OBJECTIVES: We implemented an explainable machine learning (ML) model to gain insight into the association between cardiac magnetic resonance markers and adverse outcomes of cardiovascular hospitalization and all-cause death (composite endpoint) in patients with nonischemic dilated cardiomyopathy (NICM).
BACKGROUND: Risk stratification of patients with NICM remains challenging. An explainable ML model has the potential to provide insight into the contributions of different risk markers in the prediction model.
METHODS: An explainable ML model based on extreme gradient boosting (XGBoost) machines was developed using cardiac magnetic resonance and clinical parameters. The study cohorts consist of patients with NICM from 2 academic medical centers: Beth Israel Deaconess Medical Center (BIDMC) and Brigham and Women’s Hospital (BWH), with 328 and 214 patients, respectively. XGBoost was trained on 70% of patients from the BIDMC cohort and evaluated based on the other 30% as internal validation. The model was externally validated using the BWH cohort. To investigate the contribution of different features in our risk prediction model, we used Shapley additive explanations (SHAP) analysis.
RESULTS: During a mean follow-up duration of 40 months, 34 patients from BIDMC and 33 patients from BWH experienced the composite endpoint. The area under the curve for predicting the composite endpoint was 0.71 for the internal BIDMC validation and 0.69 for the BWH cohort. SHAP analysis identified parameters associated with right ventricular (RV) dysfunction and remodeling as primary markers of adverse outcomes. High risk thresholds were identified by SHAP analysis and thus provided thresholds for top predictive continuous clinical variables.
CONCLUSIONS: An explainable ML-based risk prediction model has the potential to identify patients with NICM at risk for cardiovascular hospitalization and all-cause death. RV ejection fraction, end-systolic and end-diastolic volumes (as indicators of RV dysfunction and remodeling) were determined to be major risk markers.
Purpose: To improve the accuracy and robustness of T1 estimation by MyoMap-Net, a deep learning–based approach using 4 inversion-recovery T1-weighted images for cardiac T1 mapping.
Methods: MyoMapNet is a fully connected neural network for T1 estimation of an accelerated cardiac T1 mapping sequence, which collects 4 T1-weighted images by a single Look-Locker inversion-recovery experiment (LL4).MyoMap-Net was originally trained using in vivo data from the modified Look-Locker inversion recovery sequence, which resulted in significant bias and sensitivity to various confounders. This study sought to train MyoMapNet using signals generated from numerical simulations and phantom MR data under multiple simulated confounders. The trained model was then evaluated by phantom data scanned using new phantom vials that differed from those used for training. The performance of the new model was compared with modified Look-Locker inversion recovery sequence and saturation-recovery single-shot acquisition for measuring native and postcontrast T1 in 25 subjects.
Results: In the phantom study, T1 values measured by LL4 with MyoMapNet were highly correlated with reference values from the spin-echo sequence. Furthermore, the estimated T1 had excellent robustness to changes in flip angle and off-resonance. Native and postcontrast myocardium T1 at 3 Tesla measured by saturation-recovery single-shot acquisition, modified Look-Locker inversion recovery sequence, and MyoMapNet were 1483±46.6 ms and 791±45.8 ms, 1169±49.0 ms and 612±36.0 ms, and 1443±57.5 ms and 700±57.5 ms, respectively. The corresponding extracellular volumes were 22.90%±3.20%, 28.88%±3.48%, and 30.65%±3.60%, respectively.
Conclusion: Training MyoMapNet with numerical simulations and phantom data will improve the estimation of myocardial T1 values and increase its robustness to confounders while also reducing the overall T1 mapping estimation time to only 4 heartbeats.
Objective To investigate the performance of various deep learning (DL) architectures for MyoMapNet, a DL model for T1 estimation using accelerated cardiac T1 mapping from four T1-weighted images collected after a single inversion pulse (Look-Locker, LL4). Methods We implemented and tested three DL architectures for MyoMapNet: (a) a fully connected neural network (FC), (b) convolutional neural networks (VGG19, ResNet50), and (c) encoder-decoder networks with skip connections (ResUNet, U-Net). Modified Look-Locker Inversion Recovery (MOLLI) images from 749 patients at 3T were used for training, validation, and testing. The first four T1-weighted images from MOLLI5(3)3 and/or MOLLI4(1)3(1)2 protocols were extracted to create accelerated cardiac T1 mapping data. We also prospectively collected data from 28 subjects using MOLLI and LL4 to further evaluate model performance. Results Despite rigorous training, conventional VGG19 and ResNet50 models failed to produce anatomically correct T1 maps, and T1 values had significant errors. While ResUNet yielded good quality maps, it significantly under-estimated T1. Both FC and U-Net, however, yielded excellent image quality with good T1 accuracy for both native (FC/U-Net/MOLLI= 1217 ± 64/ 1208 ± 61/1199 ± 61 ms, all P<0.05) and post-contrast myocardial T1 (FC/U-Net/MOLLI= 578 ± 57/ 567 ± 54/574 ± 55 ms, all P<0.05). In terms of precision, the U-Net model yielded better T1 precision compared to the FC architecture (standard deviation of 61 ms vs. 67 ms for the myocardium for native (P<0.05), and 31 ms vs. 38 ms (P<0.05), for post-contrast). Similar findings were observed in prospectively collected LL4 data. Conclusion U-Net and FC DL models in MyoMapNet enable fast myocardial T1 mapping using only four T1-weighted images collected from a single Lock-Locker sequence with comparable accuracy. U-Net also provides slight improvement in precision.
Purpose: To develop and evaluate a free breathing non-electrocardiograph (ECG) myocardial T1* mapping sequence using radial imaging to quantify the changes in myocardial T1* between rest and exercise (T1*reactivity) in exercise cardiac MRI (Ex-CMR). Methods: A free-running T1* sequence was developed using a saturation pulse followed by three Look-Locker inversion-recovery experiments. Each Look-Locker continuously acquired data as radial trajectory using a low flip-angle spoiled gradient-echo readout. Self-navigation was performed with a temporal resolution of∼100 ms for retrospectively extracting respiratory motion. The mid-diastole phase for every cardiac cycle was retrospectively detected on the recorded electrocardiogram signal using an empirical model. Multiple measurements were performed to obtain mean value to reduce effects from the free-breathing acquisition. Finally, data acquired at both mid-diastole and end-expiration are picked and reconstructed by a low-rank plus sparsity constraint algorithm. The performance of this sequence was evaluated by simulations, phantoms, and in vivo studies at rest and after physiological exercise. Results: Numerical simulation demonstrated that changes in T1* are related to the changes in T1; however, other factors such as breathing motion could influence T1* measurements. Phantom T1* values measured using free-running T1* mapping sequence had good correlation with spin-echo T1 values and was insensitive to heart rate. In the Ex-CMR study, the measured T1* reactivity was 10% immediately after exercise and declined over time.
Conclusion: The free-running T1* mapping sequence allows free-breathing non-ECG quantification of changes in myocardial T1* with physiological exercise. Although, absolute myocardial T1* value is sensitive to various confounders
Purpose To develop and evaluate MyoMapNet, a rapid myocardial T1 mapping approach that uses fully connected neural networks (FCNN) to estimate T1 values from four T1-weighted images collected after a single inversion pulse in four heartbeats (Look-Locker, LL4).
Method We implemented an FCNN for MyoMapNet to estimate T1 values from a reduced number of T1-weighted images and corresponding inversion-recovery times. We studied MyoMapNet performance when trained using native, post-contrast T1, or a combination of both. We also explored the effects of number of T1-weighted images (four and five) for native T1. After rigorous training using in-vivo modified Look-Locker inversion recovery (MOLLI) T1 mapping data of 607 patients, MyoMapNet performance was evaluated using MOLLI T1 data from 61 patients by discarding the additional T1-weighted images. Subsequently, we implemented a prototype MyoMapNet and LL4 on a 3 T scanner. LL4 was used to collect T1 mapping data in 27 subjects with inline T1 map reconstruction by MyoMapNet. The resulting T1 values were compared to MOLLI.
Results MyoMapNet trained using a combination of native and post-contrast T1-weighted images had excellent native and post-contrast T1 accuracy compared to MOLLI. The FCNN model using four T1-weighted images yields similar performance compared to five T1-weighted images, suggesting that four T1 weighted images may be sufficient. The inline implementation of LL4 and MyoMapNet enables successful acquisition and reconstruction of T1 maps on the scanner. Native and post-contrast myocardium T1 by MOLLI and MyoMapNet was 1170 ± 55 ms vs. 1183 ± 57 ms (P = 0.03), and 645 ± 26 ms vs. 630 ± 30 ms (P = 0.60), and native and post-contrast blood T1 was 1820 ± 29 ms vs. 1854 ± 34 ms (P = 0.14), and 508 ± 9 ms vs. 514 ± 15 ms (P = 0.02), respectively.
Conclusion A FCNN, trained using MOLLI data, can estimate T1 values from only four T1-weighted images. MyoMapNet enables myocardial T1 mapping in four heartbeats with similar accuracy as MOLLI with inline map reconstruction.
Objectives The aim of this study was to define the variability of maximal wall thickness (MWT) measurements across modalities and predict its impact on care in patients with hypertrophic cardiomyopathy (HCM). Background Left ventricular MWT measured by echocardiography or cardiovascular magnetic resonance (CMR) contributes to the diagnosis of HCM, stratifies risk, and guides key decisions, including whether to place an implantable cardioverter-defibrillator (ICD). Methods A 20-center global network provided paired echocardiographic and CMR data sets from patients with HCM, from which 17 paired data sets of the highest quality were selected. These were presented as 7 randomly ordered pairs (at 6 cardiac conferences) to experienced readers who report HCM imaging in their daily practice, and their MWT caliper measurements were captured. The impact of measurement variability on ICD insertion decisions was estimated in 769 separately recruited multicenter patients with HCM using the European Society of Cardiology algorithm for 5-year risk for sudden cardiac death. Results MWT analysis was completed by 70 readers (from 6 continents; 91% with >5 years’ experience). Seventy-nine percent and 68% scored echocardiographic and CMR image quality as excellent. For both modalities (echocardiographic and then CMR results), intramodality inter-reader MWT percentage variability was large (range –59% to 117% [SD ±20%] and –61% to 52% [SD ±11%], respectively). Agreement between modalities was low (SE of measurement 4.8 mm; 95% CI 4.3 mm-5.2 mm; r = 0.56 [modest correlation]). In the multicenter HCM cohort, this estimated echocardiographic MWT percentage variability (±20%) applied to the European Society of Cardiology algorithm reclassified risk in 19.5% of patients, which would have led to inappropriate ICD decision making in 1 in 7 patients with HCM (8.7% would have had ICD placement recommended despite potential low risk, and 6.8% would not have had ICD placement recommended despite intermediate or high risk). Conclusions Using the best available images and experienced readers, MWT as a biomarker in HCM has a high degree of inter-reader variability and should be applied with caution as part of decision making for ICD insertion. Better standardization efforts in HCM recommendations by current governing societies are needed to improve clinical decision making in patients with HCM.
Background: Heart failure patients with preserved ejection fraction (HFpEF) are at increased risk of future hospitalization. Contrast agents are often contra-indicated in HFpEF patients due to the high prevalence of concomitant kidney disease. Therefore, the prognostic value of a noncontrast cardiac magnetic resonance imaging (MRI) for HF-hospitalization is important. Purpose: To develop and test an explainable machine learning (ML) model to investigate incremental value of noncontrast cardiac MRI for predicting HF-hospitalization. Study Type: Retrospective. Population: A total of 203 HFpEF patients (mean, 64 ± 12 years, 48% women) referred for cardiac MRI were randomly split into training validation (143 patients, ~70%) and test sets (60 patients, ~30%). Field strength: A 1.5 T, balanced steady-state free precession (bSSFP) sequence. Assessment: Two ML models were built based on the tree boosting technique and the eXtreme Gradient Boosting model (XGBoost): 1) basic clinical ML model using clinical and echocardiographic data and 2) cardiac MRI-based ML model that included noncontrast cardiac MRI markers in addition to the basic model. The primary end point was defined as HF-hospitalization. Statistical Tests: ML tool was used for advanced statistics, and the Elastic Net method for feature selection. Area under the receiver operating characteristic (ROC) curve (AUC) was compared between models using DeLong's test. To gain insight into the ML model, the SHapley Additive exPlanations (SHAP) method was leveraged. A P-value <0.05 was considered statistically significant. Results: During follow-up (mean, 50 ± 39 months), 85 patients (42%) reached the end point. The cardiac MRI-based ML model using the XGBoost algorithm provided a significantly superior prediction of HF-hospitalization (AUC: 0.81) compared to the basic model (AUC: 0.64). The SHAP analysis revealed left atrium (LA) and right atrium (RV) strains as top imaging markers contributing to its performance with cutoff values of 17.5% and −15%, respectively. Data Conclusions: Using an ML model, RV and LA strains measured in noncontrast cardiac MRI provide incremental value in predicting future hospitalization in HFpEF.
Cardiovascular disease is the leading cause of death and a significant contributor of health care costs. Noninvasive imaging plays an essential role in the management of patients with cardiovascular disease. Cardiac magnetic resonance (MR) can noninvasively assess heart and vascular abnormalities, including biventricular structure/function, blood hemodynamics, myocardial tissue composition, microstructure, perfusion, metabolism, coronary microvascular function, and aortic distensibility/stiffness. Its ability to characterize myocardial tissue composition is unique among alternative imaging modalities in cardiovascular disease. Significant growth in cardiac MR utilization, particularly in Europe in the last decade, has laid the necessary clinical groundwork to position cardiac MR as an important imaging modality in the workup of patients with cardiovascular disease. Although lack of availability, limited training, physician hesitation, and reimbursement issues have hampered widespread clinical adoption of cardiac MR in the United States, growing clinical evidence will ultimately overcome these challenges. Advances in cardiac MR techniques, particularly faster image acquisition, quantitative myocardial tissue characterization, and image analysis have been critical to its growth. In this review article, we discuss recent advances in established and emerging cardiac MR techniques that are expected to strengthen its capability in managing patients with cardiovascular disease.
BACKGROUND Quantification of myocardium scarring in late gadolinium enhanced (LGE) cardiac magnetic resonance imaging can be challenging due to low scar-to-background contrast and low image quality. To resolve ambiguous LGE regions, experienced readers often use conventional cine sequences to accurately identify the myocardium borders. PURPOSE To develop a deep learning model for combining LGE and cine images to improve the robustness and accuracy of LGE scar quantification. STUDY TYPE Retrospective POPULATION A total of 191 hypertrophic cardiomyopathy patients: 1) 162 patients from two sites randomly split into training (50%; 81 patients), validation (25%, 40 patients), and testing (25%; 41 patients); and 2) an external testing dataset (29 patients) from a third site. FIELD STRENGTH/SEQUENCE 1.5T, inversion-recovery segmented gradient-echo LGE and balanced steady-state free-precession cine sequences ASSESSMENT: Two convolutional neural networks (CNN) were trained for myocardium and scar segmentation, one with and one without LGE-Cine fusion. For CNN with fusion, the input was two aligned LGE and cine images at matched cardiac phase and anatomical location. For CNN without fusion, only LGE images were used as input. Manual segmentation of the datasets was used as reference standard. STATISTICAL TESTS Manual and CNN-based quantifications of LGE scar burden and of myocardial volume were assessed using Pearson linear correlation coefficients (r) and Bland-Altman analysis. RESULTS Both CNN models showed strong agreement with manual quantification of LGE scar burden and myocardium volume. CNN with LGE-Cine fusion was more robust than CNN without LGE-Cine fusion, allowing for successful segmentation of significantly more slices (603 [95%] vs. 562 (89%) of 635 slices; P < 0.001). Also, CNN with LGE-Cine fusion showed better agreement with manual quantification of LGE scar burden than CNN without LGE-Cine fusion (%ScarLGE-cine = 0.82 × %Scarmanual , r = 0.84 vs. %ScarLGE = 0.47 × %Scarmanual , r = 0.81) and myocardium volume (VolumeLGE-cine = 1.03 × Volumemanual , r = 0.96 vs. VolumeLGE = 0.91 × Volumemanual , r = 0.91). DATA CONCLUSION CNN based LGE-Cine fusion can improve the robustness and accuracy of automated scar quantification.
BACKGROUND Cardiac magnetic resonance (MR) images are often collected with different imaging parameters, which may impact the calculated values of myocardial radiomic features.
To investigate the sensitivity of myocardial radiomic features to changes in imaging parameters in cardiac MR images.
A total of 11 healthy participants/five patients.
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.
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.
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.
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).
In cardiac MR, a considerable number of radiomic features are sensitive to changes in sequence parameters.
PURPOSE To develop and evaluate a real-time phase contrast (PC) MRI protocol via complex-difference deep learning (DL) framework.
METHODS DL used two 3D U-nets to separately filter aliasing artifact from radial real-time velocity-compensated and complex-difference images. U-nets were trained with synthetic real-time PC generated from electrocardiograph (ECG) -gated, breathhold, segmented PC (ECG-gated segmented PC) acquired at the ascending aorta of 510 patients. In 21 patients, free-breathing, ungated real-time (acceleration rate = 28.8) and ECG-gated segmented (acceleration rate = 2) PC were prospectively acquired at the ascending aorta. Hemodynamic parameters (cardiac output [CO], stroke volume [SV], and mean velocity at peak systole [peak mean velocity]) were measured for ECG-gated segmented and DL-filtered synthetic real-time PC and compared using Bland-Altman and linear regression analyses. Additionally, hemodynamic parameters were quantified from DL-filtered, compressed-sensing (CS) -reconstructed, and gridding reconstructed prospective real-time PC and compared to ECG-gated segmented PC.
RESULTS Synthetic real-time PC with DL showed strong correlation (R > 0.98) and good agreement with ECG-gated segmented PC for quantified hemodynamic parameters (mean-difference: CO = −0.3 L/min, SV = −4.3 mL, peak mean velocity = −2.3 cm/s). On average, DL required 0.39 s/frame to filter prospective real-time PC, which was 4.6-fold faster than CS. Compared to CS, DL showed superior correlation, tighter limits of agreement (LOAs), better bias for peak mean velocity, and worse bias for CO and SV. Compared to gridding, DL showed similar correlation, tighter LOAs for CO and SV, similar bias for CO, and worse bias for SV and peak mean velocity. CONCLUSION The complex-difference DL framework accelerated real-time PCMRI by nearly 28-fold, enabling rapid free-running real-time assessment of flow hemodynamics.
PURPOSE To reduce inflow and motion artifacts in free-breathing, free-running, steady-state spoiled gradient echo T1-weighted (SPGR) myocardial perfusion imaging.
METHOD Unsaturated spins from inflowing blood or out-of-plane motion cause flashing artifacts in free-running SPGR myocardial perfusion. During free-running SPGR, 1 non-selective RF excitation was added after every 3 slice-selective RF excitations to suppress inflow artifacts by forcing magnetization in neighboring regions to steady-state. Bloch simulations and phantom experiments were performed to evaluate the impact of the flip angle and non-selective RF frequency on inflowing spins and tissue contrast. Free-running perfusion with (n = 11) interleaved non-selective RF or without (n = 11) were studied in 22 subjects (age = 60.2 ± 14.3 years, 11 male). Perfusion images were graded on a 5-point Likert scale for conspicuity of wall enhancement, inflow/motion artifact, and streaking artifact and compared using Wilcoxon sum-rank testing.
RESULT Numeric simulation showed that 1 non-selective RF excitation applied after every 3 slice-selective RF excitations produced superior out-of-plane signal suppression compared to 1 non-selective RF excitation applied after every 6 or 9 sliceselective RF excitations. In vitro experiments showed that a 30° flip angle produced near-optimal myocardial contrast. In vivo experiments demonstrated that the addition of interleaved non-selective RF significantly (P < .01) improved conspicuity of wall enhancement (mean score = 4.4 vs. 3.2) and reduced inflow/motion (mean score = 4.5 vs. 2.5) and streaking (mean score = 3.9 vs. 2.4) artifacts. CONCLUSION Non-selective RF excitations interleaved between slice-selective excitations can reduce image artifacts in free-breathing, ungated perfusion images. Further studies are warranted to assess the diagnostic accuracy of the proposed solution for evaluating myocardial ischemia.
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 þLGE 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%.
CONCLUSIONS 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.
BACKGROUND Quantification of myocardium scarring in late gadolinium enhanced (LGE) cardiac magnetic resonance imaging can be challenging due to low scar-to-background contrast and low image quality. To resolve ambiguous LGE regions, experienced readers often use conventional cine sequences to accurately identify the myocardium borders. PURPOSE To develop a deep learning model for combining LGE and cine images to improve the robustness and accuracy of LGE scar quantification. Study Type: Retrospective. Population: A total of 191 hypertrophic cardiomyopathy patients: 1) 162 patients from two sites randomly split into train- ing (50%; 81 patients), validation (25%, 40 patients), and testing (25%; 41 patients); and 2) an external testing dataset (29 patients) from a third site. FIELD STRENGTH/SEQUENCE 1.5T, inversion-recovery segmented gradient-echo LGE and balanced steady-state free-preces- sion cine sequences Assessment: Two convolutional neural networks (CNN) were trained for myocardium and scar segmentation, one with and one without LGE-Cine fusion. For CNN with fusion, the input was two aligned LGE and cine images at matched cardiac phase and anatomical location. For CNN without fusion, only LGE images were used as input. Manual segmentation of the datasets was used as reference standard. STATISTICAL TESTS Manual and CNN-based quantifications of LGE scar burden and of myocardial volume were assessed using Pearson linear correlation coefficients (r) and Bland–Altman analysis. Results: Both CNN models showed strong agreement with manual quantification of LGE scar burden and myocardium vol- ume. CNN with LGE-Cine fusion was more robust than CNN without LGE-Cine fusion, allowing for successful segmenta- tion of significantly more slices (603 [95%] vs. 562 (89%) of 635 slices; P < 0.001). Also, CNN with LGE-Cine fusion showed better agreement with manual quantification of LGE scar burden than CNN without LGE-Cine fusion (%ScarLGE-cine = 0.82 × %Scarmanual, r = 0.84 vs. %ScarLGE = 0.47 × %Scarmanual, r = 0.81) and myocardium volume (VolumeLGE-cine = 1.03 × Volumemanual, r = 0.96 vs. VolumeLGE = 0.91 × Volumemanual, r = 0.91). DATA CONCLUSION CNN based LGE-Cine fusion can improve the robustness and accuracy of automated scar quantification.
PURPOSE To develop a free-breathing sequence, that is, Multislice Joint T1-T2, for simultaneous measurement of myocardial T1 and T2 for multiple slices to achieve whole left-ventricular coverage.
METHODS Multislice Joint T1-T2 adopts slice-interleaved acquisition to collect 10 single-shot electrocardiogram-triggered images for each slice prepared by saturation and T2 preparation to simultaneously estimate myocardial T1 and T2 and achieve whole left-ventricular coverage. Prospective slice-tracking using a respiratory navigator and retrospective image registration are used to reduce through-plane and in-plane motion, respectively. Multislice Joint T1-T2 was validated through numerical simulations and phantom and in vivo experiments, and compared with saturation-recovery single-shot acquisition and T2-prepared balanced Steady-State Free Precession (T2-prep SSFP) sequences.
Phantom T1 and T2 from Multislice Joint T1-T2 had good accuracy and precision, and were insensitive to heart rate. Multislice Joint T1-T2 yielded T1 and T2 maps of nine left-ventricular slices in 1.4 minutes. The mean left-ventricular T1 difference between saturation-recovery single-shot acquisition and Multislice Joint T1-T2 across healthy subjects and patients was 191 ms (1564 ± 60 ms versus 1373 ± 50 ms; P < .05) and 111 ms (1535 ± 49 ms vs 1423 ± 49 ms; P < .05), respectively. The mean difference in left-ventricular T2 between T2-prep SSFP and Multislice Joint T1-T2 across healthy subjects and patients was −6.3 ms (42.4 ± 1.4 ms vs 48.7 ± 2.5; P < .05) and −5.7 ms (41.6 ± 2.5 ms vs 47.3 ± 2.7; P < .05), respectively.
CONCLUSION Multislice Joint T1-T2 enables quantification of whole left-ventricular T1 and T2 during free breathing within a clinically feasible scan time of less than 2 minutes.
BACKGROUND In patients with nonischemic cardiomyopathy, nonischemic fibrosis detected by late gadolinium enhancement (LGE) cardiovascular magnetic resonance is related to adverse cardiovascular outcomes. However, its relationship with left ventricular (LV) mechanical deformation parameters remains unclear. We sought to investigate the association between LV mechanics and the presence, location, and extent of fibrosis in patients with nonischemic cardiomyopathy.
METHODS AND RESULTS We retrospectively identified 239 patients with nonischemic cardiomyopathy (67% male; 55±14 years) referred for a clinical cardiovascular magnetic resonance. LGE was present in 109 patients (46%), most commonly (n=52; 22%) in the septum. LV deformation parameters did not differentiate between LGE‐positive and LGE‐negative groups. Global longitudinal, radial, and circumferential strains, twist and torsion showed no association with extent of fibrosis. Patients with septal fibrosis had a more depressed LV ejection fraction (30±12% versus 35±14%; P=0.032) and more impaired global circumferential strain (−7.9±3.5% versus −9.7±4.4%; P=0.045) and global radial strain (10.7±5.2% versus 13.3±7.7%; P=0.023) than patients without septal LGE. Global longitudinal strain was similar in both groups. While patients with septal‐only LGE (n=28) and free wall–only LGE (n=32) had similar fibrosis burden, the septal‐only LGE group had more impaired LV ejection fraction and global circumferential, longitudinal, and radial strains (all P<0.05).
CONCLUSION There is no association between LV mechanical deformation parameters and presence or extent of fibrosis in patients with nonischemic cardiomyopathy. Septal LGE was associated with poor global LV function, more impaired global circumferential and radial strains, and more impaired global strain rates.
OBJECTIVES This study sought to investigate the sensitivity of electroanatomical mapping (EAM) to detect scar as identified by late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR).
BACKGROUND Previous studies have shown correlation between low voltage electrogram amplitude and myocardial scar. However, voltage amplitude is influenced by the distance between the scar and the mapping surface and its extent.The aim of this study is to examine the reliability of low voltage EAM as a surrogate for myocardial scar using LGE-derived scar as the reference.
METHODS Twelve swine underwent anterior wall infarction by occlusion of the left anterior descending artery (LAD) (n ¼ 6) or inferior wall infarction by occlusion of the left circumflex artery (LCx) (n ¼ 6). Subsequently, animals underwent CMR and EAM using a multielectrode mapping catheter. CMR characteristics, including wall thickness, LGE location and extent, and EAM maps, were independently analyzed, and concordance between voltage maps and CMR characteristics was assessed.
RESULTS LGE volume was similar between the LCx and LAD groups (8.5 _ 2.2 ml vs. 8.3 _ 2.5 ml, respectively; p ¼ 0.852). LGE scarring in the LAD group was more subendocardial, affected a larger surface area, and resulted in significant wall thinning (4.88 _ 0.43 mm). LGE scarring in the LCx group extended from the endocardium to the epicardium with minimal reduction in wall thickness (scarred: 5.4 _ 0.67 mm vs. remote: 6.75 _ 0.38 mm). In all the animals in the LAD group, areas of low voltage corresponded with LGE and wall thinning, whereas only 2 of 6 animals in the LCx group had low voltage areas on EAM. Bipolar and unipolar voltage amplitudes were higher in thick inferior walls in the LCx group than in thin anterior walls in the LAD group, despite a similar LGE volume.
CONCLUSIONS Discordances between LGE-detected scar areas and low voltage areas by EAM highlighted the limitations of the current EAM system to detect scar in thick myocardial wall regions. (J Am Coll Cardiol EP 2020;6:1452–64)
PURPOSE To develop and validate a saturation‐delay‐inversion recovery preparation, slice tracking and multi‐slice based sequence for measuring whole‐heart native T1.
METHOD The proposed free‐breathing sequence performs T1 mapping of multiple left‐ventricular slices by slice‐interleaved acquisition to collect 10 electrocardiogram‐triggered single‐shot slice‐selective images for each slice. A saturation‐delay‐inversion recovery pulse is used for T1 preparation. Prospective slice tracking by the diaphragm navigator and retrospective registration are used to reduce through‐plane and in‐plane motion, respectively. The proposed sequence was validated in both phantom and human subjects (12 healthy subjects and 15 patients who were referred for a clinical cardiac MR exam) and compared with saturation recovery single‐shot acquisition (SASHA) and modified Look‐Locker inversion recovery (MOLLI).
RESULTS Phantom T1 measured by the proposed sequence had excellent agreement (R2 = 0.99) with the ground‐truth T1 and was insensitive to heart rate. In both healthy subjects and patients, the proposed sequence yielded nine left‐ventricular T1 maps per volume in less than 2 minutes (healthy volunteers: 1.8 ± 0.4 minutes; patients: 1.9 ± 0.2 minutes). The average T1 of whole left ventricle for all healthy subjects and patients were 1560 ± 61 and 1535 ± 49 ms by SASHA, 1208 ± 42 and 1233 ± 56 ms by MOLLI5(3)3, and 1397 ± 34 and 1433 ± 56 ms by the proposed sequence, respectively. The corresponding coefficient of variation of T1 were 6.2 ± 1.4% and 5.8 ± 1.6%, 5.3 ± 1.1% and 5.1 ± 0.8%, and 4.9 ± 0.8% and 4.5 ± 0.8%, respectively.
CONCLUSION The proposed sequence enables quantification of whole heart T1 with good accuracy and precision in less than 2 minutes during free breathing.
PURPOSE Cardiac MR cine imaging allows accurate and reproducible assessment of cardiac function. However, its long scan time not only limits the spatial and temporal resolutions but is challenging in patients with breath-holding difficulty or non-sinus rhythms. To reduce scan time, we propose a multi-domain convolutional neural network (MD-CNN) for fast reconstruction of highly undersampled radial cine images.
METHODS MD-CNN is a complex-valued network that processes MR data in k-spaceand image domains via k-space interpolation and image-domain subnetworks for residual artifact suppression. MD-CNN exploits spatio-temporal correlations across timeframes and multi-coil redundancies to enable high acceleration. Radial cine data were prospectively collected in 108 subjects (50 ± 17 y, 72 males) using retrospective gated acquisition with 80%:20% split for training/testing. Images were reconstructed by MD-CNN and k-t Radial Sparse-Sense(kt-RASPS) using an undersampled dataset (14 of 196 acquired views; relative acceleration rate = 14). MD-CNN images were evaluated quantitatively using mean-squared-error (MSE) and structural similarity index (SSIM) relative to reference images, and qualitatively by three independent readers for left ventricular (LV) border sharpness and temporal fidelity using 5-point Likert-scale (1-non-diagnostic, 2-poor, 3-fair, 4-good, and 5-excellent).
RESULTS MD-CNN showed improved MSE and SSIM compared to kt-RASPS (0.11 ±0.10 vs. 0.61 ± 0.51, and 0.87 ± 0.07 vs. 0.72 ± 0.07, respectively; P < .01). Qualitatively, MD-CCN significantly outperformed kt-RASPS in LV border sharpness (3.87 ± 0.66 vs. 2.71 ± 0.58 at end-diastole, and 3.57 ± 0.6 vs. 2.56 ± 0.6 at end-systole, respectively; P < .01) and temporal fidelity (3.27 ± 0.65 vs. 2.59 ± 0.59; P < .01).
CONCLUSION MD-CNN reduces the scan time of cine imaging by a factor of 23.3 andprovides superior image quality compared to kt-RASPS.
OBJECTIVES This study sought to determine whether myocardial tissue heterogeneity scanned by native T1mapping could improve risk stratification in patients with nonischemic dilated cardiomyopathy (NICM) evaluated for primary prevention by ICD. BACKGROUND The benefit of insertable cardiac-defibrillator (ICD) as primary prevention ICD in patients with NICM remains to be fully clarified. METHODS A total of 115 NICM candidates for primary prevention and 55 healthy controls with similar distributions of age and sex were prospectively enrolled. Imaging was performed at 1.5-T using a protocol that included cine magnetic resonance for left ventricular function, late gadolinium enhancement (LGE) for focal scarring, and 5-slice native T1 mapping for diffuse fibrosis and heterogeneity. The last method was assessed by mean absolute deviation of the segmental pixel-SD from the average pixel-SD (Mad-SD). The primary endpoint was a composite of appropriate ICD therapy and sudden cardiac death. RESULTS During a median follow-up of 24 months, 13 patients (11%) experienced the primary endpoint. Dichotomized Mad-SD > 0.24 provided a comparable outcome to the presence of LGE for the primary endpoint (annual event rate: 9.8% vs. 10.9%). The integration of Mad-SD to global native T1 showed excellent arrhythmic event-free survival (annual event rate: 0%), and high sensitivity of 85% (95% confidence interval [CI]: 55% to 98%) and moderate specificity of 72% (95% CI: 62% to 80%), with a C-statistic of 0.76 (95% CI: 0.64 to 0.87), which was comparable to the presence, location, or extent of LGE in its ability to predict arrhythmic events. CONCLUSIONS Combined myocardium tissue heterogeneity and interstitial fibrosis assessment by native T1 mapping is an important predictor of ventricular tachycardia and ventricular fibrillation and provides additive risk stratification for primary prevention ICD in NICM patients without the need for gadolinium contrast.
In patients with heart failure with preserved ejection fraction (HFpEF), diabetes mellitus (DM) and obesity are important comorbidities as well as major risk factors. Their conjoint impact on the myocardium provides insight into the HFpEF aetiology. We sought to investigate the association between obesity, DM, and their combined effect on alterations in the myocardial tissue in HFpEF patients. One hundred and sixty-two HFpEF patients (55 ± 12 years, 95 men) and 45 healthy subjects (53 ± 12 years, 27 men) were included. Patients were classified according to comorbidity prevalence (36 obese patients without DM, 53 diabetic patients without obesity, and 73 patients with both). Myocardial remodeling, fibrosis, and longitudinal contractility were quantified with cardiovascular magnetic resonance imaging using cine and myocardial native T1 images. Patients with DM and obesity had impaired global longitudinal strain (GLS) and increased myocardial native T1 compared to patients with only one comorbidity (DM + Obesity vs. DM and Obesity; GLS, - 15 ± 2.1 vs - 16.5 ± 2.4 and - 16.7 ± 2.2%; native T1, 1162 ± 37 vs 1129 ± 25 and 1069 ± 29 ms; P < 0.0001 for all). A negative synergistic effect of combined obesity and DM prevalence was observed for native T1 (np2 = 0.273, p = 0.002) and GLS (np2 = 0.288, p < 0.0001). Additionally, severity of insulin resistance was associated with GLS (R = 0.590, P < 0.0001), and native T1 (R = 0.349, P < 0.0001). The conjoint effect of obesity and DM in HFpEF patients is associated with diffuse myocardial fibrosis and deterioration in GLS. The negative synergistic effects observed on the myocardium may be related to severity of insulin resistance.
BACKGROUND: The pattern of myocardial fibrosis differs significantly between different cardiomyopathies. Fibrosis in hypertrophic cardiomyopathy (HCM) is characteristically as patchy and regional but in dilated cardiomyopathy (DCM) as diffuse and global. We sought to investigate if texture analyses on myocardial native T1 mapping can differentiate between fibrosis patterns in patients with HCM and DCM. METHODS: We prospectively acquired native myocardial T1 mapping images for 321 subjects (55±15 years, 70% male): 65 control, 116 HCM, and 140 DCM patients. To quantify different fibrosis patterns, four sets of texture descriptors were used to extract 152 texture features from native T1 maps. Seven features were sequentially selected to identify HCM- and DCM-specific patterns in 70% of data (training dataset). Pattern reproducibility and generalizability were tested on the rest of data (testing dataset) using support vector machines (SVM) and regression models. RESULTS: Pattern-derived texture features were capable to identify subjects in HCM, DCM, and controls cohorts with 202/237(85.2%) accuracy of all subjects in the training dataset using 10-fold cross-validation on SVM (AUC = 0.93, 0.93, and 0.93 for controls, HCM and DCM, respectively), while pattern-independent global native T1 mapping was poorly capable to identify those subjects with 121/237(51.1%) accuracy (AUC = 0.78, 0.51, and 0.74) (P<0.001 for all). The pattern-derived features were reproducible with excellent intra- and inter-observer reliability and generalizable on the testing dataset with 75/84(89.3%) accuracy. CONCLUSION: Texture analysis of myocardial native T1 mapping can characterize fibrosis patterns in HCM and DCM patients and provides additional information beyond average native T1 values.
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.
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 6 standard deviation age, 29 years 6 13). In addition, this study assessed repeatability in 51 patients (56 years 6 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.
BACKGROUND In patients with suspected or known hypertrophic cardiomyopathy (HCM), late gadolinium enhancement (LGE) provides diagnostic and prognostic value. However, contraindications and long-term retention of gadolinium have raised concern about repeated gadolinium administration in this population. Alternatively, native T1 -mapping enables identification of focal fibrosis, the substrate of LGE. However HCM-specific heterogeneous fibrosis distribution leads to subtle T1 -maps changes that are difficult to identify.
To apply radiomic texture analysis on native T1 -maps to identify patients with a low likelihood of LGE(+), thereby reducing the number of patients exposed to gadolinium administration.
STUDY TYPE Retrospective interpretation of prospectively acquired data.
SUBJECTS In all, 188 (54.7 ± 14.4 years, 71% men) with suspected or known HCM.
FIELD STRENGTH/SEQUENCE A 1.5T scanner; slice-interleaved native T1 -mapping (STONE) sequence and 3D LGE after administration of 0.1 mmol/kg of gadobenate dimeglumine.
ASSESSMENT Left ventricular LGE images were location-matched with native T1 -maps using anatomical landmarks. Using a split-sample validation approach, patients were randomly divided 3:1 (training/internal validation vs. test cohorts). To balance the data during training, 50% of LGE(-) slices were discarded.
STATISTICAL TESTS Four sets of texture descriptors were applied to the training dataset for capture of spatially dependent and independent pixel statistics. Five texture features were sequentially selected with the best discriminatory capacity between LGE(+) and LGE(-) T1 -maps and tested using a decision tree ensemble (DTE) classifier.
RESULTS The selected texture features discriminated between LGE(+) and LGE(-) T1 -maps with a c-statistic of 0.75 (95% confidence interval [CI]: 0.70-0.80) using 10-fold cross-validation during internal validation in the training dataset and 0.74 (95% CI: 0.65-0.83) in the independent test dataset. The DTE classifier provided adequate labeling of all (100%) LGE(+) patients and 37% of LGE(-) patients during testing.
DATA CONCLUSION Radiomic analysis of native T1 -images can identify ~1/3 of LGE(-) patients for whom gadolinium administration can be safely avoided.
PURPOSE To assess the performance of an automated myocardial T2 and extracellular volume (ECV) quantification method using transfer learning of a fully convolutional neural network (CNN) pretrained to segment the myocardium on T1 mapping images.
MATERIALS AND METHODS
A single CNN previously trained and tested using 11 550 manually segmented native T1-weighted images was used to segment the myocardium for automated myocardial T2 and ECV quantification. Reference measurements from 1525 manually processed T2 maps and 1525 ECV maps (from 305 patients) were used to evaluate the performance of the pretrained network. Correlation coefficient (R) and Bland-Altman analysis were used to assess agreement between automated and reference values on per-patient, per-slice, and per-segment analyses. Furthermore, transfer learning effectiveness in the CNN was evaluated by comparing its performance to four CNNs trained using manually segmented T2-weighted and postcontrast T1-weighted images and initialized using random-weights or weights of the pretrained CNN.
RESULTS T2 and ECV measurements using the pretrained CNN strongly correlated with reference values in per-patient (T2: R = 0.88, 95% confidence interval [CI]: 0.85, 0.91; ECV: R = 0.91, 95% CI: 0.89, 0.93), per-slice (T2: R = 0.83, 95% CI: 0.81, 0.85; ECV: R = 0.84, 95% CI: 0.82, 0.86), and per-segment (T2: R = 0.75, 95% CI: 0.74, 0.77; ECV: R = 0.76, 95% CI: 0.75, 0.77) analyses. In Bland-Altman analysis, the automatic and reference values were in good agreement in per-patient (T2: 0.3 msec ± 2.9; ECV: -0.3% ± 1.7), per-slice (T2: 0.1 msec ± 4.6; ECV: -0.3% ± 2.5), and per-segment (T2: 0.0 msec ± 6.5; ECV: -0.4% ± 3.5) analyses. The performance of the pretrained network was comparable to networks refined or trained from scratch using additional manually segmented images.
CONCLUSION Transfer learning extends the utility of pretrained CNN-based automated native T1 mapping analysis to T2 and ECV mapping without compromising performance.
AIMS Multielectrode mapping catheters can be advantageous for identifying surviving myocardial bundles in scar. This study aimed to evaluate the utility of a new multielectrode catheter with increased number of small and closely spaced electrodes for mapping ventricles with healed infarction.
METHODS AND RESULTS
In 12 swine (four healthy and eight with infarction), the left ventricle was mapped with investigational (OctarayTM) and standard (PentarayTM) multielectrode mapping catheters. The investigational catheter has more electrodes (48 vs. 20), each with a smaller surface area (0.9 vs. 2.0mm2) and spacing is fixed at 2mm (vs. 2–6–2 mm). Electrogram (EGM) characteristics, mapping efficiency and scar description were compared between the catheters and late gadolinium enhancement (LGE). Electrogram acquisition rate was faster with the investigational catheter (814 ± 126 vs.148 ± 58 EGM/min, P = 0.02) resulting in higher density maps (38 ± 10.3 vs. 10.1 ± 10.4 EGM/cm2, P = 0.02). Bipolar voltage amplitude was similar between the catheters in normal and infarcted myocardium (P = 0.265 and P = 0.44) and the infarct surface area was similar between the catheters (P = 0.12) and corresponded to subendocardial LGE. The investigational catheter identified a higher proportion of near-field local abnormal ventricular activities within the low-voltage area (53 ± 16% vs. 34 ± 16%, P = 0.03) that were considered far-field EGMs by the standard catheter. The investigational catheter was also advantageous for mapping haemodymically non-tolerated ventricular tachycardias due to its higher acquisition rate (P < 0.001).
CONCLUSION A novel multielectrode mapping catheter with higher number of small, and closely spaced electrodes increases the mapping speed, EGM density and the ability to recognize low amplitude near-field EGMs in ventricles with healed infarction.