Electronic Health Record-Based Deep Learning Prediction of Death or Severe Decompensation in Heart Failure Patients.

BACKGROUND: Surgical mechanical ventricular assistance and cardiac replacement therapies, although life-saving in many heart failure (HF) patients, remain high-risk. Despite this, the difficulty in timely identification of medical therapy nonresponders and the dire consequences of nonresponse have fueled early, less selective surgical referral. Patients who would have ultimately responded to medical therapy are therefore subjected to the risk and life disruption of surgical therapy. OBJECTIVES: The purpose of this study was to develop deep learning models based upon commonly-available electronic health record (EHR) variables to assist clinicians in the timely and accurate identification of HF medical therapy nonresponders. METHODS: The study cohort consisted of all patients (age 18 to 90 years) admitted to a single tertiary care institution from January 2009 through December 2018, with International Classification of Disease HF diagnostic coding. Ensemble deep learning models employing time-series and densely-connected networks were developed from standard EHR data. The positive class included all observations resulting in severe progression (death from any cause or referral for HF surgical intervention) within 1 year. RESULTS: A total of 79,850 distinct admissions from 52,265 HF patients met observation criteria and contributed >350 million EHR datapoints for model training, validation, and testing. A total of 20% of model observations fit positive class criteria. The model C-statistic was 0.91. CONCLUSIONS: The demonstrated accuracy of EHR-based deep learning model prediction of 1-year all-cause death or referral for HF surgical therapy supports clinical relevance. EHR-based deep learning models have considerable potential to assist HF clinicians in improving the application of advanced HF surgical therapy in medical therapy nonresponders.

Association of STS database variables with repair durability in ischemic mitral regurgitation using machine learning.

BACKGROUND: Machine learning (ML) can identify nonintuitive clinical variable combinations that predict clinical outcomes. To assess the potential predictive contribution of standardized Society of Thoracic Surgeons (STS) Database clinical variables, we used ML to detect their association with repair durability in ischemic mitral regurgitation (IMR) patients in a single institution study. METHODS: STS Database variables (n = 53) served as predictors of repair durability in ML modeling of 224 patients who underwent surgical revascularization and mitral valve repair for IMR. Follow-up mortality and echocardiography data allowed 1-year outcome analysis in 173 patients. Supervised ML analyses were performed using recurrence (≥3+ IMR) or death versus nonrecurrence (<3+ IMR) as the binary outcome classification. RESULTS: We tested standard ML and deep learning algorithms, including support vector machines, logistic regression, and deep neural networks. Following training, final models were utilized to predict class labels for the patients in the test set, producing receiver operating characteristic (ROC) curves. The three models produced similar area under the curve (AUC), and predicted class labels with promising accuracy (AUC = 0.72-0.75). CONCLUSIONS: Readily-available STS Database variables have potential to play a significant role in the development of ML models to direct durable surgical therapy in IMR patients.

Machine Learning Outcome Prediction in Dilated Cardiomyopathy Using Regional Left Ventricular Multiparametric Strain.

The clinical presentation of idiopathic dilated cardiomyopathy (IDCM) heart failure (HF) patients who will respond to medical therapy (responders) and those who will not (non-responders) is often similar. A machine learning (ML)-based clinical tool to identify responders would prevent unnecessary surgery, while targeting non-responders for early intervention. We used regional left ventricular (LV) contractile injury patterns in ML models to identify IDCM HF non-responders. MRI-based multiparametric strain analysis was performed in 178 test subjects (140 normal subjects and 38 IDCM patients), calculating longitudinal, circumferential, and radial strain over 18 LV sub-regions for inclusion in ML analyses. Patients were identified as responders based upon symptomatic and contractile improvement on medical therapy. We tested the predictive accuracy of support vector machines (SVM), logistic regression (LR), random forest (RF), and deep neural networks (DNN). The DNN model outperformed other models, predicting response to medical therapy with an area under the receiver operating characteristic curve (AUC) of 0.94. The top features were longitudinal strain in (1) basal: anterior, posterolateral and (2) mid: posterior, anterolateral, and anteroseptal sub-regions. Regional contractile injury patterns predict response to medical therapy in IDCM HF patients, and have potential application in ML-based HF patient care.

The Use of Synthetic Electronic Health Record Data and Deep Learning to Improve Timing of High-Risk Heart Failure Surgical Intervention by Predicting Proximity to Catastrophic Decompensation.

Objective: Although many clinical metrics are associated with proximity to decompensation in heart failure (HF), none are individually accurate enough to risk-stratify HF patients on a patient-by-patient basis. The dire consequences of this inaccuracy in risk stratification have profoundly lowered the clinical threshold for application of high-risk surgical intervention, such as ventricular assist device placement. Machine learning can detect non-intuitive classifier patterns that allow for innovative combination of patient feature predictive capability. A machine learning-based clinical tool to identify proximity to catastrophic HF deterioration on a patient-specific basis would enable more efficient direction of high-risk surgical intervention to those patients who have the most to gain from it, while sparing others. Synthetic electronic health record (EHR) data are statistically indistinguishable from the original protected health information, and can be analyzed as if they were original data but without any privacy concerns. We demonstrate that synthetic EHR data can be easily accessed and analyzed and are amenable to machine learning analyses. Methods: We developed synthetic data from EHR data of 26,575 HF patients admitted to a single institution during the decade ending on 12/31/2018. Twenty-seven clinically-relevant features were synthesized and utilized in supervised deep learning and machine learning algorithms (i.e., deep neural networks [DNN], random forest [RF], and logistic regression [LR]) to explore their ability to predict 1-year mortality by five-fold cross validation methods. We conducted analyses leveraging features from prior to/at and after/at the time of HF diagnosis. Results: The area under the receiver operating curve (AUC) was used to evaluate the performance of the three models: the mean AUC was 0.80 for DNN, 0.72 for RF, and 0.74 for LR. Age, creatinine, body mass index, and blood pressure levels were especially important features in predicting death within 1-year among HF patients. Conclusions: Machine learning models have considerable potential to improve accuracy in mortality prediction, such that high-risk surgical intervention can be applied only in those patients who stand to benefit from it. Access to EHR-based synthetic data derivatives eliminates risk of exposure of EHR data, speeds time-to-insight, and facilitates data sharing. As more clinical, imaging, and contractile features with proven predictive capability are added to these models, the development of a clinical tool to assist in timing of intervention in surgical candidates may be possible.

Heterogeneity of systolic dysfunction in patients with severe aortic stenosis and preserved ejection fraction.

BACKGROUND AND AIM: Left ventricular (LV) systolic strain has been shown to be an early marker of LV dysfunction in patients with severe aortic stenosis (AS) despite preserved ejection fraction (EF). Echocardiography has provided useful data on regional LV strain patterns, but is not as sensitive as magnetic resonance imaging (MRI). No prior studies have used MRI-based strain analysis to characterize regional three-dimensional strain in patients with severe AS. METHODS: Twelve patients with severe AS and preserved EF underwent MRI-based multiparametric strain analysis. Circumferential and longitudinal strain values were calculated at individual points throughout the LV and analyzed in 12 discrete regions. Strain values were compared to a database of normal controls. RESULTS: Compared to control patients, circumferential strain in AS patients was significantly reduced at the base (P = 0.002), mid (P = 0.042), and inferior walls (P < 0.001). Longitudinal strain was significantly reduced at the base (P < 0.001), mid (P < 0.001), anterior (P < 0.001), and septal (P < 0.001) walls. Among patients with AS, there was heterogeneity in the location and severity of abnormalities in circumferential and longitudinal strains despite the presence of a preserved EF and lack of prior myocardial infarction. CONCLUSIONS: LV systolic strain is significantly impaired in patients with AS and preserved EF compared to healthy volunteers. Abnormalities in circumferential and longitudinal strains were heterogeneously distributed across the LV of patients with AS, allowing us to identify sentinel regions that may reflect the earliest signs of developing LV dysfunction.

Topographic mapping of left ventricular regional contractile injury in ischemic mitral regurgitation.

OBJECTIVE: Restrictive leaflet tethering resulting from regional left ventricular (LV) contractile injury causes ischemic mitral regurgitation (MR). We hypothesized that 3-dimensional LV topographic mapping by MRI-based multiparametric strain analysis could characterize the regional contractile injury patterns that differentiate ischemic coronary artery disease patients who have ischemic MR from those who do not. METHODS: Magnetic resonance imaging-based multiparametric strain data were calculated for 15,300 LV grid points in 100 normal volunteers. Strain parameters from ischemic MR (n = 10) and ischemic no-MR (n = 36) patients were then normalized to this normal human strain database with z score quantification of standard deviation from the normal mean. Mean multiparametric strain z scores were calculated for 18 LV subregions (basilar/mid/apical levels; 6 LV regions). Mean strain z scores for papillary muscle-related (basilar/mid levels of anterolateral, posterolateral, and posterior) and nonpapillary muscle-related (all other) subregions were compared between ischemic MR and ischemic no-MR groups. RESULTS: Across all patients, contractile injury was greater in the papillary muscle-related regions compared with the nonpapillary regions (P = .007). In the papillary regions, contractile injury was greater in the ischemic MR group compared with the no-MR group (z scores, 1.91 ± 1.13 vs 1.20 ± 1.01, respectively; P < .001). Strain values in the nonpapillary muscle-related subregions were not different between the 2 groups (1.31 ± 1.04 vs 1.20 ± 1.03; P = .301). CONCLUSIONS: Multiparametric strain analysis demonstrated severe normalized contractile injury in the papillary muscle-related LV subregions in patients with ischemic MR. The mean degree of normalized injury approached 2 standard deviations and was significantly worse than the levels seen in ischemic no-MR patients.

Dilated Cardiomyopathy: Normalized Multiparametric Myocardial Strain Predicts Contractile Recovery.

BACKGROUND: Left ventricular contractile injury in dilated cardiomyopathy (DCM) may occur in a consistently heterogeneous distribution, suggesting that early-injury sentinel regions may have prognostic significance. Heightened surveillance of these regions with high-resolution contractile metrics may predict recovery in DCM. METHODS: Multiple three-dimensional strain parameters were calculated at each of 15,300 left ventricular grid points from systolic displacement data obtained from cardiac magnetic resonance imaging in 124 test subjects. In 24 DCM patients, Z-scores for two strain parameters at each grid point were calculated by comparison of patient-specific strain values to respective point-specific mean and standard deviation values from a normal human strain database (n = 100). Multiparametric strain Z-scores were averaged over six left ventricular regions at basilar, mid, and apical levels (18 subregions). Patients with DCM were stratified into three groups on the basis of a blinded review of clinical contractile recovery (complete, n = 7; incomplete, n = 7; none, n = 10). RESULTS: Basilar-septal subregions were consistently heavily injured. Basilar-septal Z-scores were significantly larger (worse) than those for the rest of the left ventricle (2.73 ± 1.27 versus 2.22 ± 0.83; p = 0.011) and lateral wall (2.73 ± 1.27 versus 1.44 ± 0.72; p < 0.001). All patients with sentinel region average multiparametric strain Z-scores less than two standard deviations (n = 6) experienced complete recovery, whereas 17 of 18 DCM patients with Z-scores greater than two standard deviations experienced incomplete or no contractile recovery. CONCLUSIONS: Contractile injury in DCM is heterogeneous, with basilar-septal regions injured more than lateral regions. The targeting of early-injury sentinel regions for heightened surveillance with high-resolution metrics of microregional contractile function may accurately predict recovery on medical therapy. A two standard deviation Z-score threshold may predict contractile recovery.

Quantifying "normalized" regional left ventricular contractile function in ischemic coronary artery disease.

OBJECTIVE: When significant coronary lesions are identified by angiography, regional left ventricular (LV) contractile function often plays a role in determining candidacy for revascularization. To improve on current subjective and nonquantitative metrics of regional LV function, we tested a z-score "normalization" of regional strain information quantified from clinically acquired high-resolution LV geometric datasets. METHODS: Test subjects (n = 120) underwent cardiac MRI with multiple 3-dimensional strain parameters calculated from tissue tag-plane displacement data. Sixty healthy volunteers contributed strain parameter data at each of 15,300 LV grid points, to form a normal human strain database. Point-specific database comparisons were made in 60 patients who had documented coronary artery disease (CAD), by angiography. Patient-specific, color-coded 3-dimensional LV maps of z-score-normalized contractile function were generated. RESULTS: Blinded clinical review indicated that 55% (33 of 60) of the patients with CAD had significant regional contractile abnormalities by 1 of 3 "gold-standard" criteria: (1) Q waves on electrocardiography (ECG); (2) infarct on radionuclide single-photon emission computed tomography (SPECT); or (3) akinesia or dyskinesia on echocardiography. Consistency among all gold-standard metrics was found for only 19% (6 of 31) of patients with CAD who had ≥2 available metrics. Blinded MRI-based, multiparametric, strain z-score localization of contractile abnormalities was accurate in 89% (ECG), 97% (SPECT), and 95% (echocardiography). CONCLUSIONS: Nonsubjective normalization of regional LV contractile function by z-score calculation from a normal human strain database can localize and quantitatively display regional wall motion abnormalities in patients with CAD. This high-resolution localization of regional wall motion abnormalities may help improve the accuracy of therapeutic intervention in patients who have CAD.

Three-dimensional regional strain computation method with displacement encoding with stimulated echoes (DENSE) in non-ischemic, non-valvular dilated cardiomyopathy patients and healthy subjects validated by tagged MRI.

PURPOSE: Fast cine displacement encoding with stimulated echoes (DENSE) MR has higher spatial resolution and enables rapid postprocessing. Thus we compared the accuracy of regional strains computation by DENSE with tagged MR in healthy and non-ischemic, non-valvular dilated cardiomyopathy (DCM) subjects. MATERIALS AND METHODS: Validation of three-dimensional regional strains computed with DENSE was conducted in reference to standard tagged MRI (TMRI) in healthy subjects and patients with DCM. Additional repeatability studies in healthy subjects were conducted to increase confidence in DENSE. A meshfree multiquadrics radial point interpolation method (RPIM) was used for computing Lagrange strains in sixteen left ventricular segments. Bland-Altman analysis and Student's t-tests were conducted to observe similarities in regional strains between sequences and in DENSE repeatability studies. RESULTS: Regional circumferential strains ranged from -0.21 ± 0.07 (Lateral-Apex) to -0.11 ± 0.05 (Posterorseptal-Base) in healthy subjects and -0.15 ± 0.04 (Anterior-Apex) to -0.02 ± 0.08 (Posterorseptal-Base) in DCM patients. Computed mean differences in regional circumferential strain from the DENSE-TMRI comparison study was 0.01 ± 0.03 (95% limits of agreement) in normal subjects, -0.01 ± 0.06 in DCM patients and 0.0 ± 0.02 in repeatability studies, with similar agreements in longitudinal and radial strains. CONCLUSION: We found agreement between DENSE and tagged MR in patients and volunteers in terms of evaluation of regional strains.

Early left ventricular regional contractile impairment in chronic mitral regurgitation occurs in a consistent, heterogeneous pattern.

OBJECTIVES: The clinical guidelines for asymptomatic patients with chronic mitral regurgitation (MR) use the ejection fraction (EF) to trigger surgical referral. We hypothesized that the EF is not sensitive enough to detect the earliest contractile injury in chronic MR and that the injury associated with chronic MR is not global but heterogeneous, occurring regionally and predictably, before the onset of global left ventricular (LV) dysfunction. METHODS: Fifteen patients with chronic MR and normal LVEF by echocardiography underwent cardiac magnetic resonance imaging with tissue tagging. Point-specific comparisons (at 15,300 LV grid points) of multiple strain parameters to a normal human strain database allowed normalization of patient-specific regional contractile function. Data were mapped over patient-specific 3-dimensional geometry and averaged across 6 LV regions. RESULTS: Global LV longitudinal and circumferential myocardial strains were normal for all 15 patients with MR compared with normal controls (P>.05). Despite preserved global function, the anteroseptum and posteroseptum demonstrated significantly worse contractile function compared with other LV regions (P=.003 and P=.035, respectively). Hypercontractile regions (lateral walls) appeared to compensate (P=.002) for the reduced septal contractile function, masking injury detection by global indexes. CONCLUSIONS: The earliest contractile injury seen in patients with MR is heterogeneous and consistently distributed along the LV septum. Compensatory responses include hypercontractility of other regions. These data suggest that rather than relying on global LV contractile metrics, which cannot detect early injury, patients might be better served by undergoing directed surveillance of "sentinel" LV regions (LV septum) with high-resolution metrics of regional contractile function.

A validation of two-dimensional in vivo regional strain computed from displacement encoding with stimulated echoes (DENSE), in reference to tagged magnetic resonance imaging and studies in repeatability.

Fast cine displacement encoding with stimulated echoes (DENSE) has comparative advantages over tagged MRI (TMRI) including higher spatial resolution and faster post-processing. This study computed regional radial and circumferential myocardial strains with DENSE displacements and validated it in reference to TMRI, according to American Heart Association (AHA) guidelines for standardized segmentation of regions in the left ventricle (LV). This study was therefore novel in examining agreement between the modalities in 16 AHA recommended LV segments. DENSE displacements were obtained with spatiotemporal phase unwrapping and TMRI displacements obtained with a conventional tag-finding algorithm. A validation study with a rotating phantom established similar shear strain between modalities prior to in vivo studies. A novel meshfree nearest node finite element method (NNFEM) was used for rapid computation of Lagrange strain in both phantom and in vivo studies in both modalities. Also novel was conducting in vivo repeatability studies for observing recurring strain patterns in DENSE and increase confidence in it. Comprehensive regional strain agreements via Bland-Altman analysis between the modalities were obtained. Results from the phantom study showed similar radial-circumferential shear strains from the two modalities. Mean differences in regional in vivo circumferential strains were -0.01 ± 0.09 (95% limits of agreement) from comparing the modalities and -0.01 ± 0.07 from repeatability studies. Differences and means from comparison and repeatability studies were uncorrelated (p > 0.05) indicating no increases in differences with increased strain magnitudes. Bland-Altman analysis and similarities in regional strain distribution within the myocardium showed good agreements between DENSE and TMRI and show their interchangeability. NNFEM was also established as a common framework for computing strain in both modalities.

MRI-based multiparametric strain analysis predicts contractile recovery after aortic valve replacement for aortic insufficiency.

BACKGROUND: Guidelines for referral of chronic aortic insufficiency (AI) patients for aortic valve replacement (AVR) suggest that surgery can be delayed until symptoms or reduction in left ventricular (LV) contractile function occur. The frequent occurrence of reduced LV contractile function after AVR for chronic AI suggests that new contractile metrics for surgical referral are needed. METHODS: In 16 chronic AI patients, cardiac MRI tagged images were analyzed before and 21.5 ± 13.8 months after AVR to calculate LV systolic strain. Average measurements of three strain parameters were obtained for each of 72 LV regions, normalized using a normal human strain database (n = 63), and combined into a composite index (multiparametric strain z score [MSZ]) representing standard deviation from the normal regional average. RESULTS: Preoperative global MSZ (72-region average) correlated with post-AVR global MSZ (R(2) = 0.825, p < 0.001). Preoperative global MSZ also predicts improvement of impaired regions (N = 271 regions from 14 AI patients, R(2) = 0.392, p < 0.001). Preoperative MRI-based LV ejection fraction (LVEF) is also predictive (r = 0.410, p < 0.001). Although global preoperative MSZ had a significantly higher correlation than preoperative LVEF with improvement of injured regions (p < 0.001), both measures convey the same phenomenon. CONCLUSIONS: Global preoperative MRI-based multiparametric strain predicts global strain postoperatively, as well as improvement of regions (n = 72 per LV) with impaired contractile function. Global contractile function is an important correlate with improvement in regionally impaired contractile function, perhaps reflecting total AI volume-overload burden (severity/duration of AI).

Heterogeneous distribution of left ventricular contractile injury in chronic aortic insufficiency.

BACKGROUND: Global systolic strain has been described previously in patients with chronic aortic insufficiency (AI). This study explored regional differences in contractile injury. METHODS: Tagged magnetic resonance images of the left ventricle (LV) were acquired and analyzed to calculate systolic strain in 42 patients with chronic AI. Multiparametric systolic strain analysis was applied to relate cardiac function in AI patients to a normal strain database (N = 60). AI patients were classified as having normal or poor function based on their results. A two-way repeated-measures analysis of variance was applied to analyze regional differences in injury. RESULTS: The mean and standard deviation of raw strain values (circumferential strain, longitudinal strain, and minimum principal strain angle) are presented over the entire LV in our normal strain database. Of the 42 patients with AI, 15 could be defined as having poor function by multiparametric systolic strain analysis. In AI patients with poor function, statistical analysis showed significant differences in injury between standard LV regions (F(3.69,44.33) = 3.47, p = 0.017) and levels (F(1.49,17.88) = 4.41, p = 0.037) of the LV, whereas no significant differences were seen in the group with normal cardiac function. CONCLUSIONS: Patients with poor function, as defined by multiparametric systolic strain z scores, exhibit a consistent, heterogeneous pattern of contractile injury in which the septum and posterior regions at the base are most injured.

Differential modulation of right ventricular strain and right atrial mechanics in mild vs. severe pressure overload.

Increased right atrial (RA) and ventricular (RV) chamber volumes are a late maladaptive response to chronic pulmonary hypertension. The purpose of the current investigation was to characterize the early compensatory changes that occur in the right heart during chronic RV pressure overload before the development of chamber dilation. Magnetic resonance imaging with radiofrequency tissue tagging was performed on dogs at baseline and after 10 wk of pulmonary artery banding to yield either mild RV pressure overload (36% rise in RV pressure; n = 5) or severe overload (250% rise in RV pressure; n = 4). The RV free wall was divided into three segments within a midventricular plane, and circumferential myocardial strain was calculated for each segment, the septum, and the left ventricle. Chamber volumes were calculated from stacked MRI images, and RA mechanics were characterized by calculating the RA reservoir, conduit, and pump contribution to RV filling. With mild RV overload, there were no changes in RV strain or RA function. With severe RV overload, RV circumferential strain diminished by 62% anterior (P = 0.04), 42% inferior (P = 0.03), and 50% in the septum (P = 0.02), with no change in the left ventricle (P = 0.12). RV filling became more dependent on RA conduit function, which increased from 30 ± 9 to 43 ± 13% (P = 0.01), than on RA reservoir function, which decreased from 47 ± 6 to 33 ± 4% (P = 0.04), with no change in RA pump function (P = 0.94). RA and RV volumes and RV ejection fraction were unchanged from baseline during either mild (P > 0.10) or severe RV pressure overload (P > 0.53). In response to severe RV pressure overload, RV myocardial strain is segmentally diminished and RV filling becomes more dependent on RA conduit rather than reservoir function. These compensatory mechanisms of the right heart occur early in chronic RV pressure overload before chamber dilation develops.

Magnetic resonance imaging detects significant sex differences in human myocardial strain.

BACKGROUND: The pathophysiology responsible for the significant outcome disparities between men and women with cardiac disease is largely unknown. Further investigation into basic cardiac physiological differences between the sexes is needed. This study utilized magnetic resonance imaging (MRI)-based multiparametric strain analysis to search for sex-based differences in regional myocardial contractile function. METHODS: End-systolic strain (circumferential, longitudinal, and radial) was interpolated from MRI-based radiofrequency tissue tagging grid point displacements in each of 60 normal adult volunteers (32 females). RESULTS: The average global left ventricular (LV) strain among normal female volunteers (n = 32) was significantly larger in absolute value (functionally better) than in normal male volunteers (n = 28) in both the circumferential direction (Male/Female = -0.19 ± 0.02 vs. -0.21 ± 0.02; p = 0.025) and longitudinal direction (Male/Female = -0.14 ± 0.03 vs. -0.16 ± 0.02; p = 0.007). CONCLUSIONS: The finding of significantly larger circumferential and longitudinal LV strain among normal female volunteers suggests that baseline contractile differences between the sexes may contribute to the well-recognized divergence in cardiovascular disease outcomes. Further work is needed in order to determine the pathologic changes that occur in LV strain between women and men with the onset of cardiovascular disease.

Regional myocardial contractile function: multiparametric strain mapping.

Magnetic resonance imaging (MRI) with tissue tagging enables the quantification of multiple strain indices that can be combined through normalization into a single multiparametric index of regional myocardial contractile function. The aim of this study was to test the ability of multiparametric strain analysis to quantify regional differences in contractile function in an ovine model of myocardial injury. Regional variance in myocardial contractile function was induced in eight sheep by the ligation of the blood supply to the anterior and apical left ventricular (LV) myocardial walls. LV systolic strain was obtained from tissue tagged MRI images. A normal strain database (n=50) defines all parameters of systolic strain and allows normalization of regional function at 15,300 LV points by calculation of a z-score. Multiparametric systolic strain z-scores were therefore determined for 15,300 points in each injured sheep left ventricle. Multiparametric z-scores were found to vary significantly by region (P<0.001). z-Scores in regions remote to the infarct were found to be significantly smaller than those in the regions most likely to include infarcted myocardium. In this pre-clinical evaluation of MRI-based multiparametric strain analysis, it accurately quantified and visually defined regional differences in myocardial contractile function.

Magnetic resonance imaging-based multiparametric systolic strain analysis and regional contractile heterogeneity in patients with dilated cardiomyopathy.

BACKGROUND: Myocardial systolic strain patterns in dilated cardiomyopathy are considered non-homogeneous but have not been investigated with magnetic resonance imaging (MRI)-based multiparametric systolic strain analysis. Left ventricular (LV) 3-dimensional (3D) multiparametric systolic strain analysis is sensitive to regional contractility and is generated from sequential MRI of tissue-tagging gridline-point displacements. METHODS: Sixty normal human volunteers underwent MRI-based 3D systolic strain analysis to supply normal average and standard deviation values for each of three strain parameters at each of 15,300 individual LV grid-points. Patient-specific multiparametric systolic strain data from each dilated cardiomyopathy patient (n = 10) were then subjected to a point-by-point comparison (n = 15,300 LV points) to the normal strain database for three individual strain components (45,900 database comparisons per patient). The resulting composite multiparametric Z-score values (standard deviation from normal average) were color contour mapped over patient-specific 3D LV geometry to detect the normalized regional contractile patterns associated with dilated cardiomyopathy. RESULTS: Average multiparametric strain Z-score values varied significantly according to ventricular level (p = 0.001) and region (p = 0.003). Apical Z-scores were significantly less than those in both the base (p = 0.037) and mid-ventricle (p = 0.002), whereas anterolateral wall Z-scores were less than those in the anteroseptal (p = 0.023) and posteroseptal walls (p = 0.028). CONCLUSIONS: MRI-based multiparametric systolic strain analysis suggests that myocardial systolic strain in patients with dilated cardiomyopathy has a heterogeneous regional distribution and, on average, falls almost 2 standard deviations from normal.

Myocardial viability mapping by magnetic resonance-based multiparametric systolic strain analysis.

BACKGROUND: Regional myocardial contractility can be characterized by three-dimensional left ventricular (LV) multiparametric strain maps generated from sequential magnetic resonance imaging of radiofrequency tissue-tagging grid point displacements. METHODS: Normal average and standard deviation values for each of three strain indices at 15,300 LV points were determined from a normal volunteer human strain database (n = 50) by application of magnetic resonance-based three-dimensional strain analysis. Patient-specific multiparametric strain data from each ischemic cardiomyopathy patient (n = 20) were then submitted to a point-by-point comparison (n = 15,300 LV points) to the normal strain database. The resulting 15,300 composite multiparametric Z-score values (standard deviation from normal average) were color-contour mapped over patient-specific three-dimensional LV geometry to detect the abnormal contractile patterns associated with myocardial infarction and nonviable myocardium. RESULTS: The average multiparametric strain composite Z-score from each LV region (n = 120) was compared with the respective clinical standard viability testing result and used to construct a receiver-operator characteristic curve. The area under the curve was 0.941 (p < 0.001; 95% confidence interval: 0.897 to 0.985). A regional average Z-score threshold of 1.525 (> 1.525 being nonviable) resulted in a sensitivity of 90% and a specificity of 90%. Corresponding positive and negative predictive values were 84% and 95%, respectively. CONCLUSIONS: The clinical application of magnetic resonance-based multiparametric strain analysis allowed accurate regional characterization and visualization of LV myocardial viability.

Left ventricular wall stress in patients with severe aortic insufficiency with finite element analysis.

BACKGROUND: Severe aortic insufficiency (AI) with preserved left ventricular (LV) function may be associated with a long asymptomatic period and unpredictable course on medical therapy. Since myocardial wall stress is closely related to both pathologic cardiac remodeling and ultimately to LV decompensation, a more accurate description of regional wall stress may improve our ability to appropriately manage these patients. The objective of this study was to define differences in instantaneous global and regional three-dimensional end-systolic maximum principal stress (ESS) between normal patients and patients with AI, both before and after aortic valve replacement (AVR) using magnetic resonance imaging (MRI) and finite element analysis (FEA). METHODS: Magnetic resonance imaging was performed on 20 normal volunteers and 14 patients with moderate to severe AI with normal systolic function (ejection fraction: 57 +/- 0.6) before and after AVR. Finite element analysis was utilized to estimate global and regional ESS. RESULTS: Both global (p < 0.001) and regional (p < 0.001 in all segments) ESS were significantly higher in the preoperative AI patients when compared with their postoperative values and normal controls. Postoperative ESS was significantly lower than the normal controls (p = 0.002). CONCLUSIONS: Three-dimensional regional and global end-systolic LV wall stress can be determined by MRI and finite element analysis. Values of ESS in patients with chronic AI were elevated prior to AVR and normalized after AVR. This method may have considerable potential as a noninvasive, clinically applicable index of regional LV geometry and function that may help with the serial evaluation of patients with AI.