. In addition, it might reduce the possible biological effects of long-term anesthesia on the cat brain in investigations using animal-dedicated PET scanners.

Impact of metal artefacts due to EEG electrodes in brain PET/CT imaging

The goal of this study is to investigate the impact of electroencephalogram (EEG) electrodes on the visual quality and quantification of 18F-FDG PET images in neurological PET/CT examinations. For this purpose, the scans of 20 epilepsy patients with EEG monitoring were used. The CT data were reconstructed with filtered backprojection (FBP) and with a metal artefact reduction (MAR) algorithm. Both data sets were used for CT-based attenuation correction (AC) of the PET data. Also, a calculated AC (CALC) technique was considered. A volume of interest (VOI)-based analysis and a voxel-based quantitative analysis were performed to compare the different AC methods. Images were also evaluated visually by two observers. It was shown with simulations and phantom measurements that from the considered AC methods, the MAR-AC can be used as the reference in this setting. The visual assessment of PET images showed local hot spots outside the brain corresponding to the locations of the electrodes when using FBP-AC. In the brain, no abnormalities were observed. The quantitative analysis showed a very good correlation between PET-FBP-AC and PET-MAR-AC, with a statistically significant positive bias in the PET-FBP-AC images of about 5–7% in most brain voxels. There was also good correlation between PET-CALC-AC and PET-MAR-AC, but in the PET-CALC-AC images, regions with both a significant positive and negative bias were observed. EEG electrodes give rise to local hot spots outside the brain and a positive quantification bias in the brain. However, when diagnosis is made by mere visual assessment, the presence of EEG electrodes does not seem to alter the diagnosis. When quantification is performed, the bias becomes an issue especially when comparing brain images with and without EEG monitoring.

Content-Specific Delusions From Right Caudate Lacunar Stroke: Association with Prefrontal Hypometabolism

Content-Specific Delusions From Right Caudate Lacunar Stroke: Association with Prefrontal Hypometabolism

The Usefulness of Calculated Attenuation Correction in the Evaluation of Metallic Artifacts on Brain PET/CT Imaging

A known limitation of CT-based attenuation correction (CTAC) is the creation of artifacts that mimic increased FDG uptake in the location of metallic prostheses. Calculated attenuation correction (CAC) uses a constant value for the attenuation coefficient of brain tissue at 511 keV and corrects the reconstructed data for attenuation effects. Comparison of CTAC and NAC PET images should always be the first step when evaluating possible metallic artifacts on CTAC PET brain images. If the etiology responsible for the increased FDG uptake remains unclear, reprocessing using CAC is an additional useful tool to differentiate between artifactual and true FDG uptake.

Validation of quantitative accuracy of the post-injection transmission-based and transmissionless attenuation correction techniques in neurological FDG-PET

We evaluated the quantitative accuracy of the post-injection transmission-based segmented attenuation correction (SAC) technique and transmissionless calculated attenuation correction (CAC) technique in both 2D and 3D scanning for the brain, and compared the results with those obtained using the pre-injection transmission-based measured attenuation correction (MAC) technique, which is generally accepted as the 'gold standard'. We examined [18F]fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) on both a cylindrical phantom and in 10 patients with epilepsy. The statistical analyses were performed using both the region of interest (ROI) method and statistical parametric mapping (SPM). In the ROI analysis, [18F] activity concentration values obtained by the SAC technique were well correlated with those obtained by the MAC technique (2D, R2=0.94; 3D, R2=0.98; P<0.001), although these values were underestimated over the entire brain. The CAC technique was also found to have significant correlation with the MAC technique (2D, R2=0.84; 3D, R2=0.86; P<0.001), but this technique showed apparent overestimation or underestimation in several parts of the brain. In the SPM analysis, there were no significant differences between the SAC and MAC technique, while those values obtained by the CAC technique were significantly lower in the parieto-occipital region and higher in the lower frontal region (P<0.001, uncorrected). The SAC technique was superior to the CAC technique in both 2D and 3D scanning, although we found that both the SAC and CAC techniques had some problems in quantitative evaluation. We considered that the SAC technique may yield adequate qualitative measurements of the [18F] activity concentration value after global normalization.

Assessment of the impact of model-based scatter correction on [18F]-FDG 3D brain PET in healthy subjects using statistical parametric mapping

It is recognized that scatter correction can supply more accurate absolute quantification, and that iterative reconstruction results in better noise properties and significantly reduces streak artefacts; however, it is not entirely clear whether they produce significant changes in [18F]-FDG distribution of reconstructed 3D brain PET images relative to not scatter corrected images and analytic reconstruction procedures. The current study assesses the effect of model-based scatter correction using the single-scatter simulation algorithm and iterative reconstruction in 3D brain PET studies, using statistical parametric mapping (SPM) analysis. The study population consisted of 14 healthy volunteers (6 males, 8 females; age 63–80 years). PET images were reconstructed using an analytic 3DRP reprojection algorithm with (SC) and without explicit scatter correction (NSC), as well as using an iterative ordered subset–expectation maximization (OSEM) algorithm. Calculated attenuation correction was performed assuming uniform attenuation (μ = 0.096 cm−1) for brain tissues when data are precorrected for scatter. The broad-beam attenuation coefficient (μ = 0.06 cm−1) determined from phantom studies was applied to NSC images. The images were coregistered and normalized using the default [15O]-H2O template supplied with SPM99 and an [18F]-FDG template. A t statistic image for the contrast condition effect was then constructed. The contrast comparing SC to NSC images suggest that regional brain metabolic activity decreases significantly in the frontal gyri, in addition to the middle temporal and postcentral gyri. On the other hand, activity increases in the cerebellum, thalamus, insula, brainstem, temporal lobe, and the frontal cortex. No significant changes were detected when comparing images reconstructed using analytic and iterative algorithms. It is concluded that, for some cerebral areas, significant differences in [18F]-FDG distribution arise when images are reconstructed with and without explicit SC. This needs to be considered when interpreting [18F]-FDG 3D brain PET images after applying SC.

Attenuation correction of PET activation studies in the presenceof task-related motion

Motion-induced misalignment between transmission and emission scans can result in erroneous estimation of regional tissue activity concentrations. If this motion is of a random nature, mismatch between transmission and emission scans is likely to result in diminished signal-to-noise ratios. In the case of task-related motion, however, corresponding systematic reconstruction artefacts may lead to false-positive or false-negative results. The purpose of the present study was to investigate whether implementation of an image registration (IR) method, which allows for motion-corrected attenuation correction, would improve accuracy of H 2 15O PET activation studies. To evaluate the performance of this method, phantom studies as well as studies in human subjects were performed. Results were compared with three alternative methods: standard attenuation correction without motion correction, calculated attenuation correction, and no attenuation correction. The phantom measurements showed that, for quantitative assessment of regional activity concentrations, the IR method was superior to the other attenuation correction methods. In a single-subject study with intentional task-related motion during a visual stimulation paradigm, false-positive results, obtained with the standard attenuation correction method, disappeared after attenuation correction using the IR method. Finally, a group analysis of 11 patients indicated that an increase in signal-to-noise ratio was obtained with the IR method. Therefore, in our view, the IR method should be considered as a first choice for attenuation correction in PET activation studies.

Interscan Displacement-Induced Variance in PET Activation Data Is Excluded by a Scan-Specific Attenuation Correction

In PET activation studies, linear changes in regional cerebral blood flow may be caused by subject interscan displacements rather than by changes in cognitive state. The aim of this study was to investigate the impact of these artifacts and to assess whether they can be removed by applying a scan-specific calculated attenuation correction (CAC) instead of the default measured attenuation correction (MAC). Two independent data sets were analyzed, one with large (data I) and one with small (data II) interscan displacements. After attenuation correction (CAC or MAC), data were analyzed using SPM99. Interscan displacement parameters (IDP), obtained during scan realignment, were included as additional regressors in the General Linear Model and their impact was assessed by variance statistics revealing the affected brain volume. For data I, this volume reduced dramatically from 579 to 12 cm 3 (approximately 50-fold) at P uncorr ≤ 0.001 and from 100 to 0 cm 3 at P corr ≤ 0.05 when CAC was applied instead of MAC. Surprisingly, for data II, applying CAC instead of MAC still resulted in a substantial (approximately 10-fold) reduction of the affected volume from 23 to 2 cm 3 at P uncorr ≤ 0.001. We conclude that interscan displacement-induced variance can be prevented by applying a (realigned attenuation correction scan (e.g., CAC). With MAC data, introducing IDP covariates is not an alternative since they model only this variance. Even in data with minor interscan displacements, applying a (realigned attenuation correction method (e.g., CAC) is superior to a nonaligned MAC with IDP covariates.

Signal-to-noise ratio in neuro activation PET studies

It has become commonplace to compare scanner sensitivity characteristics by comparing noise equivalent count rate curves. However, because a 20-cm diameter uniform phantom is drastically different from a human brain, these curves give misleading information when planning a neuro activation PET experiment. Signal-to-noise ratio (SNR) calculations have been performed using measured data (Siemens 921 scanner) from the three-dimensional (3-D) Hoffman brain phantom for the purpose of determining the optimal injection and scanning protocol for [ (15)O] labeled activation experiments. Region of interest (ROI) values along with the variance due to prompt (trues plus randoms) and random events were determined for various regions and radioactivity concentrations. Calculated attenuation correction was used throughout. Scatter correction was not used when calculating the SNR in activation studies because the number of scattered events is almost identical in each data acquisition and hence cancels. The authors results indicate that randoms correction should not be performed and that rather than being limited by the scanner capabilities, neuro activation experiments are limited by the amount of radioactivity that can be injected and the length of time the patient can stay in the scanner.

Implementation and evaluation of a calculated attenuation correction for PET

A highly reliable and consistent method for automated calculated attenuation correction for PET (positron emission tomography) was developed. The method was tested in nearly 300 brain studies and shown to accurately find the object outline, regardless of the tracer used and the activity distribution. The Bergstrom method of determining the contour of a convex object from its emission projection sinogram was implemented to eliminate the initial reconstruction and ellipse drawing steps. The efficacy of this method is highly dependent on a robust algorithm for determining the border between the activity uptake and scatter background. An effective, tracer-distribution-independent, edge detection algorithm is presented that has primarily been applied to brain studies using a variety of tracers (/sup 15/O-H/sub 2/O, /sup 18/F-FGD, /sup 18/F-FDOPA), in extreme uptake distributions, and tested on three different PET systems. The algorithm has also been extended for abdominal PET scans and tested in a limited series of kidney flow studies using /sup 13/N-NH/sub 3/. >