Quantitative and qualitative assessments of deep learning image reconstruction in low-keV virtual monoenergetic dual-energy CT.

OBJECTIVES: To evaluate a novel deep learning image reconstruction (DLIR) technique for dual-energy CT (DECT) derived virtual monoenergetic (VM) images compared to adaptive statistical iterative reconstruction (ASIR-V) in low kiloelectron volt (keV) images. METHODS: We analyzed 30 venous phase acute abdominal DECT (80/140 kVp) scans. Data were reconstructed to ASIR-V and DLIR-High at four different keV levels (40, 50, 74, and 100) with 1- and 3-mm slice thickness. Quantitative Hounsfield unit (HU) and noise assessment were measured within the liver, aorta, fat, and muscle. Subjective assessment of image noise, sharpness, texture, and overall quality was performed by two board-certified radiologists. RESULTS: DLIR reduced image noise by 19.9-35.5% (p < 0.001) compared to ASIR-V in all reconstructions at identical keV levels. Contrast-to-noise ratio (CNR) increased by 49.2-53.2% (p < 0.001) in DLIR 40-keV images compared to ASIR-V 50 keV, while no significant difference in noise was identified except for 1 and 3 mm in aorta and for 1-mm liver measurements, where ASIR-V 50 keV showed 5.5-6.8% (p < 0.002) lower noise levels. Qualitative assessment demonstrated significant improvement particularly in 1-mm reconstructions (p < 0.001). Lastly, DLIR 40 keV demonstrated comparable or improved image quality ratings when compared to ASIR-V 50 keV (p < 0.001 to 0.22). CONCLUSION: DLIR significantly reduced image noise compared to ASIR-V. Qualitative assessment showed that DLIR significantly improved image quality particularly in thin sliced images. DLIR may facilitate 40 keV as a new standard for routine low-keV VM reconstruction in contrast-enhanced abdominal DECT. KEY POINTS: • DLIR enables 40 keV as the routine low-keV VM reconstruction. • DLIR significantly reduced image noise compared to ASIR-V, across a wide range of keV levels in VM DECT images. • In low-keV VM reconstructions, improvements in image quality using DLIR were most evident and consistent in 1-mm sliced images.

Transesophageal Doppler reliably tracks changes in cardiac output in comparison with intermittent pulmonary artery thermodilution in cardiac surgery patients.

In this study a comparison of cardiac output (CO) measurements obtained with CardioQ transesophageal Doppler (TED) and pulmonary artery catheter (PAC) thermodilution (TD) technique was done in a systematic set-up, with induced changes in preload, afterload and heart rate. Twenty-five patients completed the study. Each patient were placed in the following successive positions: (1) supine, (2) head-down tilt, (3) head-up tilt, (4) supine, (5) supine with phenylephrine administration, (6) pace heart rate 80 beats per minute (bpm), (7) pace heart rate 110 bpm. The agreement of compared data was investigated by Bland-Altman plots, and to assess trending ability a four quadrants plot and a polar plot were constructed. Both methods showed an acceptable precision 6.4 % (PAC TD) and 12.8 % (TED). In comparison with PAC TD, the TED was associated with a mean bias in supine position of -0.30 l min-1 (95 % CI -0.88; 0.27), wide limits of agreement, a percentage error of 69.5 %, and a trending ability with a concordance rate of 92 %, angular bias of 1.1° and a radial sector size of 40.0° corresponding to an acceptable trending ability. In comparison with PAC TD, the CardioQ TED showed a low mean bias, wide limits of agreement and a larger percentage error than should be expected from the precision of the two methods. However, an acceptable trending ability was found. Thus, the CardioQ TED should not replace CO measurements done by PAC TD, but could be a valuable tool in guiding therapy.

Evaluation of thin-slice abdominal DECT using deep-learning image reconstruction in 74 keV virtual monoenergetic images: an image quality comparison.

PURPOSE: To compare noise, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR) and image quality using deep-learning image reconstruction (DLIR) vs. adaptive statistical iterative reconstruction (ASIR-V) in 0.625 and 2.5 mm slice thickness gray scale 74 keV virtual monoenergetic (VM) abdominal dual-energy CT (DECT). METHODS: This retrospective study was approved by the institutional review board and regional ethics committee. We analysed 30 portal-venous phase abdominal fast kV-switching DECT (80/140kVp) scans. Data were reconstructed to ASIR-V 60% and DLIR-High at 74 keV in 0.625 and 2.5 mm slice thickness. Quantitative HU and noise assessment were measured within liver, aorta, adipose tissue and muscle. Two board-certified radiologists evaluated image noise, sharpness, texture and overall quality based on a five-point Likert scale. RESULTS: DLIR significantly reduced image noise and increased CNR as well as SNR compared to ASIR-V, when slice thickness was maintained (p < 0.001). Slightly higher noise of 5.5-16.2% was measured (p < 0.01) in liver, aorta and muscle tissue at 0.625 mm DLIR compared to 2.5 mm ASIR-V, while noise in adipose tissue was 4.3% lower with 0.625 mm DLIR compared to 2.5 mm ASIR-V (p = 0.08). Qualitative assessments demonstrated significantly improved image quality for DLIR particularly in 0.625 mm images. CONCLUSIONS: DLIR significantly reduced image noise, increased CNR and SNR and improved image quality in 0.625 mm slice images, when compared to ASIR-V. DLIR may facilitate thinner image slice reconstructions for routine contrast-enhanced abdominal DECT.

New technology - demonstration of a vector velocity technique.

With conventional Doppler ultrasound it is not possible to estimate direction and velocity of blood flow, when the angle of insonation exceeds 60-70°. Transverse oscillation is an angle independent vector velocity technique which is now implemented on a conventional ultrasound scanner. In this paper a few of the possibilities with transverse oscillation are demonstrated.

Data on the quantitative assessment pulmonary ground-glass opacification from coronary computed tomography angiography datasets.

We assessed the CT attenuation density of the pulmonary tissue adjacent to the heart in patients with acute non-ST segment elevation myocardial infarction (J.T. Kuhl, T.S. Kristensen, A.F. Thomsen et al., 2016) [1]. This data was related to the level of ground-glass opacification evaluated by a radiologist, and data on the interobserver variability of semi-automated assessment of pulmonary attenuation density was provided.

Clinical and prognostic correlates of pulmonary congestion in coronary computed tomography angiography data sets.

BACKGROUND: Signs of pulmonary congestion obtained from cardiac computed tomography angiographic (coronary CTA) images have not previously been related to clinical congestion or outcome and the clinical value is, therefore, unknown. Our objective was to test the hypothesis that signs of pulmonary congestion predict clinical heart failure and adverse outcome in patients with myocardial infarction. METHODS: Coronary CTA was performed before invasive treatment in 400 prospectively included patients with non ST segment elevation myocardial infarction in an observational study. Using a previously described chest computed tomography evaluation algorithm, patients were classified as having "no congestion", "mild to moderate congestion" or "severe congestion". RESULTS: Using multivariate analyses, presence of pulmonary congestion on coronary CTA images was associated with age, female gender, left ventricular ejection fraction (LVEF) and left atrial size. The diagnostic accuracy for predicting clinical heart failure, defined as Killip class >1, was: sensitivity: 83%, specificity: 69%, positive predictive value: 25%, and negative predictive value: 97%. The median follow-up time was 50 months and the study end-point of death or hospitalization due to heart failure was reached in 68 (16%) patients. In a Cox proportional hazards model with adjustments for known risk factors and Killip class, the presence of "mild to moderate congestion" and "severe congestion" was independently associated with adverse outcome (Hazard ratio: 2.6 (95% CI:1.3-5.0) and 3.2 (1.3-7.5)). CONCLUSION: Signs of pulmonary congestion on coronary CTA images are closely correlated to cardiac dysfunction, predict clinical heart failure, and provide prognostic value independent of LVEF and Killip class.

Surveillance for hemodialysis access stenosis: usefulness of ultrasound vector volume flow.

PURPOSE: To investigate if ultrasound vector-flow imaging (VFI) is equal to the reference method ultrasound dilution technique (UDT) in estimating volume flow and changes over time in arteriovenous fistulas (AVFs) for hemodialysis. MATERIALS AND METHODS: From January 2014 to January 2015, patients with end-stage renal disease and matured functional AVFs were consecutively solicited to participate in this prospective study. All patients were included after written informed consent and approval by the National Committee on Biomedical Research Ethics and the local Ethics Committee (journal no. H-4-2014-FSP). VFI and UDT measurements were performed monthly over a six-month period. Nineteen patients were included in the study. VFI measurements were performed before dialysis, and UDT measurements after. Statistical analyses were performed with Bland-Altman plot, Student's t-test, four-quadrant plot, and regression analysis. Repeated measurements and precision analysis were used for reproducibility determination. RESULTS: Precision measurements for UDT and VFI were 32% and 20%, respectively (p = 0.33). Average volume flow measured with UDT and VFI were 1161 mL/min (±778 mL/min) and 1213 mL/min (±980 mL/(min), respectively (p = 0.3). The mean difference was -51 mL/min (CI: -150 mL/min to 46 mL/min) with limits of agreement from -35% to 54%, with a strong correlation (r2 = 0.87). A large change in volume flow between dialysis sessions detected by UDT was confirmed by VFI (p = 0.0001), but the concordance rate was poor (0.72). CONCLUSIONS: VFI is an acceptable method for volume flow estimation and volume flow changes over time in AVFs.

Improving Accuracy for Image Fusion in Abdominal Ultrasonography.

Image fusion involving real-time ultrasound (US) is a technique where previously recorded computed tomography (CT) or magnetic resonance images (MRI) are reformatted in a projection to fit the real-time US images after an initial co-registration. The co-registration aligns the images by means of common planes or points. We evaluated the accuracy of the alignment when varying parameters as patient position, respiratory phase and distance from the co-registration points/planes. We performed a total of 80 co-registrations and obtained the highest accuracy when the respiratory phase for the co-registration procedure was the same as when the CT or MRI was obtained. Furthermore, choosing co-registration points/planes close to the area of interest also improved the accuracy. With all settings optimized a mean error of 3.2 mm was obtained. We conclude that image fusion involving real-time US is an accurate method for abdominal examinations and that the accuracy is influenced by various adjustable factors that should be kept in mind.

In vivo validation of a blood vector velocity estimator with MR angiography.

Conventional Doppler methods for blood velocity estimation only estimate the velocity component along the ultrasound beam direction. This implies that a Doppler angle under examination close to 90 degrees results in unreliable information about the true blood direction and blood velocity. The novel method transverse oscillation (TO), which combines estimates of the axial and the transverse velocity components in the scan plane, makes it possible to estimate the vector velocity of the blood regardless of the Doppler angle. The present study evaluates the TO method with magnetic resonance phase contrast angiography (MRA) by comparing in vivo measurements of stroke volume. Eleven healthy volunteers were included in this prospective study. From the obtained data sets recorded with the 2 modalities, vector velocity sequences were constructed and stroke volume calculated. Angle of insonation was approximately 90 degrees for TO measurements. The correlation between the stroke volume estimated by TO and MRA was 0.91 (p < 0.01) with the equation for the line of regression: MRA = 1.1.TO-0.4. A Bland-Altman plot was additionally constructed where the mean difference was 0.2 ml with limits of agreement at -1.4 ml and 1.9 ml. The results indicate that reliable vector velocity estimates can be obtained in vivo using the presented angle-independent 2-D vector velocity method. The TO method can be a useful alternative to conventional Doppler systems by avoiding the angle artifact, thus giving quantitative velocity information.