Automatic Contrast Generation from Contrastless Computed Tomography.

The use of contrast-enhanced computed tomography (CTCA) for detection of coronary artery disease (CAD) exposes patients to the risks of iodine contrast-agents and excessive radiation, increases scanning time and healthcare costs. Deep learning generative models have the potential to artificially create a pseudo-enhanced image from non-contrast computed tomography (CT) scans.In this work, two specific models of generative adversarial networks (GANs) - the Pix2Pix-GAN and the Cycle-GAN - were tested with paired non-contrasted CT and CTCA scans from a private and public dataset. Furthermore, an exploratory analysis of the trade-off of using 2D and 3D inputs and architectures was performed. Using only the Structural Similarity Index Measure (SSIM) and the Peak Signal-to-Noise Ratio (PSNR), it could be concluded that the Pix2Pix-GAN using 2D data reached better results with 0.492 SSIM and 16.375 dB PSNR. However, visual analysis of the output shows significant blur in the generated images, which is not the case for the Cycle-GAN models. This behavior can be captured by the evaluation of the Fréchet Inception Distance (FID), that represents a fundamental performance metric that is usually not considered by related works in the literature.Clinical relevance- Contrast-enhanced computed tomography is the first line imaging modality to detect CAD resulting in unnecessary exposition to the risk of iodine contrast and radiation in particularly in young patients with no disease. This algorithm has the potential of being translated into clinical practice as a screening method for CAD in asymptomatic subjects or quick rule-out method of CAD in the acute setting or centres with no CTCA service. This strategy can eventually represent a reduction in the need for CTCA reducing its burden and associated costs.

Exercise Mode in Heart Failure: A Systematic Review and Meta-Analysis.

BACKGROUND: Optimising exercise prescription in heart failure (HF) with a preserved (HFpEF) or reduced (HFrEF) ejection fraction is clinically important. As such, the aim of this meta-analysis was to compare traditional moderate intensity training (MIT) against combined aerobic and resistance training (CT) and high-intensity interval training (HIIT) for improving aerobic capacity (VO2), as well as other clinically relevant parameters. METHODS: A comprehensive systematic search was performed to identify randomised controlled trials published between 1990 and May 2021. Research trials reporting the effects of MIT against CT or HIIT on peak VO2 in HFpEF or HFrEF were considered. Left-ventricular ejection fraction (LVEF) and various markers of diastolic function were also analysed. RESULTS: Seventeen studies were included in the final analysis, 4 of which compared MIT against CT and 13 compared MIT against HIIT. There were no significant differences between MIT and CT for peak VO2 (weighted mean difference [WMD]: 0.521 ml min-1 kg-1, [95% CI] = - 0.7 to 1.8, Pfixed = 0.412) or LVEF (WMD: - 1.129%, [95% CI] = - 3.8 to 1.5, Pfixed = 0.408). However, HIIT was significantly more effective than MIT at improving peak VO2 (WMD: 1.62 ml min-1 kg-1, [95% CI] = 0.6-2.6, Prandom = 0.002) and LVEF (WMD: 3.24%, [95% CI] = 1.7-4.8, Prandom < 0.001) in HF patients. When dichotomized by HF phenotype, HIIT remained significantly more effective than MIT in all analyses except for peak VO2 in HFpEF. CONCLUSIONS: HIIT is significantly more effective than MIT for improving peak VO2 and LVEF in HF patients. With the exception of peak VO2 in HFpEF, these findings remain consistent in both phenotypes. Separately, there is no difference in peak VO2 and LVEF change following MIT or CT, suggesting that the addition of resistance exercise does not inhibit aerobic adaptations in HF.

Candidate microRNAs as prognostic biomarkers in heart failure: A systematic review.

BACKGROUND: Heart failure (HF) is a high prevalent syndrome with significant burden worldwide. B-type natriuretic peptide (BNP) and N-terminal proBNP are the gold standard biomarkers in HF management. Although useful in clinical practice, they have limitations as their expression can be influenced by ventricular function, aging, obesity, renal failure and atrial arrhythmias. MicroRNAs have recently emerged as potential diagnostic and prognostic biomarkers, given that they are related to cell growth, proliferation, differentiation, and metabolism. An increasing amount of research has highlighted some microRNAs for their potential as HF biomarkers. However, different study designs, methods and study groups have led to inconsistent results. METHODS AND RESULTS: We performed a systematic search of available literature on Pubmed and Scopus reporting the prognostic value of microRNAs in HF, followed by a review of risk of bias, according to Quadas Group Standards. Simultaneously, microRNAs' potential as differential diagnosis and severity biomarkers was also analyzed. Studies have described circulating microRNA as potential diagnostic, prognostic, and severity markers. Mir-622, -519 and -499 were significantly related to HF with reduced ejection fraction, whereas miR-22-3p revealed greater ability as a severity biomarker. Let-7i-5p, miR-223-5p, miR-423-5p, miR-21, miR-1306-5p and miR-122 serum expressions presented a consistent correlation with HF prognosis. Furthermore, identified miR targets were associated with signaling pathways already known to be involved in HF progression. CONCLUSION: Several miRs were related to HF pathophysiology and demonstrated potential as biomarkers for disease progression. MicroRNAs have a promising role in HF, and although unquestionable, we require a deeper and broader understanding of their role and function for future research.

Decoding the radiomic and proteomic phenotype of epicardial adipose tissue associated with adverse left atrial remodelling and post-operative atrial fibrillation in aortic stenosis.

AIMS: Epicardial adipose tissue (EAT) volume and attenuation on computed tomography (CT) have been associated with atrial fibrillation. Beyond these conventional CT measures, radiomics allows extraction of high-dimensional data and deep quantitative adipose tissue phenotyping, which may capture its underlying biology. We aimed to explore the EAT proteomic and CT-radiomic signatures associated with impaired left atrial (LA) remodelling and post-operative atrial fibrillation (POAF). METHODS AND RESULTS: We prospectively included 132 patients with severe aortic stenosis with no prior atrial fibrillation referred for aortic valve replacement. Pre-operative non-contrast CT images were obtained for extraction of EAT volume and other radiomic features describing EAT texture. The LA function was assessed by 2D-speckle-tracking echocardiography peak atrial longitudinal strain and peak atrial contraction strain. The EAT biopsies were performed during surgery for proteomic analysis by sequential windowed acquisition of all theoretical fragment ion mass spectra (SWATH-MS). The POAF incidence was monitored from surgery until discharge. Impaired LA function and incident POAF were associated with EAT up-regulation of inflammatory and thrombotic proteins, and down-regulation of cardioprotective proteins with anti-inflammatory and anti-lipotoxic properties. The EAT volume was independently associated with LA enlargement, impaired function, and POAF risk. On CT images, EAT texture of patients with POAF was heterogeneous and exhibited higher maximum grey-level values than sinus rhythm patients, which correlated with up-regulation of inflammatory and down-regulation of lipid droplet-formation EAT proteins. The CT radiomics of EAT provided an area under the curve of 0.80 (95% confidence interval: 0.68-0.92) for discrimination between patients with POAF and sinus rhythm. CONCLUSION: Pre-operative CT-radiomic profile of EAT detected adverse EAT proteomics and identified patients at risk of developing POAF.

Myocardial oedema: pathophysiological basis and implications for the failing heart.

Myocardial fluid homeostasis relies on a complex interplay between microvascular filtration, interstitial hydration, cardiomyocyte water uptake and lymphatic removal. Dysregulation of one or more of these mechanisms may result in myocardial oedema. Interstitial and intracellular fluid accumulation disrupts myocardial architecture, intercellular communication, and metabolic pathways, decreasing contractility and increasing myocardial stiffness. The widespread use of cardiac magnetic resonance enabled the identification of myocardial oedema as a clinically relevant imaging finding with prognostic implications in several types of heart failure. Furthermore, growing experimental evidence has contributed to a better understanding of the physical and molecular interactions in the microvascular barrier, myocardial interstitium and lymphatics and how they might be disrupted in heart failure. In this review, we summarize current knowledge on the factors controlling myocardial water balance in the healthy and failing heart and pinpoint the new potential therapeutic avenues.

A systematic review and meta-analysis of randomized controlled studies comparing off-pump versus on-pump coronary artery bypass grafting in the elderly.

INTRODUCTION: Comparison of short and mid-term outcomes between off-pump CABG (OPCAB) and on-pump CABG (ONCAB) in patients older than 65 throughout a meta-analysis of randomized clinical trials (RCTs). EVIDENCE ACQUISITION: A literature search was conducted using 3 databases. RCTs reporting mortality outcomes of OPCAB versus ONCAB among the elderly were included. Data on myocardial infarction, stroke, re-revascularization, renal failure and composite endpoints after CABG were also collected. Random effects models were used to compute statistical combined measures and 95% confidence intervals (CI). EVIDENCE SYNTHESIS: Five RCTs encompassing 6221 patients were included (3105 OPCAB and 3116 ONCAB). There were no significant differences on mid-term mortality (pooled HR: 1.02, 95%CI: 0.89-1.17, P=0.80) and composite endpoint incidence (pooled HR: 0.98, 95%CI: 0.88-1.09, P=0.72) between OPCAB and ONCAB. At 30-day, there were no differences in mortality, myocardial infarction, stroke and renal complications. The need for early re-revascularization was significantly higher in OPCAB (pooled OR: 3.22, 95%CI: 1.28-8.09, P=0.01), with a higher percentage of incomplete revascularization being reported for OPCAB in trials included in this pooled result (34% in OPCAB vs. 29% in ONCAB, P<0.01). CONCLUSIONS: Data from RCTs in elderly patients showed that OPCAB and ONCAB provide similar mid-term results. OPCAB was associated with a higher risk of early re-revascularization. As CABG on the elderly is still insufficiently explored, further RCTs, specifically designed targeting this population, are needed to establish a better CABG strategy for these patients.

Machine learning phenotyping of scarred myocardium from cine in hypertrophic cardiomyopathy.

AIMS: Cardiovascular magnetic resonance (CMR) with late-gadolinium enhancement (LGE) is increasingly being used in hypertrophic cardiomyopathy (HCM) for diagnosis, risk stratification, and monitoring. However, recent data demonstrating brain gadolinium deposits have raised safety concerns. We developed and validated a machine-learning (ML) method that incorporates features extracted from cine to identify HCM patients without fibrosis in whom gadolinium can be avoided. METHODS AND RESULTS: An XGBoost ML model was developed using regional wall thickness and thickening, and radiomic features of myocardial signal intensity, texture, size, and shape from cine. A CMR dataset containing 1099 HCM patients collected using 1.5T CMR scanners from different vendors and centres was used for model development (n=882) and validation (n=217). Among the 2613 radiomic features, we identified 7 features that provided best discrimination between +LGE and -LGE using 10-fold stratified cross-validation in the development cohort. Subsequently, an XGBoost model was developed using these radiomic features, regional wall thickness and thickening. In the independent validation cohort, the ML model yielded an area under the curve of 0.83 (95% CI: 0.77-0.89), sensitivity of 91%, specificity of 62%, F1-score of 77%, true negatives rate (TNR) of 34%, and negative predictive value (NPV) of 89%. Optimization for sensitivity provided sensitivity of 96%, F2-score of 83%, TNR of 19% and NPV of 91%; false negatives halved from 4% to 2%. CONCLUSION: An ML model incorporating novel radiomic markers of myocardium from cine can rule-out myocardial fibrosis in one-third of HCM patients referred for CMR reducing unnecessary gadolinium administration.

Multiple versus single arterial grafting in the elderly: a meta-analysis of randomized controlled trials and propensity score studies.

INTRODUCTION: The benefit of adding a second arterial conduit is still controversial, mainly in specific subgroups. We conducted a meta-analysis of randomized controlled trials (RCTs) and propensity score (PS) studies comparing survival and early results in elderly patients who underwent coronary artery bypass grafting (CABG) with multiple (MAG) versus single arterial grafting (SAG). EVIDENCE ACQUISITION: MEDLINE, Web of Science and Cochrane Library were used to find relevant literature (1960-April 2020). Survival at a ≥1-year follow-up and early outcomes were evaluated. Outcomes were collected from matched samples or PS adjusted analysis: hazard ratio (HR) along with their variance, frequencies or odds ratios. Random effect models were used to compute combined statistical measures and 95% confidence intervals (CI) through generic inverse variance method (time-to-event) or Mantel-Haenszel method (binary events). EVIDENCE SYNTHESIS: Eleven PS cohorts and 1 RCT comprising >18,800 patients older than 70 (>6200 MAG and >12,500 SAG) were included in this meta-analysis. MAG was associated with lower long-term mortality (pooled HR: 0.81, 95% CI: 0.72-0.91, P<0.01, I2=64%) in the absence of higher risk of early mortality (pooled OR: 0.74, 95% CI: 0.44 to 1.25, P=0.27, I2=0%). In a meta-regression, MAG survival advantage was more pronounced in studies with a higher MAG usage rate (ß=-0.0052, P=0.021). CONCLUSIONS: Current evidence suggests that advanced age should not limit MAG's use considering its benefits in long-term survival. Of note, an individualized patient selection for this approach is warranted.

Impact of oral anticoagulation therapy on postoperative atrial fibrillation outcomes: a systematic review and meta-analysis.

BACKGROUND: Post-operative atrial fibrillation (POAF) is the most common complication after cardiac surgery. Recent studies had shown this phenomenon is no longer considered transitory and is associated with higher risk of thromboembolic events or death. The aim of this study was to systematically review and analyze previous studies comparing oral anticoagulation therapy with no anticoagulation, regarding these long-term outcomes. METHODS: PubMed/MEDLINE, EMBASE, Web of Science and Cochrane Database were systematically searched to identify the studies comparing the risk of stroke, or thromboembolic events or mortality of POAF patients who received anticoagulation compared with those who were not anticoagulated. Incidence of stroke, thromboembolic events and all-cause mortality were evaluated up to 10 years after surgery. Time-to-event outcomes were collected through hazard ratio (HR) along with their variance and the early endpoints using frequencies or odds ratio (OR). Random effect models were used to compute statistical combined measures and 95% confidence intervals (CI). Heterogeneity was evaluated through Q statistic-related measures of variance (Tau2, I2, Chi-squared test). RESULTS: Eight observational cohort studies were selected, including 15,335 patients (3492 on Oral Anticoagulants (OAC) vs 11,429 without OAC) that met the inclusion criteria for qualitative synthesis. Patients had a wide gender distribution (38.6-82.3%), each study with a mean age above 65 years (67.5-85). Vitamin K antagonists were commonly prescribed anticoagulants (74.3-100%). OAC was associated with a protective impact on all-cause mortality at a mean of 5.0 years of follow-up (HR is 0.85 [0.72-1.01]; p = 0.07; I2 = 48%). Thromboembolic events did not differ between the two treatment arms (HR 0.68 [0.40-1.15], p = 0.15). CONCLUSION: Current literature suggests a possibly protective impact of OAC therapy for all-cause mortality in patients with new-onset atrial fibrillation after cardiac surgery. However, it does not appear to impact thromboembolism rate.

Pericardial NT-Pro-BNP and GDF-15 as Biomarkers of Atrial Fibrillation and Atrial Matrix Remodeling in Aortic Stenosis.

AIMS: This study aimed to evaluate the association of GDF-15 and NT-pro-BNP in two different biological matrices with AF in severe aortic stenosis patients undergoing aortic valve replacement surgery (AVR), its association with atrial matrix remodeling, as well as with 30-day postoperative outcomes. MAIN METHODS: One hundred and twenty-six patients between 2009 and 2019 with severe aortic stenosis undergoing AVR surgery in a tertiary hospital were assessed. KEY FINDINGS: pericardial fluid GDF-15 and pericardial fluid and serum NT-pro-BNP were increased in AF patients with aortic stenosis. COL1A1 and COL3A1 gene expression increased when pericardial fluid NT-pro-BNP values were higher. TIMP4 was positively correlated with pericardial fluid GDF-15. SIGNIFICANCE: GDF-15 and NT-pro-BNP in the pericardial fluid are biomarkers of atrial fibrillation in aortic stenosis and correlate with atrial matrix remodeling. AKI is predicted by both serum and pericardial fluid GDF-15.

Left ventricular reverse remodeling and function by strain analysis in aortic stenosis: A CMR analysis of the EPICHEART study.

INTRODUCTION AND OBJECTIVES: In severe aortic stenosis (AS), the impact of aortic valve replacement (AVR) on left ventricular (LV) systolic function assessed by strain and measured by echocardiography or cardiac magnetic resonance (CMR) has been controversial. We aimed to investigate LV systolic myocardial function changes six months after AVR using global longitudinal (GLS), circumferential (GCS) and radial (GRS) strain derived from CMR imaging. METHODS: We included 39 severe AS patients (69.3±7.8 years; 61.5% male) with preserved LV ejection fraction (LVEF) who were recruited as part of the EPICHEART study and underwent successful AVR (aortic valvular area: 0.8 cm2 (IQR: 0.2) pre- to 1.8 cm2 (IQR:0.5) post-AVR). Structural and functional parameters were assessed at baseline and six months after AVR, including LV GRS, GCS and GLS analysis by CMR, using cine short-axial and two-, three-, and four-chamber long-axial view. Comparison between baseline and postoperative LV remodeling was performed using Student t-test and Wilcoxon test. RESULTS: At six-month follow-up, LV mass, end-diastolic and end-systolic volumes, stroke volume, cardiac output, lateral E/e', tricuspid annular plane systolic excursion, right ventricular (RV) S wave velocity, GLS [-15.6% (IQR: 4.39) to -13.7% (IQR: 4.62)] and GCS [-17.8±3.58% to -16.1±2.94%] reduced significantly, while LVEF and GRS remained unchanged and lateral e' velocity increased. CONCLUSIONS: Despite favorable reverse LV structural and diastolic functional remodeling six months following AVR, GLS and GCS assessed by CMR reduced compared to baseline, LVEF remained unchanged. The clinical utility and timing of assessment of postoperative strain changes as a marker of systolic function progression needs further research.

Sensitivity of Myocardial Radiomic Features to Imaging Parameters in Cardiac MR Imaging.

BACKGROUND: Cardiac magnetic resonance (MR) images are often collected with different imaging parameters, which may impact the calculated values of myocardial radiomic features. PURPOSE: To investigate the sensitivity of myocardial radiomic features to changes in imaging parameters in cardiac MR images. STUDY TYPE: Prospective. POPULATION: A total of 11 healthy participants/five patients. FIELD STRENGTH/ SEQUENCE: A 3 T/cine balanced steady-state free-precession, T1 -weighted spoiled gradient-echo, T2 -weighted turbo spin-echo, and quantitative T1 and T2 mapping. For each sequence, the flip angle, in-plane resolution, slice thickness, and parallel imaging technique were varied to study the sensitivity of radiomic features to alterations in imaging parameters. ASSESSMENT: Myocardial contours were manually delineated by experienced readers, and a total of 1023 radiomic features were extracted using PyRadiomics with 11 image filters and six feature families. STATISTICAL TESTS: Sensitivity was defined as the standardized mean difference (D effect size), and the robust features were defined at sensitivity < 0.2. Sensitivity analysis was performed on predefined sets of reproducible features. The analysis was performed using the entire cohort of 16 subejcts. RESULTS: 64% of radiomic features were robust (sensitivity < 0.2) to changes in any imaging parameter. In qualitative sequences, radiomic features were most sensitive to changes in in-plane spatial resolution (spatial resolution: 0.6 vs. flip angle: 0.19, parallel imaging: 0.18, slice thickness: 0.07; P < 0.01 for all); in quantitative sequences, radiomic features were least sensitive to changes in spatial resolution (spatial resolution: 0.07 vs. slice thickness: 0.16, flip angle: 0.24; P < 0.01 for all). In an individual feature level, no singular feature family/image filter was identified as robust (sensitivity < 0.2) across sequences; however, highly sensitive features were predominantly associated with high-frequency wavelet filters across all sequences (32/50 features). DATA CONCLUSION: In cardiac MR, a considerable number of radiomic features are sensitive to changes in sequence parameters. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 1.

Influence of EPICardial adipose tissue in HEART diseases (EPICHEART) study: Protocol for a translational study in coronary atherosclerosis.

INTRODUCTION: Accumulation of epicardial adipose tissue (EAT) is associated with coronary artery disease (CAD) and increased risk of coronary events in asymptomatic subjects and low-risk patients, suggesting that EAT promotes atherosclerosis in its early stage. Recent studies have shown that the presence of CAD affects the properties of adjacent EAT, leading to dynamic changes in the molecular players involved in the interplay between EAT and the coronary arteries over the history of the disease. The role of EAT in late-stage CAD has not been investigated. OBJECTIVES: In a comparative analysis with mediastinal and subcutaneous adipose tissue, we aim to investigate whether the volume of EAT assessed by computed tomography and its proteome assessed by SWATH-MS mass spectrometry are associated with late stages of CAD in an elderly cohort of severe aortic stenosis patients. METHODS: The EPICHEART study (NCT03280433) is a prospective study enrolling patients with severe degenerative aortic stenosis referred for elective aortic valve replacement, whose protocol includes preoperative clinical, nutritional, echocardiographic, cardiac computed tomography and invasive coronary angiographic assessments. During cardiac surgery, samples of EAT and mediastinal and subcutaneous thoracic adipose tissue are collected for proteomics analysis by SWATH-MS. In addition, pericardial fluid and peripheral and coronary sinus blood samples are collected to identify circulating and local adipose tissue-derived biomarkers of CAD. CONCLUSION: We designed a translational study to explore the association of EAT quantity and quality with advanced CAD. We expect to identify new biochemical factors and biomarkers in the crosstalk between EAT and the coronary arteries that are involved in the pathogenesis of late coronary atherosclerosis, especially coronary calcification, which might be translated into new therapeutic targets and imaging tools by biomedical engineering.

Reproducibility of Segmentation-based Myocardial Radiomic Features with Cardiac MRI.

PURPOSE: To investigate reproducibility of myocardial radiomic features with cardiac MRI. MATERIALS AND METHODS: Test-retest studies were performed with a 3-T MRI system using commonly used cardiac MRI sequences of cine balanced steady-state free precession (cine bSSFP), T1-weighted and T2-weighted imaging, and quantitative T1 and T2 mapping in phantom experiments and 10 healthy participants (mean ± standard deviation age, 29 years ± 13). In addition, this study assessed repeatability in 51 patients (56 years ± 14) who underwent imaging twice during the same session. Three readers independently delineated the myocardium to investigate inter- and intraobserver reproducibility of radiomic features. A total of 1023 radiomic features were extracted by using PyRadiomics (https://pyradiomics.readthedocs.io/) with 11 image filters and six feature families. The intraclass correlation coefficient (ICC) was estimated to assess reproducibility and repeatability, and features with ICCs greater than or equal to 0.8 were considered reproducible. RESULTS: Different reproducibility patterns were observed among sequences in in vivo test-retest studies. In cine bSSFP, the gray-level run-length matrix was the most reproducible feature family, and the wavelet low-pass filter applied horizontally and vertically was the most reproducible image filter. In T1 and T2 maps, intensity-based statistics (first-order) and gray-level co-occurrence matrix features were the most reproducible feature families, without a dominant reproducible image filter. Across all sequences, gray-level nonuniformity was the most frequently identified reproducible feature name. In inter- and intraobserver reproducibility studies, respectively, only 32%-47% and 61%-73% of features were identified as reproducible. CONCLUSION: Only a small subset of myocardial radiomic features was reproducible, and these reproducible radiomic features varied among different sequences. Supplemental material is available for this article. © RSNA, 2020See also the commentary by Leiner in this issue.

Deep complex convolutional network for fast reconstruction of 3D late gadolinium enhancement cardiac MRI.

Several deep-learning models have been proposed to shorten MRI scan time. Prior deep-learning models that utilize real-valued kernels have limited capability to learn rich representations of complex MRI data. In this work, we utilize a complex-valued convolutional network (ℂNet) for fast reconstruction of highly under-sampled MRI data and evaluate its ability to rapidly reconstruct 3D late gadolinium enhancement (LGE) data. ℂNet preserves the complex nature and optimal combination of real and imaginary components of MRI data throughout the reconstruction process by utilizing complex-valued convolution, novel radial batch normalization, and complex activation function layers in a U-Net architecture. A prospectively under-sampled 3D LGE cardiac MRI dataset of 219 patients (17 003 images) at acceleration rates R = 3 through R = 5 was used to evaluate ℂNet. The dataset was further retrospectively under-sampled to a maximum of R = 8 to simulate higher acceleration rates. We created three reconstructions of the 3D LGE dataset using (1) ℂNet, (2) a compressed-sensing-based low-dimensional-structure self-learning and thresholding algorithm (LOST), and (3) a real-valued U-Net (realNet) with the same number of parameters as ℂNet. LOST-reconstructed data were considered the reference for training and evaluation of all models. The reconstructed images were quantitatively evaluated using mean-squared error (MSE) and the structural similarity index measure (SSIM), and subjectively evaluated by three independent readers. Quantitatively, ℂNet-reconstructed images had significantly improved MSE and SSIM values compared with realNet (MSE, 0.077 versus 0.091; SSIM, 0.876 versus 0.733, respectively; p < 0.01). Subjective quality assessment showed that ℂNet-reconstructed image quality was similar to that of compressed sensing and significantly better than that of realNet. ℂNet reconstruction was also more than 300 times faster than compressed sensing. Retrospective under-sampled images demonstrate the potential of ℂNet at higher acceleration rates. ℂNet enables fast reconstruction of highly accelerated 3D MRI with superior performance to real-valued networks, and achieves faster reconstruction than compressed sensing.