Echocardiographic Detection of Regional Wall Motion Abnormalities Using Artificial Intelligence Compared to Human Readers

BACKGROUND

Although regional wall motion abnormality (RWMA) detection is foundational to transthoracic echocardiography, current methods are prone to interobserver variability. We aimed to develop a deep learning (DL) model for RWMA assessment and compare it to expert and novice readers.

METHODS

We used 15,746 transthoracic echocardiography studies-including 25,529 apical videos-which were split into training, validation, and test datasets. A convolutional neural network was trained and validated using apical 2-, 3-, and 4-chamber videos to predict the presence of RWMA in 7 regions defined by coronary perfusion territories, using the ground truth derived from clinical transthoracic echocardiography reports. Within the test cohort, DL model accuracy was compared to 6 expert and 3 novice readers using F1 score evaluation, with the ground truth of RWMA defined by expert readers. Significance between the DL model and novices was assessed using the permutation test.

RESULTS

Within the test cohort, the DL model accurately identified any RWMA with an area under the curve of 0.96 (0.92-0.98). The mean F1 scores of the experts and the DL model were numerically similar for 6 of 7 regions: anterior (86 vs 84), anterolateral (80 vs 74), inferolateral (83 vs 87), inferoseptal (86 vs 86), apical (88 vs 87), inferior (79 vs 81), and any RWMA (90 vs 94), respectively, while in the anteroseptal region, the F1 score of the DL model was lower than the experts (75 vs 89). Using F1 scores, the DL model outperformed both novices 1 (P = .002) and 2 (P = .02) for the detection of any RWMA.

CONCLUSIONS

Deep learning provides accurate detection of RWMA, which was comparable to experts and outperformed a majority of novices. Deep learning may improve the efficiency of RWMA assessment and serve as a teaching tool for novices.

Limitations of Apical Sparing Pattern in Cardiac Amyloidosis: A Multicenter Echocardiographic Study

BACKGROUND

Although impaired left ventricular (LV) global longitudinal strain (GLS) with apical sparing is a feature of cardiac amyloidosis (CA), its diagnostic accuracy has varied across studies. We aimed to determine the ability of apical sparing ratio (ASR) and most common echocardiographic parameters to differentiate patients with confirmed CA from those with clinical and/or echocardiographic suspicion of CA, but with this diagnosis ruled out.

METHODS

We identified 544 patients with confirmed CA and 200 controls as defined above (CTRL Patients). Measurements from transthoracic echocardiograms (TTE) were performed using artificial intelligence software (Us2.AI, Singapore) and audited by an experienced echocardiographer. Receiver-operating characteristic curve analysis was used to evaluate the diagnostic performance and optimal cutoffs for the differentiation of CA patients from CTRL Patients. Additionally, a group of 174 healthy subjects (Healthy CTRL) was included to provide insight on how Patients and Healthy controls differed echocardiographically.

RESULTS

LV GLS was more impaired (-13.9 ± 4.6% vs -15.9 ± 2.7%, p < 0.0005) and ASR was higher (2.4 ± 1.2 vs 1.7 ± 0.9, p < 0.0005) in the CA group vs. CTRL Patients. Relative wall thickness and ASR were the most accurate parameters for differentiating CA from CTRL Patients (AUC: 0.77 and 0.74, respectively). However, even with the optimal cutoff of 1.67, ASR was only 72% sensitive and 66% specific for CA, indicating presence of apical sparing in 32% of CTRL Patients and even in 6% Healthy CTRLs.

CONCLUSIONS

Apical sparing did not prove to be a CA-specific biomarker for accurate identification of CA, when compared to clinically similar controls with no CA.

Noninvasive assessment of left ventricular end-diastolic pressure using machine learning-derived phasic left atrial strain

AIMS

While transthoracic echocardiography (TTE) assessment of left ventricular end-diastolic pressure (LVEDP) is critically important, the current paradigm is subject to error and indeterminate classification. Recently, peak left atrial strain (LAS) was found to be associated with LVEDP. We aimed to test the hypothesis that integration of the entire LAS time curve into a single parameter could improve the accuracy of peak LAS in the noninvasive assessment of LVEDP with TTE.

METHODS AND RESULTS

We retrospectively identified 294 patients who underwent left heart catheterization and TTE within 24 h. LAS curves were trained using machine learning (100 patients) to detect LVEDP ≥ 15 mmHg, yielding the novel parameter LAS index (LASi). The accuracy of LASi was subsequently validated (194 patients), side by side with peak LAS and ASE/EACVI guidelines, against invasive filling pressures. Within the validation cohort, invasive LVEDP was elevated in 116 (59.8%) patients. The overall accuracy of LASi, peak LAS, and American Society of Echocardiography/European Association for Cardiovascular Imaging (ASE/EACVI) algorithm was 79, 75, and 76%, respectively (excluding 37 patients with indeterminate diastolic function by ASE/EACVI guidelines). When the number of LASi indeterminates (defined by near-zero LASi values) was matched to the ASE/EACVI guidelines (n = 37), the accuracy of LASi improved to 87%. Importantly, among the 37 patients with ASE/EACVI-indeterminate diastolic function, LASi had an accuracy of 81%, compared with 76% for peak LAS.

CONCLUSION

LASi allows the detection of elevated LVEDP using invasive measurements as a reference, at least as accurately as peak LAS and current diastolic function guideline algorithm, with the advantage of no indeterminate classifications in patients with measurable LAS.

Preservation of Circumferential and Radial Left Ventricular Function as a Mitigating Mechanism for Impaired Longitudinal Strain in Early Cardiac Amyloidosis

BACKGROUND

In patients with cardiac amyloidosis (CA), left ventricular ejection fraction (LVEF) is frequently preserved, despite commonly reduced global longitudinal strain (GLS). We hypothesized that nonlongitudinal contraction may initially serve as a mitigating mechanism to maintain cardiac output and studied the relationship between global circumferential (GCS) and radial (GRS) strain with LVEF and extracellular volume (ECV), a marker of amyloid burden.

METHODS

Patients with CA who underwent cardiac magnetic resonance (CMR; n = 140, 70.7 ± 11.5 years, 66% male) or echocardiography (n = 67, 71 ± 13 years, 66% male) and normal controls (CMR, n = 20; echocardiography, n = 45) were retrospectively identified, and GCS, GLS, and GRS were quantified using feature-tracking CMR or speckle-tracking echocardiography and compared between CA patients with preserved and reduced LVEF (CAHFpEF, CAHFrEF) and controls. The prevalence of impaired strain (magnitudes <2.5th percentile of the controls) was compared between CAHFpEF and CAHFrEF and between ECV quartiles.

RESULTS

While echocardiography-derived GLS was impaired in both CAHFpEF (-13.4% ± 3.1%, P < .003) and CAHFrEF (-9.1% ± 3.2%, P < .003), compared with controls (-20.8% ± 2.4%), GCS was more impaired in CAHFrEF compared with both controls (-15.6% ± 5.0% vs -32.3% ± 3.3%, P < .003) and CAHFpEF (-30.4% ± 5.7%, P < .003) and did not differ between CAHFpEF and controls (P = .24). The prevalence of abnormal CMR-derived GCS (P < .0001) and GRS (P < .0001) but not GLS (P = .054) varied significantly across ECV quartiles.

CONCLUSIONS

Among CA patients with preserved LVEF, preserved GCS and GRS, despite near-universally impaired GLS, may be explained by an initial predominantly subendocardial involvement, where mostly longitudinal fibers are located. If confirmed in future studies, these findings may facilitate identification of patients with early stages of CA, when treatments may be most effective.

The Sword and the Crown: Echocardiography for the Detection of a Rare Combination of Congenital Heart Disease

•Unexplained right heart enlargement should always prompt evaluation for shunts. •Scimitar syndrome occurs when the right pulmonary veins drain into the IVC. •Coronary sinus dilation should prompt suspicion for PLSVC. •PLSVC can be diagnosed with a bubble study through left-sided IV.

Cardiovascular computed tomography for the detection of quadricuspid aortic valve: A case report

Aortic regurgitation (AR) represents a significant cause of morbidity and mortality. Due to its low cost and widespread availability, echocardiography remains the frontline for aortic valve (AV) assessment. However, poor sonographic windows may limit the assessment of valve morphology with this technique. Cardiovascular computed tomography (CCT) is increasingly utilized prior to structural AV interventions. Due to its excellent spatial resolution, CCT provides exceptional characterization of aortic leaflets. Accordingly, we present a case of a quadricuspid valve diagnosed by CCT. Here, CCT led to a new diagnosis of quadricuspid valve, highlighting the potential for CCT for the characterization of aortic leaflet morphology. CCT may be particularly useful in patients with contraindications to transesophageal echocardiography or those undergoing structural or robotic interventions.

Normative values of the aortic valve area and Doppler measurements using two-dimensional transthoracic echocardiography: results from the Multicentre World Alliance of Societies of Echocardiography Study

AIMS

Aortic valve area (AVA) used for echocardiographic assessment of aortic stenosis (AS) has been traditionally interpreted independently of sex, age and race. As differences in normal values might impact clinical decision-making, we aimed to establish sex-, age- and race-specific normative values for AVA and Doppler parameters using data from the World Alliance Societies of Echocardiography (WASE) Study.

METHODS AND RESULTS

Two-dimensional transthoracic echocardiographic studies were obtained from 1903 healthy adult subjects (48% women). Measurements of the left ventricular outflow tract (LVOT) diameter and Doppler parameters, including AV and LVOT velocity time integrals (VTIs), AV mean pressure gradient, peak velocity, were obtained according to ASE/EACVI guidelines. AVA was calculated using the continuity equation. Compared with men, women had smaller LVOT diameters and AVA values, and higher AV peak velocities and mean gradients (all P < 0.05). LVOT and AV VTI were significantly higher in women (P < 0.05), and both parameters increased with age in both sexes. AVA differences persisted after indexing to body surface area. According to the current diagnostic criteria, 13.5% of women would have been considered to have mild AS and 1.4% moderate AS. LVOT diameter and AVA were lower in older subjects, both men and women, and were lower in Asians, compared with whites and blacks.

CONCLUSION

WASE data provide clinically relevant information about significant differences in normal AVA and Doppler parameters according to sex, age, and race. The implementation of this information into clinical practice should involve development of specific normative values for each ethnic group using standardized methodology.

Artificial intelligence based left ventricular ejection fraction and global longitudinal strain in cardiac amyloidosis

BACKGROUND

Assessment of left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) plays a key role in the diagnosis of cardiac amyloidosis (CA). However, manual measurements are time consuming and prone to variability. We aimed to assess whether fully automated artificial intelligence (AI) calculation of LVEF and GLS provide similar estimates and can identify abnormalities in agreement with conventional manual methods, in patients with pre-clinical and clinical CA.

METHODS

We identified 51 patients (age 80 ± 10 years, 53% male) with confirmed CA according to guidelines, who underwent echocardiography before and/or at the time of CA diagnosis (median (IQR) time between observations 3.87 (1.93, 5.44 years). LVEF and GLS were quantified from the apical 2- and 4-chamber views using both manual and fully automated methods (EchoGo Core 2.0, Ultromics). Inter-technique agreement was assessed using linear regression and Bland-Altman analyses and two-way ANOVA. The diagnostic accuracy and time for detecting abnormalities (defined as LVEF ≤ 50% and GLS ≥ -15.1%, respectively) using AI was assessed by comparisons to manual measurements as a reference.

RESULTS

There were no significant differences in manual and automated LVEF and GLS values in either pre-CA (p = .791 and p = .105, respectively) or at diagnosis (p = .463 and p = .722). The two methods showed strong correlation on both the pre-CA (r = .78 and r = .83) and CA echoes (r = .74 and r = .80) for LVEF and GLS, respectively. The sensitivity and specificity of AI-derived indices for detecting abnormal LVEF were 83% and 86%, respectively, in the pre-CA echo and 70% and 79% at CA diagnosis. The sensitivity and specificity of AI-derived indices for detecting abnormal GLS was 82% and 86% in the pre-CA echo and 100% and 67% at the time of CA diagnosis. There was no significant difference in the relationship between LVEF (p = .99) and GLS (p = .19) and time to abnormality between the two methods.

CONCLUSION

Fully automated AI-calculated LVEF and GLS are comparable to manual measurements in patients pre-CA and at the time of CA diagnosis. The widespread implementation of automated LVEF and GLS may allow for more rapid assessment in different disease states with comparable accuracy and reproducibility to manual methods.

Non-invasive diagnosis of transthyretin cardiac amyloidosis utilizing typical late gadolinium enhancement pattern on cardiac magnetic resonance and light chains

AIMS

While cardiac magnetic resonance (CMR) is often obtained early in the evaluation of suspected cardiac amyloidosis (CA), it currently cannot be utilized to differentiate immunoglobulin (AL) and transthyretin (ATTR) CA. We aimed to determine whether a novel CMR and light-chain biomarker-based algorithm could accurately diagnose ATTR-CA.

METHODS AND RESULTS

Patients with confirmed AL or ATTR-CA with typical late gadolinium enhancement (LGE) and Look-Locker pattern for CA on CMR were retrospectively identified at three academic medical centres. Comprehensive light-chain analysis including free light chains, serum, and urine electrophoresis/immunofixation was performed. The diagnostic accuracy of the typical CMR pattern for CA in combination with negative light chains for the diagnosis of ATTR-CA was determined both in the entire cohort and in the subset of patients with invasive tissue biopsy as the gold standard. A total of 147 patients (age 70 ± 11, 76% male, 51% black) were identified: 89 ATTR-CA and 58 AL-CA. Light-chain biomarkers were abnormal in 81 (55%) patients. Within the entire cohort, the sensitivity and specificity of a typical LGE and Look-Locker CMR pattern and negative light chains for ATTR-CA was 73 and 98%, respectively. Within the subset with biopsy-confirmed subtype, the CMR and light-chain algorithm were 69% sensitive and 98% specific.

CONCLUSION

The combination of a typical LGE and Look-Locker pattern on CMR with negative light chains is highly specific for ATTR-CA. The successful non-invasive diagnosis of ATTR-CA using CMR has the potential to reduce diagnostic and therapeutic delays and healthcare costs for many patients.

Three-Dimensional Echocardiographic Deconstruction: Feasibility of Clinical Evaluation from Two-Dimensional Views Derived from a Three-Dimensional Data Set

BACKGROUND

Three-dimensional echocardiography (3DE) makes it possible to capture the entire heart in a single data set that theoretically could be used to extract any two-dimensional (2D) views and potentially replace the standard practice of serial 2D acquisitions. The aim of this study was to test the hypothesis that the quality of 3DE-derived 2D images is sufficient to allow the visualization of the left ventricular (LV), right ventricular (RV), and left atrial (LA) endocardium, on par with images from conventional two-dimensional echocardiography (2DE), and potentially more accurate quantification of chamber size and function.

METHODS

First, the investigators prospectively studied 36 patients who underwent 2DE in 14 standard views, and full-volume data sets from 3DE, from which the same views were extracted offline. The ability to visualize the LV endocardium, RV free wall, and LA endocardium was scored. LV linear dimensions, LV volumes, and LV ejection fraction (LVEF), LA volume, and RV basal dimension were measured and compared between both types of images. Thereafter, 40 patients who underwent 2DE, 3DE, and cardiac magnetic resonance (CMR) imaging on the same day were retrospectively studied. LV volumes and LVEF derived from 2DE and 3DE were compared side by side against the CMR reference.

RESULTS

Intertechnique agreement in visualization scores was 87% for LV segments, 86% for the RV free wall, and 83% for the LA endocardium. The correlations between 2DE- and 3DE-derived measurements were 0.95, 0.97, and 0.97 for LV volumes and LVEF, respectively, and 0.88 for RV basal dimension. Three-dimensional echocardiography-derived measurements of LV volumes and LVEF were more similar to those on CMR than those obtained on conventional 2DE.

CONCLUSIONS

The feasibility of segmental assessment of cardiac chambers using 3DE-derived 2D views is similar to that using conventional 2DE. This approach provides similar quantitative information, including more accurate LV volumes and LVEF measurements compared with CMR, and thus promises to significantly shorten the duration of the echocardiographic examination.

Imaging-Based Risk Stratification for Recurrence Risk in Women with a History of Peripartum Cardiomyopathy

OBJECTIVE

Peripartum cardiomyopathy (PPCM) affects 1:1,000 U.S. pregnancies, and while many recover from the disease, the risk of recurrence in subsequent pregnancy (SSP) is high. This study aims to evaluate the utility of left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) to predict the risk of recurrence of PPCM in SSP.

STUDY DESIGN

We retrospectively evaluated outcomes in women with a history of PPCM and SSP at a large-volume cardioobstetrics program (2008-2019).

RESULTS

There were 18 women who had incident PPCM and pursued SSP. Of 24 pregnancies in these women, 8 (33%) were complicated by the development of recurrent PPCM. LVEF ≥ 52% or GLS ≤ -16 was associated with a low risk of recurrent PPCM.

CONCLUSION

Approximately one-third of women with PPCM developed recurrent PPCM in SSP. LVEF and GLS on prepregnancy echocardiography may predict the risk of recurrence. Additional studies evaluating risk for recurrence are required to better understand which women are the safest to consider SSP.

KEY POINTS

· Peripartum cardiomyopathy affects 1:1000 US pregnancies.. · Approximately one third of women with a history of peripartum cardiomyopathy developed recurrent disease in a subsequent pregnancy.. · A left ventricular ejection fraction ≥52% or global longitudinal strain ≤-16 on echocardiogram is associated with a low risk of recurrence..

Atrial function and geometry differences in transthyretin versus immunoglobulin light chain amyloidosis: a cardiac magnetic resonance study

To determine the differences in left atrial (LA) function and geometry assessed by cardiac magnetic resonance (CMR) between transthyretin (ATTR) and immunoglobulin light chain (AL) cardiac amyloidosis (CA). We performed a retrospective analysis of 54 consecutive patients (68.5% male, mean age 67 ± 11 years) with confirmed CA (24 ATTR, 30 AL) who underwent comprehensive CMR examinations. LA structural and functional assessment including LA volume, LA sphericity index, and LA strain parameters were compared between both subtypes. In addition, 15 age-matched controls were compared to all groups. Patients with ATTR-CA were older (73 ± 9 vs. 62 ± 10 years, p < 0.001) and more likely to be male (83.3% vs. 56.7%, p = 0.036) when compared to AL-CA. No significant difference existed in LA maximum volume and LA sphericity index between ATTR-CA and AL-CA. LA minimum volumes were larger in ATTR-CA when compared with AL-CA. There was a significant difference in LA function with worse strain values in ATTR vs AL: left atrial reservoir [7.4 (6.3–12.8) in ATTR vs. 13.8 (6.90–24.8) in AL, p = 0.017] and booster strains [3.6 (2.6–5.5) in ATTR vs. 5.2 (3.6–12.1) in AL, p = 0.039]. After adjusting for age, LA reservoir remained significantly lower in ATTR-CA compared to AL-CA (p = 0.03), but not LA booster (p = 0.16). We demonstrate novel differences in LA function between ATTR-CA and AL-CA despite similar LA geometry. Our findings of more impaired LA function in ATTR may offer insight into higher AF burden in these patients.

Recent Advances and Current Dilemmas in the Diagnosis and Management of Transthyretin Cardiac Amyloidosis

Cardiac amyloidosis (CA) is an increasingly recognized cause of heart failure, arrhythmias, and sudden cardiac death. While CA was previously rapidly fatal, recent advances in diagnosis and treatment have significantly improved outcomes. Advances in cardiac imaging and biomarkers have critically improved the accuracy and efficiency with which CA is diagnosed, even allowing for the noninvasive diagnosis of transthyretin CA. Cardiac magnetic resonance imaging, technetium nuclear imaging, echocardiography, and blood-based biomarkers have established important and complementary roles in the management and advancement of care. At the same time, the development of novel targeted amyloid therapies has allowed patients with CA to live longer and potentially achieve better quality of life. Still, despite this significant progress, there remain critical ongoing questions in the field. Accordingly, within this review we will highlight recent advances in cardiac imaging and therapeutics for CA, while focusing on key opportunities for further optimization of care and outcomes among this growing population. Specifically, we will discuss ongoing debates in the diagnosis of CA, including the interpretation of indeterminate cardiac imaging findings, the best technique to screen asymptomatic transthyretin amyloidosis gene mutation carriers for cardiac involvement, and the ideal method for monitoring response to CA treatment. We will additionally focus on recent advances in treatment for transthyretin amyloidosis-CA, including a discussion of available agents as well as highlighting ongoing clinical trials. Together, these data will allow clinicians to emerge with a greater understanding of the present and future of diagnosis, management, and potentially enhanced outcomes in this rapidly advancing field.

Indexed left ventricular mass to QRS voltage ratio is associated with heart failure hospitalizations in patients with cardiac amyloidosis

In cardiac amyloidosis (CA), amyloid infiltration results in increased left ventricular (LV) mass disproportionate to electrocardiographic (EKG) voltage. We assessed the relationship between LV mass-voltage ratio with subsequent heart failure hospitalization (HHF) and mortality in CA. Patients with confirmed CA and comprehensive cardiovascular magnetic resonance (CMR) and EKG exams were included. CMR-derived LV mass was indexed to body surface area. EKG voltage was assessed using Sokolow, Cornell, and Limb-voltage criteria. The optimal LV mass-voltage ratio for predicting outcomes was determined using receiver operating characteristic curve analysis. The relationship between LV mass-voltage ratio and HHF was assessed using Cox proportional hazards analysis adjusting for significant covariates. A total of 85 patients (mean 69 ± 11 years, 22% female) were included, 42 with transthyretin and 43 with light chain CA. At a median of 3.4-year follow-up, 49% of patients experienced HHF and 60% had died. In unadjusted analysis, Cornell LV mass-voltage ratio was significantly associated with HHF (HR, 1.05; 95% CI 1.02-1.09, p = 0.001) and mortality (HR, 1.05; 95% CI 1.02-1.07, p = 0.001). Using ROC curve analysis, the optimal cutoff value for Cornell LV mass-voltage ratio to predict HHF was 6.7 gm/m2/mV. After adjusting for age, NYHA class, BNP, ECV, and LVEF, a Cornell LV mass-voltage ratio > 6.7 gm/m2/mV was significantly associated with HHF (HR 2.25, 95% CI 1.09-4.61; p = 0.03) but not mortality. Indexed LV mass-voltage ratio is associated with subsequent HHF and may be a useful prognostic marker in cardiac amyloidosis.

Current Trends in Anticoagulation Bridging for Patients With Chronic Atrial Fibrillation on Warfarin Undergoing Endoscopy

For warfarin-treated patients with atrial fibrillation (AF) at low thromboembolic risk, recent studies have shown harm associated with periprocedural bridging using low-molecular-weight heparin. Clinician surveys have indicated a preference toward excessive bridging, especially among noncardiologists; however, little is known about actual practice patterns in these patients. We performed a retrospective evaluation of bridging in the setting of gastrointestinal endoscopy. We identified 938 patients with AF on warfarin who underwent esophagogastroduodenoscopy or colonoscopy between 2012 and 2016 at a tertiary health center. Urgent, inpatient, or advanced endoscopic procedures were excluded. Clinical variables were abstracted using a predefined data dictionary. Values were expressed as means and compared using a t test or a chi-squared test as appropriate. Three hundred seventy-four patients met criteria for analysis. Twenty-five percent of these patients received bridging therapy, including 11% of patients with CHADS₂ scores of 0 to 2 without valvular AF or previous venous thromboembolism. Of the clinical variables assessed, CHADS₂, CHA₂DS₂-VASc, and a history of stroke were the strongest predictors of bridging. Cardiologists were also significantly less likely to prescribe bridging than noncardiology providers (18% vs 30%, p = 0.011); this effect was significant when controlling for CHADS₂, CHA₂DS₂-VASc, or stroke history. In conclusion, patients with AF on warfarin receive excessive low-molecular-weight heparin bridging in the setting of endoscopy; the lower rates of bridging observed among cardiologists suggests a need for their increased involvement in this decision making.