Cardiovascular Magnetic Resonance Radiomics to Identify Components of the Extracellular Matrix in Dilated Cardiomyopathy.

BACKGROUND: Current cardiovascular magnetic resonance sequences cannot discriminate between different myocardial extracellular space (ECSs), including collagen, noncollagen, and inflammation. We sought to investigate whether cardiovascular magnetic resonance radiomics analysis can distinguish between noncollagen and inflammation from collagen in dilated cardiomyopathy. METHODS: We identified data from 132 patients with dilated cardiomyopathy scheduled for an invasive septal biopsy who underwent cardiovascular magnetic resonance at 3 T. Cardiovascular magnetic resonance imaging protocol included native and postcontrast T1 mapping and late gadolinium enhancement (LGE). Radiomic features were computed from the midseptal myocardium, near the biopsy region, on native T1, extracellular volume (ECV) map, and LGE images. Principal component analysis was used to reduce the number of radiomic features to 5 principal radiomics. Moreover, a correlation analysis was conducted to identify radiomic features exhibiting a strong correlation (r>0.9) with the 5 principal radiomics. Biopsy samples were used to quantify ECS, myocardial fibrosis, and inflammation. RESULTS: Four histopathological phenotypes were identified: low collagen (n=20), noncollagenous ECS expansion (n=49), mild to moderate collagenous ECS expansion (n=42), and severe collagenous ECS expansion (n=21). Noncollagenous expansion was associated with the highest risk of myocardial inflammation (65%). Although native T1 and ECV provided high diagnostic performance in differentiating severe fibrosis (C statistic, 0.90 and 0.90, respectively), their performance in differentiating between noncollagen and mild to moderate collagenous expansion decreased (C statistic: 0.59 and 0.55, respectively). Integration of ECV principal radiomics provided better discrimination and reclassification between noncollagen and mild to moderate collagen (C statistic, 0.79; net reclassification index, 0.83 [95% CI, 0.45-1.22]; P<0.001). There was a similar trend in the addition of native T1 principal radiomics (C statistic, 0.75; net reclassification index, 0.93 [95% CI, 0.56-1.29]; P<0.001) and LGE principal radiomics (C statistic, 0.74; net reclassification index, 0.59 [95% CI, 0.19-0.98]; P=0.004). Five radiomic features per sequence were identified with correlation analysis. They showed a similar improvement in performance for differentiating between noncollagen and mild to moderate collagen (native T1, ECV, LGE C statistic, 0.75, 0.77, and 0.71, respectively). These improvements remained significant when confined to a single radiomic feature (native T1, ECV, LGE C statistic, 0.71, 0.70, and 0.64, respectively). CONCLUSIONS: Radiomic features extracted from native T1, ECV, and LGE provide incremental information that improves our capability to discriminate noncollagenous expansion from mild to moderate collagen and could be useful for detecting subtle chronic inflammation in patients with dilated cardiomyopathy.

Radiomics of Late Gadolinium Enhancement Reveals Prognostic Value of Myocardial Scar Heterogeneity in Hypertrophic Cardiomyopathy.

BACKGROUND: Late gadolinium enhancement (LGE) scar burden by cardiac magnetic resonance is a major risk factor for sudden cardiac death (SCD) in hypertrophic cardiomyopathy (HCM). However, there is currently limited data on the incremental prognostic value of integrating myocardial LGE radiomics (ie, shape and texture features) into SCD risk stratification models. OBJECTIVES: The purpose of this study was to investigate the incremental prognostic value of myocardial LGE radiomics beyond current European Society of Cardiology (ESC) and American College of Cardiology (ACC)/American Heart Association (AHA) models for SCD risk prediction in HCM. METHODS: A total of 1,229 HCM patients (62% men; age 52 ± 16 years) from 3 medical centers were included. Left ventricular myocardial radiomic features were calculated from LGE images. Principal component analysis was used to reduce the radiomic features and calculate 3 principal radiomics (PrinRads). Cox and logistic regression analyses were then used to evaluate the significance of the extracted PrinRads of LGE images, alone or in combination with ESC or ACC/AHA models, to predict SCD risk. The ACC/AHA risk markers include LGE burden using a dichotomized 15% threshold of LV scar. RESULTS: SCD events occurred in 30 (2.4%) patients over a follow-up period of 49 ± 28 months. Risk prediction using PrinRads resulted in higher c-statistics than the ESC (0.69 vs 0.57; P = 0.02) and the ACC/AHA (0.69 vs 0.67; P = 0.75) models. Risk predictions were improved by combining the 3 PrinRads with ESC (0.73 vs 0.57; P < 0.01) or ACC/AHA (0.76 vs 0.67; P < 0.01) risk scores. The net reclassification index was improved by combining the PrinRads with ESC (0.25 [95% CI: 0.08-0.43]; P = 0.005) or ACC/AHA (0.05 [95% CI: -0.07 to 0.16]; P = 0.42) models. One PrinRad was a significant predictor of SCD risk (HR: 0.57 [95% CI: 0.39-0.84]; P = 0.01). LGE heterogeneity was a major component of PrinRads and a significant predictor of SCD risk (HR: 0.07 [95% CI: 0.01-0.75]; P = 0.03). CONCLUSIONS: Myocardial LGE radiomics are strongly associated with SCD risk in HCM and provide incremental risk stratification beyond current ESC or AHA/ACC risk models. Our proof-of-concept study warrants further validation.

Scanner-Independent MyoMapNet for Accelerated Cardiac MRI T1 Mapping Across Vendors and Field Strengths.

BACKGROUND: In cardiac T1 mapping, a series of T1 -weighted (T1 w) images are collected and numerically fitted to a two or three-parameter model of the signal recovery to estimate voxel-wise T1 values. To reduce the scan time, one can collect fewer T1 w images, albeit at the cost of precision or/and accuracy. Recently, the feasibility of using a neural network instead of conventional two- or three-parameter fit modeling has been demonstrated. However, prior studies used data from a single vendor and field strength; therefore, the generalizability of the models has not been established. PURPOSE: To develop and evaluate an accelerated cardiac T1 mapping approach based on MyoMapNet, a convolution neural network T1 estimator that can be used across different vendors and field strengths by incorporating the relevant scanner information as additional inputs to the model. STUDY TYPE: Retrospective, multicenter. POPULATION: A total of 1423 patients with known or suspected cardiac disease (808 male, 57 ± 16 years), from three centers, two vendors (Siemens, Philips), and two field strengths (1.5 T, 3 T). The data were randomly split into 60% training, 20% validation, and 20% testing. FIELD STRENGTH/SEQUENCE: A 1.5 T and 3 T, Modified Look-Locker inversion recovery (MOLLI) for native and postcontrast T1 . ASSESSMENT: Scanner-independent MyoMapNet (SI-MyoMapNet) was developed by altering the deep learning (DL) architecture of MyoMapNet to incorporate scanner vendor and field strength as inputs. Epicardial and endocardial contours and blood pool (by manually drawing a large region of interest in the blood pool) of the left ventricle were manually delineated by three readers, with 2, 8, and 9 years of experience, and SI-MyoMapNet myocardial and blood pool T1 values (calculated from four T1 w images) were compared with conventional MOLLI T1 values (calculated from 8 to 11 T1 w images). STATISTICAL TESTS: Equivalency test with 95% confidence interval (CI), linear regression slope, Pearson correlation coefficient (r), Bland-Altman analysis. RESULTS: The proposed SI-MyoMapNet successfully created T1 maps. Native and postcontrast T1 values measured from SI-MyoMapNet were strongly correlated with MOLLI, despite using only four T1 w images, at both field-strengths and vendors (all r > 0.86). For native T1 , SI-MyoMapNet and MOLLI were in good agreement for myocardial and blood T1 values in institution 1 (myocardium: 5 msec, 95% CI [3, 8]; blood: -10 msec, 95%CI [-16, -4]), in institution 2 (myocardium: 6 msec, 95% CI [0, 11]; blood: 0 msec, [-18, 17]), and in institution 3 (myocardium: 7 msec, 95% CI [-8, 22]; blood: 8 msec, [-14, 30]). Similar results were observed for postcontrast T1 . DATA CONCLUSION: Inclusion of field strength and vendor as additional inputs to the DL architecture allows generalizability of MyoMapNet across different vendors or field strength. EVIDENCE LEVEL: 2. TECHNICAL EFFICACY: Stage 2.

Radiomics and deep learning for myocardial scar screening in hypertrophic cardiomyopathy.

BACKGROUND: Myocardial scar burden quantified using late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR), has important prognostic value in hypertrophic cardiomyopathy (HCM). However, nearly 50% of HCM patients have no scar but undergo repeated gadolinium-based CMR over their life span. We sought to develop an artificial intelligence (AI)-based screening model using radiomics and deep learning (DL) features extracted from balanced steady state free precession (bSSFP) cine sequences to identify HCM patients without scar. METHODS: We evaluated three AI-based screening models using bSSFP cine image features extracted by radiomics, DL, or combined DL-Radiomics. Images for 759 HCM patients (50 ± 16 years, 66% men) in a multi-center/vendor study were used to develop and test model performance. An external dataset of 100 HCM patients (53 ± 14 years, 70% men) was used to assess model generalizability. Model performance was evaluated using area-under-receiver-operating curve (AUC). RESULTS: The DL-Radiomics model demonstrated higher AUC compared to DL and Radiomics in the internal (0.83 vs 0.77, p = 0.006 and 0.78, p = 0.05; n = 159) and external (0.74 vs 0.64, p = 0.006 and 0.71, p = 0.27; n = 100) datasets. The DL-Radiomics model correctly identified 43% and 28% of patients without scar in the internal and external datasets compared to 42% and 16% by Radiomics model and 42% and 23% by DL model, respectively. CONCLUSIONS: A DL-Radiomics AI model using bSSFP cine images outperforms DL or Radiomics models alone as a scar screening tool prior to gadolinium administration. Despite its potential, the clinical utility of the model remains limited and further investigation is needed to improve the accuracy and generalizability.

An Explainable Machine Learning Approach Reveals Prognostic Significance of Right Ventricular Dysfunction in Nonischemic Cardiomyopathy.

OBJECTIVES: The authors 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.

Noncontrast Cardiac Magnetic Resonance Imaging Predictors of Heart Failure Hospitalization in Heart Failure With Preserved Ejection Fraction.

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. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Multiple two-dimensional active shape model framework for right ventricular segmentation.

Segmentation of the right ventricle (RV) in MRI short axis images is very challenging due to its complex shape and various appearance among the different subjects and cross-sections. Active shape models (ASM) have shown potential for segmenting the complex structures, including the RV, through two formulations: two- and three-dimensional modeling with a reported trade-off between accuracy and complexity of each formulation. In this work, we propose a new framework for modeling the RV surface using multiple 2D contours, where information from multiple cross-sectional images are incorporated into the same model. The proposed method was tested using cardiac MRI images from 56 human subjects. Compared to a golden reference of manually delineated RV contours, the proposed method resulted in significantly lower error than (almost one half) that of the conventional 2D ASM especially at the apical slices. The mean absolute distance of the proposed method was 2.9 ± 2 mm while the conventional 2D ASM resulted in an error of 6.6 ± 4.5 mm. In addition, the computation time of the proposed method was 5 s compared to 4 ± 1 min previously reported for the 3D ASM formulation.

Machine Learning for Predicting Heart Failure Progression in Hypertrophic Cardiomyopathy.

Background: Development of advanced heart failure (HF) symptoms is the most common adverse pathway in hypertrophic cardiomyopathy (HCM) patients. Currently, there is a limited ability to identify HCM patients at risk of HF. Objectives: In this study, we present a machine learning (ML)-based model to identify individual HCM patients who are at high risk of developing advanced HF symptoms. Methods: From a consecutive cohort of HCM patients evaluated at the Tufts HCM Institute from 2001 to 2018, we extracted a set of 64 potential risk factors measured at baseline. Only patients with New York Heart Association (NYHA) functional class I/II and LV ejection fraction (LVEF) by echocardiography >35% were included. The study cohort (n = 1,427 patients) was split into three disjoint subsets: development (50%), model selection (10%), and independent validation (40%). The least absolute shrinkage and selection operator was used to select the most influential clinical variables. An ensemble of ML classifiers, including logistic regression, was used to identify patients with high risk of developing a HF outcome. Study outcomes were defined as progression to NYHA class III/IV, drop in LVEF below 35%, septal reduction procedure, and/or heart transplantation. Results: During a mean follow-up of 4.7 ± 3.7 years, advanced HF occurred in 283 (20% out of 1,427) patients. The model features included patients' sex, NYHA class (I or II), HCM type (i.e., obstructive or not), LV wall thickness, LVEF, presence of HF symptoms (e.g., dyspnea, presyncope), comorbidities (atrial fibrillation, hypertension, mitral regurgitation, and systolic anterior motion), and type of cardiac medications. The developed risk stratification model showed strong differentiation power to identify patients at advanced HF risk in the testing dataset (c-statistics = 0.81; 95% confidence interval [CI]: 0.76, 0.86). The model allowed correct identification of high-risk patients with accuracy 74% (CI: 0.70, 0.78), sensitivity 80% (CI: 0.77, 0.83), and specificity 72% (CI: 0.68, 0.76). The model performance was comparable among different sex and age groups. Conclusions: A 5-year risk prediction of progressive HF in HCM patients can be accurately estimated using ML analysis of patients' clinical and imaging parameters. A set of 17 clinical and imaging variables were identified as the most important predictors of progressive HF in HCM.

Improved Quantification of Myocardium Scar in Late Gadolinium Enhancement Images: Deep Learning Based Image Fusion Approach.

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. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: 1.

Multi-domain convolutional neural network (MD-CNN) for radial reconstruction of dynamic cardiac MRI.

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-space and 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 and provides superior image quality compared to kt-RASPS.

Deep convolution neural networks based artifact suppression in under-sampled radial acquisitions of myocardial T 1 mapping images.

We developed a deep convolutional neural network (CNN) based method to remove streaking artefact from accelerated radial acquisitions of myocardial T 1-mapping images. A deep CNN based on a modified U-Net architecture was developed and trained to remove the streaking artefacts from under-sampled T 1 mapping images. A total of 2090 T 1-weighted images for 33 patients (55 ± 15 years, 19 males) and five healthy subjects (30 ± 14 years, 2 males) were used for training and testing the network. The images were acquired using radial slice interleaved T 1 mapping sequence (STONE) and retrospectively under-sampled to achieve acceleration rate of 4 (corresponding to 48 spokes). The dataset was split into training and testing subsets with 23 subjects (60%) and 15 subjects (40%), respectively. For generating voxel-wise T 1 maps, a two-parameter fitting model was used. Network performance was evaluated using normalized mean square error (NMSE), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR) metrics. The proposed network allowed fast (<0.3 s/image) removal of the artefact from all T 1-weighted testing images and the corresponding T 1 maps with PSNR = 64.3 ± 1.02, NMSE = 0.2 ± 0.09 and SSIM = 0.9 ± 0.3 × 10-4. There was no statistically significant difference between the measured T 1 maps for both per-subject (reference: 1085 ± 37 ms, CNN: 1088 ± 37 ms, p = 0.4) and per-segment (reference: 1084 ± 48 ms, CNN: 1083 ± 58 ms, p = 0.9) analyses. In summary, deep CNN allows fast and reliable removal of streaking artefact from under-sampled radial T 1 mapping images. Our results show that the highly non-linear operations of deep CNN processing of T 1 mapping images do not impact accurate reconstruction of myocardial T 1 maps.

Automated Myocardial T2 and Extracellular Volume Quantification in Cardiac MRI Using Transfer Learning-based Myocardium Segmentation.

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. Supplemental material is available for this article. © RSNA, 2020.

Detecting liver fibrosis using a machine learning-based approach to the quantification of the heart-induced deformation in tagged MR images.

Liver disease causes millions of deaths per year worldwide, and approximately half of these cases are due to cirrhosis, which is an advanced stage of liver fibrosis that can be accompanied by liver failure and portal hypertension. Early detection of liver fibrosis helps in improving its treatment and prevents its progression to cirrhosis. In this work, we present a novel noninvasive method to detect liver fibrosis from tagged MRI images using a machine learning-based approach. Specifically, coronal and sagittal tagged MRI imaging are analyzed separately to capture cardiac-induced deformation of the liver. The liver is manually delineated and a novel image feature, namely, the histogram of the peak strain (HPS) value, is computed from the segmented liver region and is used to classify the liver as being either normal or fibrotic. Classification is achieved using a support vector machine algorithm. The in vivo study included 15 healthy volunteers (10 males; age range 30-45 years) and 22 patients (15 males; age range 25-50 years) with liver fibrosis verified and graded by transient elastography, and 10 patients only had a liver biopsy and were diagnosed with a score of F3-F4. The proposed method demonstrates the usefulness and efficiency of extracting the HPS features from the sagittal slices for patients with moderate fibrosis. Cross-validation of the method showed an accuracy of 83.7% (specificity = 86.6%, sensitivity = 81.8%).

Three-dimensional Deep Convolutional Neural Networks for Automated Myocardial Scar Quantification in Hypertrophic Cardiomyopathy: A Multicenter Multivendor Study.

Background Cardiac MRI late gadolinium enhancement (LGE) scar volume is an important marker for outcome prediction in patients with hypertrophic cardiomyopathy (HCM); however, its clinical application is hindered by a lack of measurement standardization. Purpose To develop and evaluate a three-dimensional (3D) convolutional neural network (CNN)-based method for automated LGE scar quantification in patients with HCM. Materials and Methods We retrospectively identified LGE MRI data in a multicenter (n = 7) and multivendor (n = 3) HCM study obtained between November 2001 and November 2011. A deep 3D CNN based on U-Net architecture was used for LGE scar quantification. Independent CNN training and testing data sets were maintained with a 4:1 ratio. Stacks of short-axis MRI slices were split into overlapping substacks that were segmented and then merged into one volume. The 3D CNN per-site and per-vendor performances were evaluated with respect to manual scar quantification performed in a core laboratory setting using Dice similarity coefficient (DSC), Pearson correlation, and Bland-Altman analyses. Furthermore, the performance of 3D CNN was compared with that of two-dimensional (2D) CNN. Results This study included 1073 patients with HCM (733 men; mean age, 49 years ± 17 [standard deviation]). The 3D CNN-based quantification was fast (0.15 second per image) and demonstrated excellent correlation with manual scar volume quantification (r = 0.88, P < .001) and ratio of scar volume to total left ventricle myocardial volume (%LGE) (r = 0.91, P < .001). The 3D CNN-based quantification strongly correlated with manual quantification of scar volume (r = 0.82-0.99, P < .001) and %LGE (r = 0.90-0.97, P < .001) for all sites and vendors. The 3D CNN identified patients with a large scar burden (>15%) with 98% accuracy (202 of 207) (95% confidence interval [CI]: 95%, 99%). When compared with 3D CNN, 2D CNN underestimated scar volume (r = 0.85, P < .001) and %LGE (r = 0.83, P < .001). The DSC of 3D CNN segmentation was comparable among different vendors (P = .07) and higher than that of 2D CNN (DSC, 0.54 ± 0.26 vs 0.48 ± 0.29; P = .02). Conclusion In the hypertrophic cardiomyopathy population, a three-dimensional convolutional neural network enables fast and accurate quantification of myocardial scar volume, outperforms a two-dimensional convolutional neural network, and demonstrates comparable performance across different vendors. © RSNA, 2019 Online supplemental material is available for this article.

Cardiovascular magnetic resonance feature tracking strain analysis for discrimination between hypertensive heart disease and hypertrophic cardiomyopathy.

BACKGROUND: Hypertensive heart disease (HHD) and hypertrophic cardiomyopathy (HCM) are both associated with an increased left ventricular (LV) wall thickness. Whilst LV ejection fraction is frequently normal in both, LV strain assessment could differentiate between the diseases. We sought to establish if cardiovascular magnetic resonance myocardial feature tracking (CMR-FT), an emerging method allowing accurate assessment of myocardial deformation, differentiates between both diseases. Additionally, CMR assessment of fibrosis and LV hypertrophy allowed association analyses and comparison of diagnostic capacities. METHODS: Two-hundred twenty-four consecutive subjects (53 HHD, 107 HCM, and 64 controls) underwent 1.5T CMR including native myocardial T1 mapping and late gadolinium enhancement (LGE). Global longitudinal strain (GLS) was assessed by CMR-FT (CVi42, Circle Cardiovascular Imaging Inc.). RESULTS: GLS was significantly higher in HCM patients (-14.7±3.8 vs. -16.5±3.3% [HHD], P = 0.004; or vs. -17.2±2.0% [controls], P<0.001). GLS was associated with LV mass index (HHD, R = 0.419, P = 0.002; HCM, R = 0.429, P<0.001), and LV ejection fraction (HHD, R = -0.493, P = 0.002; HCM, R = -0.329, P<0.001). In HCM patients, GLS was also associated with global native T1 (R = 0.282, P = 0.003), and LGE volume (ρ = 0.380, P<0.001). Discrimination between HHD and HCM by GLS (c = 0.639, 95% confidence interval [CI] 0.550-0.729) was similar to LV mass index (c = 0.643, 95% CI 0.556-0.731), global myocardial native T1 (c = 0.718, 95% CI 0.638-0.799), and LGE volume (c = 0.680, 95% CI 0.585-0.775). CONCLUSION: CMR-FT GLS differentiates between HHD and HCM. In HCM patients GLS is associated with myocardial fibrosis. The discriminatory capacity of CMR-FT GLS is similar to LV hypertrophy and fibrosis imaging markers.

Automated analysis of cardiovascular magnetic resonance myocardial native T1 mapping images using fully convolutional neural networks.

BACKGROUND: Cardiovascular magnetic resonance (CMR) myocardial native T1 mapping allows assessment of interstitial diffuse fibrosis. In this technique, the global and regional T1 are measured manually by drawing region of interest in motion-corrected T1 maps. The manual analysis contributes to an already lengthy CMR analysis workflow and impacts measurements reproducibility. In this study, we propose an automated method for combined myocardium segmentation, alignment, and T1 calculation for myocardial T1 mapping. METHODS: A deep fully convolutional neural network (FCN) was used for myocardium segmentation in T1 weighted images. The segmented myocardium was then resampled on a polar grid, whose origin is located at the center-of-mass of the segmented myocardium. Myocardium T1 maps were reconstructed from the resampled T1 weighted images using curve fitting. The FCN was trained and tested using manually segmented images for 210 patients (5 slices, 11 inversion times per patient). An additional image dataset for 455 patients (5 slices and 11 inversion times per patient), analyzed by an expert reader using a semi-automatic tool, was used to validate the automatically calculated global and regional T1 values. Bland-Altman analysis, Pearson correlation coefficient, r, and the Dice similarity coefficient (DSC) were used to evaluate the performance of the FCN-based analysis on per-patient and per-slice basis. Inter-observer variability was assessed using intraclass correlation coefficient (ICC) of the T1 values calculated by the FCN-based automatic method and two readers. RESULTS: The FCN achieved fast segmentation (< 0.3 s/image) with high DSC (0.85 ± 0.07). The automatically and manually calculated T1 values (1091 ± 59 ms and 1089 ± 59 ms, respectively) were highly correlated in per-patient (r = 0.82; slope = 1.01; p < 0.0001) and per-slice (r = 0.72; slope = 1.01; p < 0.0001) analyses. Bland-Altman analysis showed good agreement between the automated and manual measurements with 95% of measurements within the limits-of-agreement in both per-patient and per-slice analyses. The intraclass correllation of the T1 calculations by the automatic method vs reader 1 and reader 2 was respectively 0.86/0.56 and 0.74/0.49 in the per-patient/per-slice analyses, which were comparable to that between two expert readers (=0.72/0.58 in per-patient/per-slice analyses). CONCLUSION: The proposed FCN-based image processing platform allows fast and automatic analysis of myocardial native T1 mapping images mitigating the burden and observer-related variability of manual analysis.

Imaging sequence for joint myocardial T1 mapping and fat/water separation.

PURPOSE: To develop and evaluate an imaging sequence to simultaneously quantify the epicardial fat volume and myocardial T1 relaxation time. METHODS: We introduced a novel simultaneous myocardial T1 mapping and fat/water separation sequence (joint T1 -fat/water separation). Dixon reconstruction is performed on a dual-echo data set to generate water/fat images. T1 maps are computed using the water images, whereas the epicardial fat volume is calculated from the fat images. A phantom experiment using vials with different T1 /T2 values and a bottle of oil was performed. Additional phantom experiment using vials of mixed fat/water was performed to show the potential of this sequence to mitigate the effect of intravoxel fat on estimated T1 maps. In vivo evaluation was performed in 17 subjects. Epicardial fat volume, native myocardial T1 measurements and precision were compared among slice-interleaved T1 mapping, Dixon, and the proposed sequence. RESULTS: In the first phantom, the proposed sequence separated oil from water vials and there were no differences in T1 of the fat-free vials (P = .1). In the second phantom, the T1 error decreased from 22%, 36%, 57%, and 73% to 8%, 9%, 16%, and 26%, respectively. In vivo there was no difference between myocardial T1 values (1067 ± 17 ms versus 1077 ± 24 ms, P = .6). The epicardial fat volume was similar for both sequences (54.3 ± 33 cm3 versus 52.4 ± 32 cm3 , P = .8). CONCLUSION: The proposed sequence provides simultaneous quantification of native myocardial T1 and epicardial fat volume. This will eliminate the need for an additional sequence in the cardiac imaging protocol if both measurements are clinically indicated.

Non-contrast myocardial infarct scar assessment using a hybrid native T1 and magnetization transfer imaging sequence at 1.5T.

PURPOSE: To develop a gadolinium-free cardiac MR technique that simultaneously exploits native T1 and magnetization transfer (MT) contrast for the imaging of myocardial infarction. METHODS: A novel hybrid T one and magnetization transfer (HYTOM) method was developed based on the modified look-locker inversion recovery (MOLLI) sequence, with a train of MT-prep pulses placed before the balanced SSFP (bSSFP) readout pulses. Numerical simulations, based on Bloch-McConnell equations, were performed to investigate the effects of MT induced by (1) the bSSFP readout pulses, and (2) the MT-prep pulses, on the measured, "apparent," native T1 values. The HYTOM method was then tested on 8 healthy adult subjects, 6 patients, and a swine with prior myocardial infarction (MI). The resulting imaging contrast between normal myocardium and infarcted tissues was compared with that of MOLLI. Late gadolinium enhancement (LGE) images were also obtained for infarct assessment in patients and swine. RESULTS: Numerical simulation and in vivo studies in healthy volunteers demonstrated that MT effects, resulting from on-resonance bSSFP excitation pulses and off-resonance MT-prep pulses, reduce the measured T1 in both MOLLI and HTYOM. In vivo studies in patients and swine showed that the HYTOM sequence can identify locations of MI, as seen on LGE. Furthermore, the HYTOM method yields higher myocardium-to-scar contrast than MOLLI (contrast-to-noise ratio: 7.33 ± 1.67 vs. 3.77 ± 0.66, P < 0.01). CONCLUSION: The proposed HYTOM method simultaneously exploits native T1 and MT contrast and significantly boosts the imaging contrast for myocardial infarction.

Gray blood late gadolinium enhancement cardiovascular magnetic resonance for improved detection of myocardial scar.

BACKGROUND: Low scar-to-blood contrast in late gadolinium enhanced (LGE) MRI limits the visualization of scars adjacent to the blood pool. Nulling the blood signal improves scar detection but results in lack of contrast between myocardium and blood, which makes clinical evaluation of LGE images more difficult. METHODS: GB-LGE contrast is achieved through partial suppression of the blood signal using T2 magnetization preparation between the inversion pulse and acquisition. The timing parameters of GB-LGE sequence are determined by optimizing a cost-function representing the desired tissue contrast. The proposed 3D GB-LGE sequence was evaluated using phantoms, human subjects (n = 45) and a swine model of myocardial infarction (n = 5). Two independent readers subjectively evaluated the image quality and ability to identify and localize scarring in GB-LGE compared to black-blood LGE (BB-LGE) (i.e., with complete blood nulling) and conventional (bright-blood) LGE. RESULTS: GB-LGE contrast was successfully generated in phantoms and all in-vivo scans. The scar-to-blood contrast was improved in GB-LGE compared to conventional LGE in humans (1.1 ± 0.5 vs. 0.6 ± 0.4, P < 0.001) and in animals (1.5 ± 0.2 vs. -0.03 ± 0.2). In patients, GB-LGE detected more tissue scarring compared to BB-LGE and conventional LGE. The subjective scores of the GB-LGE ability for localizing LV scar and detecting papillary scar were improved as compared with both BB-LGE (P < 0.024) and conventional LGE (P < 0.001). In the swine infarction model, GB-LGE scores for the ability to localize LV scar scores were consistently higher than those of both BB-LGE and conventional-LGE. CONCLUSION: GB-LGE imaging improves the ability to identify and localize myocardial scarring compared to both BB-LGE and conventional LGE. Further studies are warranted to histologically validate GB-LGE.