Algorithm Improves Image Quality Research Articles

Application of a Deep Learning-Based Contrast-Boosting Algorithm to Low-Dose Computed Tomography Pulmonary Angiography With Reduced Iodine Load.

The aim of this study was to assess the effectiveness of a deep learning-based image contrast-boosting algorithm by enhancing the image quality of low-dose computed tomography pulmonary angiography at reduced iodine load. This study included 179 patients who underwent low-dose computed tomography pulmonary angiography with a reduced iodine load using 64 mL of a 1:1 mixture of contrast medium from January 1 to June 30, 2023. For single-energy computed tomography, the noise index was set at 15.4 to maintain a CTDIvol of <2 mGy at 80 kVp, and for dual-energy computed tomography, fast kV-switching between 80 and 140 kVp was employed with a fixed tube current of 145 mA. Images were reconstructed by 50% adaptive statistical iterative reconstruction (AR50) and a commercially available deep learning image reconstruction (TrueFidelity) package at a high strength level (TFH). In addition, AR50 images were further processed using a deep learning-based contrast-boosting algorithm (AR50-CB). Quantitative and qualitative image qualities and numbers of involved vessels with thrombus at each pulmonary artery level were compared in the 3 image types using the Friedman test and Wilcoxon signed rank test. Five hundred thirty-seven reconstructed image datasets of 179 patients were analyzed. Quantitative image analysis showed AR50-CB (30.8 ± 10.0 and 28.1 ± 9.6, respectively) had significantly higher signal-to-noise ratio and contrast-to-noise ratio values than AR50 (20.2 ± 6.2 and 17.8 ± 6.2, respectively) (P < 0.001) or TFH (28.3 ± 8.3 and 24.9 ± 8.1, respectively) (P < 0.001). Qualitative image analysis showed that contrast enhancement and noise scores of AR50-CB were significantly greater than those of AR50 (P < 0.001) and that AR50-CB enhancement scores were significantly higher than TFH enhancement scores (P < 0.001). The number of subsegmental pulmonary arteries affected by thrombus detected was significantly greater for AR50-CB (30 for AR50, 30 for TFH, and 55 for AR50-CB, P < 0.001). The use of a deep learning-based contrast-boosting algorithm improved image quality in terms of signal-to-noise ratio and contrast-to-noise ratio values and the detection of thrombi in subsegmental pulmonary arteries.

Comparative study of image quality and radiation dose in thoracic-abdominal-pelvic CT Enhancement with different tube voltages and reconstruction algorithms

ObjectivesTo evaluate the image quality and radiation dose in thoracic-abdominal-pelvic computed tomography enhancement using ATVM (Automatic Tube Voltage Modulation)coupled with AIIR versus routine tube voltage combined with Karl-3D IR. The optimal noise level for the AIIR in thoracic-abdominal-pelvic CT enhancement was also determined. MethodsGroup A was scanned using ATVM, and images were reconstructed using AIIR with 1–5 noise levels. Group B was scanned and reconstructed using the fixed tube voltage (120 kVp) combined with Karl 3D level 5 IR. The image quality of the reconstructed images of AIIR with 1–5 noise levels were compared and the best image reconstruction noise level for AIIR was preferred. Image quality and radiation dose were statistically analyzed for group A (best image reconstruction noise level for AIIR) and group B. ResultsAIIR level 3 is the optimal noise level for CT-enhanced image reconstruction of the thorax, abdomen, and pelvis. The mean SNR, CNR, and subjective evaluation of AIIR level 3 Group A images were better than those of Karl 3D level 5 Group B images (p < 0.05). The mean SSDE and the mean ED of the AIIR Group A patients were reduced by 46% and 41%, respectively, compared with those of Group B. ConclusionsATVM technology combined with the AIIR algorithm improved image quality and reduced patient radiation dose in thoracic-abdominal-pelvic CT scans. The optimal noise level for the reconstruction of high-quality arterial and venous-phase images was AIIR level 3.

Aberration correction in 2D echocardiography.

An aberration correction algorithm has been implemented and demonstrated in an echocardiographic clinical trial using two-dimensional (2D) imaging. The method estimates and compensates arrival time errors between different sub-aperture processor (SAP) signals in a matrix array probe. Five standard views of channel data cineloops were recorded from 22 patients (11 male and 11 female) resulting in a total of 116 cineloops. The channel data were processed with and without the aberration correction algorithm, allowing for side-by-side comparison of images processed from the same channel data cineloops. The aberration correction algorithm improved image quality, as quantified by a coherence metric, in all 7,380 processed frames. In a blinded and left-right-randomized side-by-side evaluation, four cardiologists (two experienced and two in training) preferred the aberration corrected cineloops in 97% of the cases. The clinicians reported that the corrected cineloops appeared sharper with better contrast and less noise. Many structures like valve leaflets, chordae, endocardium, and endocardial borders appeared narrower and more clearly defined in the aberration corrected images. An important finding is that aberration correction improves contrast between the endocardium and ventricle cavities for every processed image. The gain difference was confirmed by the cardiologists in their feedback and quantified with a median global gain difference estimate between the aberration-corrected and non-corrected images of 1.2 dB. The study shows the potential value of aberration correction in clinical echocardiography. Systematic improvement of images acquired with state-of-art equipment was observed both with quantitative metrics of image quality and clinician preference.

Optimization of Bayesian penalized likelihood reconstruction for 68Ga-prostate-specific membrane antigen-11 PET/computed tomography.

The objective of this study is to determine the optimal β value for clinical use in digital 68Ga-prostate-specific membrane antigen (PSMA-11) PET/computed tomography (CT) imaging. 68Ga PSMA PET/CT of 21 patients with prostate cancer were reconstructed using block-sequential regularized expectation maximization (β value of 400-1600) and ordered subsets expectation maximization. Nine independent blinded readers evaluated each reconstruction for overall image quality, noise level and lesion detectability. Maximum standardized uptake value (SUVmax) of the most intense lesion, liver SUVmean and liver SUVSD were recorded. Lesions were then subdivided according to uptake and size; the SUVmax of these lesions were analyzed. There is a statistically significant correlation between improvement in image quality and β value, with the best being β 1400. This trend was also seen in image noise (P < 0.001), with the least image noise reported with β 1400. Lesion detectability was not significantly different between the different β values (P = 0.6452). There was no statistically significant difference in SUVmax of the most intense lesion (P = 0.9966) and SUVmean of liver background between the different β values (P = 0.9999); however, the SUVSD of the liver background showed a clear trend, with the lowest with β 1400 (P = 0.0008). There was a decreasing trend observed in SUVmax when β values increased from 800 to 1400 for all four subgroups, and this decrease was greatest in small and low uptake lesions. Bayesian penalized likelihood reconstruction algorithms improve image quality without affecting lesion detectability. A β value of 1400 is optimal.

Post-hoc motion correction for coronary computed tomography angiography without additional radiation dose - Improved image quality and interpretability for “free”

AbstractObjectiveTo evaluate the impact of a motion-correction (MC) algorithm, applicable post-hoc and not dependent on extended padding, on the image quality and interpretability of coronary computed tomography angiography (CCTA).MethodsNinety consecutive patients undergoing CCTA on a latest-generation 256-slice CT device were prospectively included. CCTA was performed with prospective electrocardiogram-triggering and the shortest possible acquisition window (without padding) at 75% of the R-R-interval. All datasets were reconstructed without and with MC of the coronaries. The latter exploits the minimal padding inherent in cardiac CT scans with this device due to data acquisition also during the short time interval needed for the tube to reach target currents and voltage (“free” multiphase). Two blinded readers independently assessed image quality on a 4-point Likert scale for all segments.ResultsA total of 1,030 coronary segments were evaluated. Application of MC both with automatic and manual coronary centerline tracking resulted in a significant improvement in image quality as compared to the standard reconstruction without MC (mean Likert score 3.67 [3.50;3.81] vs 3.58 [3.40;3.73], P = 0.005, and 3.7 [3.55;3.82] vs 3.58 [3.40;3.73], P &lt; 0.001, respectively). Furthermore, MC significantly reduced the proportion of non-evaluable segments and patients with at least one non-evaluable coronary segment from 2% to as low as 0.3%, and from 14% to as low as 3%. Reduction of motion artifacts was predominantly observed in the right coronary artery.ConclusionsA post-hoc device-specific MC algorithm improves image quality and interpretability of prospectively electrocardiogram-triggered CCTA and reduces the proportion of non-evaluable scans without any additional radiation dose exposure.

The impact of deep learning reconstruction on image quality and coronary CT angiography-derived fractional flow reserve values.

To explore the impact of deep learning reconstruction (DLR) on image quality and machine learning-based coronary CT angiography (CTA)-derived fractional flow reserve (CT-FFRML) values. Thirty-three consecutive patients with known or suspected coronary artery disease who underwent coronary CTA and subsequent invasive coronary angiography were enrolled. DLR was compared with filtered back projection (FBP), statistical-based iterative reconstruction (SBIR), model-based iterative reconstruction (MBIR) Cardiac, and MBIR Cardiac sharp for objective image qualities of coronary CTA. Invasive fractional flow reserve (FFR) and quantitative flow ratio (QFR) were used as the reference standards. The diagnostic performances of different reconstruction approach-based CT-FFRML were calculated. A total of 182 lesions in 33 patients were enrolled for analysis. The image quality of DLR was superior to the others. There were no significant differences in the CT-FFRML values among these five approaches (all p > 0.05). Of the 182 lesions, 17 had invasive FFR results, and 70 had QFR results. Using FFR as a reference, MBIR Cardiac, MBIR Cardiac sharp, and DLR achieved equal diagnostic performance, slightly higher than the other reconstruction approaches (MBIR Cardiac, MBIR Cardiac sharp, and DLR: AUC = 0.82, FBP and AIDR: AUC = 0.78, all p > 0.05). Using QFR as a reference, the AUCs of FBP, SBIR, MBIR Cardiac, MBIR Cardiac sharp, and DLR were 0.83, 0.81, 0.86, 0.84, and 0.83, respectively (all p > 0.05). Our study showed that the DLR algorithm improved image quality, but there were no significant differences in the CT-FFRML values and diagnostic performance among different reconstruction approaches. • Deep learning-based image reconstruction (DLR) improves the image quality of coronary CTA. • CT-FFRML values and diagnostic performance of DLR revealed no significant differences compared to other reconstruction approaches.

Feasibility of MMWF noise reduction algorithm in brain SPECT images according to various reconstruction methods: A phantom study

In order to improve image quality in single-photon emission computed tomography (SPECT)/computed tomography (CT), various studies focusing on reconstruction methods and noise reduction algorithms have been conducted. Therefore, the purpose of this study was to evaluate the efficiency of the filtered back projection (FBP) and ordered subset expectation maximization (OSEM) reconstruction methods, and to evaluate image quality after applying the median modified Wiener filter (MMWF) as a complementary noise reduction filter using 3 × 3 and 5 × 5 mask sizes. The 3D Hoffman brain phantom was used, and the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were evaluated in regions of interest for quantitative analysis. Based on the results, the image quality obtained by the OSEM reconstruction method was 37.7% and 25.9% higher than that of the FBP reconstruction method regarding SNR and CNR, respectively. In addition, the average SNR and CNR values obtained by applying the MMWF algorithm were 35.9% and 17.1% higher than those without the MMWF algorithm. Therefore, our results indicate that the OSEM reconstruction method is more efficient than the FBP reconstruction method, and that the MMWF algorithm improves image quality in SPECT images, regardless of which reconstruction method is used.

A feasibility study of realizing low-dose abdominal CT using deep learning image reconstruction algorithm.

To explore the feasibility of achieving diagnostic images in low-dose abdominal CT using a Deep Learning Image Reconstruction (DLIR) algorithm. Prospectively enrolled 47 patients requiring contrast-enhanced abdominal CT scans. The late-arterial phase scan was added and acquired using lower-dose mode (tube current range, 175-545 mA; 80 kVp for patients with BMI ≤24 kg/m2 and 100 kVp for patients with BMI > 24 kg/m2) and reconstructed with DLIR at medium setting (DLIR-M) and high setting (DLIR-H), ASIR-V at 0% (FBP), 40% and 80% strength. Both the quantitative measurement and qualitative analysis of the five types of reconstruction methods were compared. In addition, radiation dose and image quality between the early-arterial phase ASIR-V images using standard-dose and the late-arterial phase DLIR images using low-dose were compared. For the late-arterial phase, all five reconstructions had similar CT value (P > 0.05). DLIR-H, DLIR-M and ASIR-V80% images significantly reduced the image noise and improved the image contrast noise ratio, compared with the standard ASIR-V40% images (P < 0.05). ASIR-V80% images had undesirable image characteristics with obvious "waxy" artifacts, while DLIR-H images maintained high spatial resolution and had the highest subjective image quality. Compared with the early-arterial scans, the late-arterial phase scans significantly reduced the radiation dose (P < 0.05), while the DLIR-H images exhibited lower image noise and good display of the specific image details of lesions. DLIR algorithm improves image quality under low-dose scan condition and may be used to reduce the radiation dose without adversely affecting the image quality.

Improving Image Quality and Reducing Radiation Dose for Pediatric CT by Using Deep Learning Reconstruction.

Background CT deep learning reconstruction (DLR) algorithms have been developed to remove image noise. How the DLR affects image quality and radiation dose reduction has yet to be fully investigated. Purpose To investigate a DLR algorithm's dose reduction and image quality improvement for pediatric CT. Materials and Methods DLR was compared with filtered back projection (FBP), statistical-based iterative reconstruction (SBIR), and model-based iterative reconstruction (MBIR) in a retrospective study by using data from CT examinations of pediatric patients (February to December 2018). A comparison of object detectability for 15 objects (diameter, 0.5-10 mm) at four contrast difference levels (50, 150, 250, and 350 HU) was performed by using a non-prewhitening-matched mathematical observer model with eye filter (d'NPWE), task transfer function, and noise power spectrum analysis. Object detectability was assessed by using area under the curve analysis. Three pediatric radiologists performed an observer study to assess anatomic structures with low object-to-background signal and contrast to noise in the azygos vein, right hepatic vein, common bile duct, and superior mesenteric artery. Observers rated from 1 to 10 (worst to best) for edge definition, quantum noise level, and object conspicuity. Analysis of variance and Tukey honest significant difference post hoc tests were used to analyze differences between reconstruction algorithms. Results Images from 19 patients (mean age, 11 years ± 5 [standard deviation]; 10 female patients) were evaluated. Compared with FBP, SBIR, and MBIR, DLR demonstrated improved object detectability by 51% (16.5 of 10.9), 18% (16.5 of 13.9), and 11% (16.5 of 14.8), respectively. DLR reduced image noise without noise texture effects seen with MBIR. Radiologist ratings were 7 ± 1 (DLR), 6.2 ± 1 (MBIR), 6.2 ± 1 (SBIR), and 4.6 ± 1 (FBP); two-way analysis of variance showed a difference on the basis of reconstruction type (P < .001). Radiologists consistently preferred DLR images (intraclass correlation coefficient, 0.89; 95% CI: 0.83, 0.93). DLR demonstrated 52% (1 of 2.1) greater dose reduction than SBIR. Conclusion The DLR algorithm improved image quality and dose reduction without sacrificing noise texture and spatial resolution. © RSNA, 2020 Online supplemental material is available for this article.

UCT‐GAN: underwater image colour transfer generative adversarial network

Underwater image enhancement algorithms improve image quality and indirectly enhance underwater visibility. Although many underwater image enhancement neural networks have been proposed, they require large amounts of data. To reduce the amount of data required while providing better image enhancement, this study proposes an underwater image colour transfer generative adversarial network (UCT-GAN). The authors first design a non-linear mapping function to generate colour cast images according to original images. Then, the authors utilise these image pairs (i.e. colour cast images and corresponding original images) to guide the UCT-GAN in learning the inverse function of the designed non-linear mapping function. Finally, colour cast images are restored via the inverse function. A data augmentation method based on Poisson fusion and block combination is also proposed to overcome the problem of requiring a large amount of training data. Moreover, the authors extend UCT-GAN into a multi-class colour transfer network to achieve an array of underwater image enhancements. Experimental results indicate that the proposed UCT-GAN can more effectively resolve underwater image colour cast compared to existing algorithms.

The Effect of Various β Values on Image Quality and Semiquantitative Measurements in 68Ga-RM2 and 68Ga-PSMA-11 PET/MRI Images Reconstructed With a Block Sequential Regularized Expectation Maximization Algorithm.

To compare the block sequential regularized expectation maximization (BSREM) algorithm with the ordered subsets expectation maximization (OSEM) algorithm and to evaluate how different penalty factors (b values) influence image quality and SUV measurements. We analyzed data from 78 prostate cancer patients who underwent Ga-RM2 (n = 42) or Ga-prostate-specific membrane antigen (PSMA)-11 (n = 36) PET/MRI. The raw PET data were retrospectively reconstructed using both time-of-flight (TOF)-BSREM with b values of 250, 350, 500, 750, and 1000 and TOF-OSEM. Each reconstruction was reviewed independently by 3 nuclear medicine physicians and scored qualitatively using a Likert scale (1 = poor, 5 = excellent quality). SUV measurements were analyzed as well. Fifty-seven lesions were detected (21 on Ga-RM2 and 36 on Ga-PSMA-11 PET/MRI); SUVmax decreased with the increase of β values for both tracers. Background noise (SUVsd) decreased with increasing of β values for both tracers. The mean ± SD scores for Ga-RM2 PET images were 2.4 ± 0.5 for b = 250 reconstructions, 3.2 ± 0.6 for b = 350, 4 ± 0.6 for b = 500, 4.5 ± 0.5 for b = 750, 4.4 ± 0.7 for b = 1000, and 3.4 ± 0.6 for TOF-OSEM. The mean ± SD scores for Ga-PSMA-11 PET images were 3.2 ± 0.8 for b = 250 reconstructions, 4.1 ± 0.8 for b = 350, 4.7 ± 0.6 for b = 500, 4.8 ± 0.4 for b = 750, 4.7 ± 0.6 for b = 1000, and 3.8 ± 0.5 for TOF-OSEM. Time-of-flight-BSREM algorithm improves image quality. Different b values should be used for different Ga-labeled radiopharmaceuticals such as those targeting GRPR and PSMA receptors. Once selected, the same b value should be consistently used because SUVmax measurements differ with different b values.

Comparison of NEMA characterizations for Discovery MI and Discovery MI-DR TOF PET/CT systems at different sites and with other commercial PET/CT systems

BackgroundThis article compares the physical performance of the 4-ring digital Discovery MI (DMI) and PMT-based Discovery MI-DR (DMI-DR) PET/CT systems. Physical performance was assessed according to the NEMA NU 2-2012 standards. Performance measures included spatial resolution, image quality, scatter fraction and count rate performance, and sensitivity. Energy and timing resolutions were also measured. Published DMI and DMI-DR performance studies from other centers are reviewed and compared.Results4-ring DMI spatial resolution at 1-cm radial offset in the radial, tangential and axial directions was 4.62, 4.18 and 4.57 mm, respectively, compared with the DMI-DR system values of 4.58, 4.52, and 5.31 mm. Measured sensitivity was 13.3 kcps/MBq at the center of the FOV and 13.4 kcps/MBq 10 cm off-center for the SiPM-based DMI system. DMI-DR system sensitivity was 6.3 kcps/MBq at the center of the FOV and 6.8 kcps/MBq at 10 cm off-center. DMI measured noise equivalent count rate peak was 175.6 kcps at 20.1 kBq/ml; DMI-DR was 146.7 kcps at 31.7 kBq/ml. Scatter fraction was 40.5% and 36.6%, respectively. DMI image contrast recovery (CR) values ranged from 73.2% (10 mm sphere) to 91.0% (37 mm sphere); DMI-DR, values ranged from 68.4% to 91.4%. DMI background variability (BV) was 1.8%–6.5%; DMI-DR was 2.3%–9.1%. The Q.Clear algorithm improved image quality, increasing CR and decreasing BV in both systems. The photopeak energy resolution was 9.63% and 12.19% for DMI and DMI-DR, respectively. The time-of-flight (TOF) resolution was 377.26 ps and 552.71 ps, respectively. Compared with measurements in other centers, results were similar and showed an absolute mean relative deviation of 6% for DMI and 7% for DMI-DR overall performance results.ConclusionsPerformance measures were higher for the 4-ring DMI than the DMI-DR system. The biggest advantages of the 4-ring DMI vs DMI-DR are improved sensitivity and count rate performance. This should allow a better image signal-to-noise ratio (SNR) for the same acquisition times or, similar SNR with lower acquisition times or injected activity. In its 3-ring configuration, the DMI showed worse performance results than the PMT-based system in terms of count rate scatter fraction and image quality (for similar axial FOV).

The SnapShot Freeze Algorithm Improves Image Quality of Coronary Computed Tomography Angiography Using High-Resolution Wide-Detector CT

Objective: The aim of this study was to investigate the effect the motion-correction reconstruction algorithm SnapShot Freeze (SSF) on the diagnostic performance of coronary computed tomography angiography (CCTA) using high-resolution, wide-detector CT. Methods: CCTA was performed using an electrocardiography-triggered auto-gating protocol in 62 patients. All images were reconstructed using both standard algorithm and motioncorrection algorithm (SSF). According to the American Heart Association (AHA) guidelines for coronary angiography, the coronary artery was divided into 15 segments. The coronary artery segments were evaluated using a 4-point Likert scale to determine the image quality. Arteries with a score greater than 1 were further evaluated by two expert radiologists blinded to conditions. The diagnostic performance of SSF was compared in the patients with high heart rate (HR ≥ 75 beats per minute) compared to low HR (&lt; 75 beats per minute). Results: At the segmental level, the Likert score for segments 1–4 and 6–14 were higher in SSF compared to standard algorithm (p &lt; 0.05). The proportion of images that was improved by SSF was greater for the high HR systolic phase imaging (66.7%) than for the low HR diastolic phase imaging (34.3%, p &lt; 0.05). Conclusion: The SSF algorithm improves coronary CTA images acquired using a 256-row wide detector CT, compared to standard algorithm, particularly in patients with high HR.

Evaluation and Optimization of a New PET Reconstruction Algorithm, Bayesian Penalized Likelihood Reconstruction, for Lung Cancer Assessment According to Lesion Size.

OBJECTIVE. The purpose of this study was to characterize the Bayesian penalized likelihood (BPL) reconstruction algorithm in comparison with an ordered subset expectation maximization (OSEM) reconstruction algorithm and to determine its optimal penalization factor (expressed as a beta value) for clinical use. MATERIALS AND METHODS. FDG PET/CT scans of 46 patients with lung cancer were reconstructed using OSEM and BPL with beta values of 200, 300, 400, 500, and 1000. The liver signal-to-noise ratio, mean standardized uptake value (SUVmean) of the liver, and maximum standardized uptake value (SUVmax) and SUVmean of the cancers were measured. Tumors were categorized into three size groups, and the percentage difference in the tumor SUVmax between OSEM and BPL with a beta value of 200 as well as the percentage difference in the SUVmax between BPL with a beta value of 200 and BPL with a beta value of 1000 were calculated. Image quality was assessed by visual scoring. RESULTS. BPL showed a significantly higher liver signal-to-noise ratio than OSEM, except for BPL with a beta value of 200. The liver SUVmean showed no statistical difference among all algorithms. The SUVmax and SUVmean of tumors decreased as the beta value increased. BPL with a beta value of 200 produced a significantly higher tumor SUVmax than did OSEM (p < 0.01), and BPL with a beta value of 400, 500, or 1000 produced a significantly lower tumor SUVmax than did OSEM (p < 0.01). Visual analysis showed the highest and lowest scores for BPL with beta values of 500 and 200, respectively. In the small size group, the percentage difference in the SUVmax between OSEM and BPL with a beta value of 200 and the percentage difference in the SUVmax between BPL with a beta value of 200 and BPL with a beta value of 1000 were significantly larger than that in the other size groups (p < 0.01). CONCLUSION. The BPL algorithm improves image quality without compromising image quantification. A beta value of 500 appeared to be optimal in this study. Smaller tumors were more influenced by BPL.

The value of iterative metal artifact reduction algorithms during antenna positioning for CT-guided microwave ablation

Objectives: To compare image quality between filtered back projection (FBP) and iterative reconstruction algorithm and dedicated metal artifact reduction (iMAR) algorithms during antenna positioning for computed tomography-guided microwave ablation (MWA).Materials and methods: An MWA antenna was positioned in the liver of five pigs under CT guidance. Different exposure settings (120kVp/200mAs-120kVp/50mAs) and image reconstruction techniques (FBP, iterative reconstruction with and without iMAR) were applied. Quantitative image analysis included density measurements in six positions (e.g., liver in extension of the antenna [ANTENNA] and liver >3 cm away from the antenna [LIVER-1]). Qualitative image analysis included assessment of overall quality, image noise, artifacts at the antenna tip, artifacts in liver parenchyma bordering antenna tip and newly generated artifacts. Two independent observers performed the analyses twice and interreader agreement was compared with Bland–Altman analysis.Results: For all exposure and reconstruction settings, density measurements for ANTENNA were significantly higher for the I30-1 iMAR compared with FBP and I30-1 (e.g., 8.3−17.2HU vs. −104.5 to 155.1HU; p ≤ 0.01, respectively). In contrast, for all exposure settings, density measurements for LIVER-1 were comparable between FBP and I30-1 iMAR (e.g., 49.4−50.4HU vs. 50.1−52.5U, respectively). For all exposure and reconstruction settings, subjective image quality for LIVER-1 was better for the I30-1 iMAR algorithm compared with FBP and I30-1. Bland–Altman interobserver agreement was from −0.2 to 0.2 for FBP and iMAR, and Cohen’s kappa was 0.74.Conclusion: Iterative algorithms I30-1 with iMAR algorithm improves image quality during antenna positioning and placement for CT-guided MWA and is applicable over a range of exposure settings.

Prototype metal artefact reduction algorithm in flat panel computed tomography - evaluation in patients undergoing transarterial hepatic radioembolisation.

To investigate the effect of an on-site prototype metal artefact reduction (MAR) algorithm in cone-beam CT-catheter-arteriography (CBCT-CA) in patients undergoing transarterial radioembolisation (RE) of hepatic masses. Ethical board approved retrospective study of 29 patients (mean 63.7±13.7 years, 11 female), including 16 patients with arterial metallic coils, undergoing CBCT-CA (8s scan, 200 degrees rotation, 397 projections). Image reconstructions with and without prototype MAR algorithm were evaluated quantitatively (streak-artefact attenuation changes) and qualitatively (visibility of hepatic parenchyma and vessels) in near- (<1cm) and far-field (>3cm) of artefact sources (metallic coils and catheters). Quantitative and qualitative measurements of uncorrected and MAR corrected images and different artefact sources were compared RESULTS: Quantitative evaluation showed significant reduction of near- and far-field streak-artefacts with MAR for both artefact sources (p<0.001), while remaining stable for unaffected organs (all p>0.05). Inhomogeneities of attenuation values were significantly higher for metallic coils compared to catheters (p<0.001) and decreased significantly for both after MAR (p<0.001). Qualitative image scores were significantly improved after MAR (all p<0.003) with by trend higher artefact degrees for metallic coils compared to catheters. In patients undergoing CBCT-CA for transarterial RE, prototype MAR algorithm improves image quality in proximity of metallic coil and catheter artefacts. • Metal objects cause artefacts in cone-beam computed tomography (CBCT) imaging. • These artefacts can be corrected by metal artefact reduction (MAR) algorithms. • Corrected images show significantly better visibility of nearby hepatic vessels and tissue. • Better visibility may facilitate image interpretation, save time and radiation exposure.

Iterative metal artefact reduction in CT: can dedicated algorithms improve image quality after spinal instrumentation?

To investigate the value of dedicated computed tomography (CT) iterative metal artefact reduction (iMAR) algorithms in patients after spinal instrumentation. Post-surgical spinal CT images of 24 patients performed between March 2015 and July 2016 were retrospectively included. Images were reconstructed with standard weighted filtered back projection (WFBP) and with two dedicated iMAR algorithms (iMAR-Algo1, adjusted to spinal instrumentations and iMAR-Algo2, adjusted to large metallic hip implants) using a medium smooth kernel (B30f) and a sharp kernel (B70f). Frequencies of density changes were quantified to assess objective image quality. Image quality was rated subjectively by evaluating the visibility of critical anatomical structures including the central canal, the spinal cord, neural foramina, and vertebral bone. Both iMAR algorithms significantly reduced artefacts from metal compared with WFBP (p<0.0001). Results of subjective image analysis showed that both iMAR algorithms led to an improvement in visualisation of soft-tissue structures (median iMAR-Algo1=3; interquartile range [IQR]:1.5-3; iMAR-Algo2=4; IQR: 3.5-4) and bone structures (iMAR-Algo1=3; IQR:3-4; iMAR-Algo2=4; IQR:4-5) compared to WFBP (soft tissue: median 2; IQR: 0.5-2 and bone structures: median 2; IQR: 1-3; p<0.0001). Compared with iMAR-Algo1, objective artefact reduction and subjective visualisation of soft-tissue and bone structures were improved with iMAR-Algo2 (p<0.0001). Both iMAR algorithms reduced artefacts compared with WFBP, however, the iMAR algorithm with dedicated settings for large metallic implants was superior to the algorithm specifically adjusted to spinal implants.

Dual-Energy Computed Tomography Angiography of the Head and Neck With Single-Source Computed Tomography

Dual-energy computed tomographic angiography (DE-CTA) has been demonstrated to improve the visualization of the head and neck vessels. The aim of this study was to test the potential of split-filter single-source dual-energy CT to automatically remove bone from the final CTA data set. Dual-energy CTA was performed in 50 consecutive patients to evaluate the supra-aortic arteries, either to grade carotid artery stenosis or to rule out traumatic dissections. Dual-energy CTA was performed on a 128-slice single-source CT system equipped with a special filter array to separate the 120-kV spectrum into a high- and a low-energy spectrum for DE-based automated bone removal. Image quality of fully automated bone suppression and subsequent manual optimization was evaluated by 2 radiologists on maximum intensity projections using a 4-grade scoring system. The effect of image reconstruction with an iterative metal artifact reduction algorithm on DE postprocessing was tested using a 3-grade scoring system, and the time demand for each postprocessing step was measured. Two patients were excluded due to insufficient arterial contrast enhancement; in the remaining 48 patients, automated bone removal could be performed successfully. The addition of iterative metal artifact reduction algorithm improved image quality in 58.3% of the cases. After manual optimization, DE-CTA image quality was rated excellent in 7, good in 29, and moderate in 10 patients. Interobserver agreement was high (κ = 0.85). Stenosis grading was not influenced using DE-CTA with bone removal as compared with the original CTA. The time demand for DE image reconstruction was significantly higher than for single-energy reconstruction (42.1 vs 20.9 seconds). Our results suggest that bone removal in DE-CTA of the head and neck vessels with a single-source CT is feasible and can be performed within acceptable time and moderate user interaction.

SU‐F‐J‐74: High Z Geometric Integrity and Beam Hardening Artifact Assessment Using a Retrospective Metal Artifact Reduction (MAR) Reconstruction Algorithm

Purpose:To test the efficacy of a retrospective metal artifact reduction (MAR) reconstruction algorithm for a commercial computed tomography (CT) scanner for radiation therapy purposes.Methods:High Z geometric integrity and artifact reduction analysis was performed with three phantoms using General Electric's (GE) Discovery CT. The three phantoms included: a Computerized Imaging Reference Systems (CIRS) electron density phantom (Model 062) with a 6.5 mm diameter titanium rod insert, a custom spine phantom using Synthes Spine hardware submerged in water, and a dental phantom with various high Z fillings submerged in water. Each phantom was reconstructed using MAR and compared against the original scan. Furthermore, each scenario was tested using standard and extended Hounsfield Unit (HU) ranges. High Z geometric integrity was performed using the CIRS phantom, while the artifact reduction was performed using all three phantoms.Results:Geometric integrity of the 6.5 mm diameter rod was slightly overestimated for non‐MAR scans for both standard and extended HU. With MAR reconstruction, the rod was underestimated for both standard and extended HU. For artifact reduction, the mean and standard deviation was compared in a volume of interest (VOI) in the surrounding material (water and water equivalent material, ∼0HU). Overall, the mean value of the VOI was closer to 0 HU for the MAR reconstruction compared to the non‐MAR scan for most phantoms. Additionally, the standard deviations for all phantoms were greatly reduced using MAR reconstruction.Conclusion:GE's MAR reconstruction algorithm improves image quality with the presence of high Z material with minimal degradation of its geometric integrity. High Z delineation can be carried out with proper contouring techniques. The effects of beam hardening artifacts are greatly reduced with MAR reconstruction. Tissue corrections due to these artifacts can be eliminated for simple high Z geometries and greatly reduced for more complex geometries.

Research on Blurred Image Restoration Quality Evaluation Method

Accurate evaluation of image quality is always being a problem. With blurry cotton images as samples, using variance methods, average gradient method, gray prediction error statistics method and information entropy method, this research evaluated restoration quality of 4 kinds of images from the perspective of grayscale contrast, sharpness, contrast and ambiguity. The results show that the 4 image restoration algorithms all improved image quality, and Wiener algorithm improved image quality best comparing with other 3 algorithms, Reg second, Blind algorithm and K-L algorithm were worse.