Real-Time Artificial Intelligence-Based Guidance of Echocardiographic Imaging by Novices: Image Quality and Suitability for Diagnostic Interpretation and Quantitative Analysis
BACKGROUND/UNASSIGNED:We aimed to assess in a prospective multicenter study the quality of echocardiographic exams performed by inexperienced users guided by a new artificial intelligence software and evaluate their suitability for diagnostic interpretation of basic cardiac pathology and quantitative analysis of cardiac chamber and function. METHODS/UNASSIGNED:The software (UltraSight, Ltd) was embedded into a handheld imaging device (Lumify; Philips). Six nurses and 3 medical residents, who underwent minimal training, scanned 240 patients (61±16 years; 63% with cardiac pathology) in 10 standard views. All patients were also scanned by expert sonographers using the same device without artificial intelligence guidance. Studies were reviewed by 5 certified echocardiographers blinded to the imager's identity, who evaluated the ability to assess left and right ventricular size and function, pericardial effusion, valve morphology, and left atrial and inferior vena cava sizes. Finally, apical 4-chamber images of adequate quality, acquired by novices and sonographers in 100 patients, were analyzed to measure left ventricular volumes, ejection fraction, and global longitudinal strain by an expert reader using conventional methodology. Measurements were compared between novices' and experts' images. RESULTS/UNASSIGNED:Of the 240 studies acquired by novices, 99.2%, 99.6%, 92.9%, and 100% had sufficient quality to assess left ventricular size and function, right ventricular size, and pericardial effusion, respectively. Valve morphology, right ventricular function, and left atrial and inferior vena cava size were visualized in 67% to 98% exams. Images obtained by novices and sonographers yielded concordant diagnostic interpretation in 83% to 96% studies. Quantitative analysis was feasible in 83% images acquired by novices and resulted in high correlations (r≥0.74) and small biases, compared with those obtained by sonographers. CONCLUSIONS/UNASSIGNED:After minimal training with the real-time guidance software, novice users can acquire images of diagnostic quality approaching that of expert sonographers in most patients. This technology may increase adoption and improve accuracy of point-of-care cardiac ultrasound.
An unusual case of a solitary cardiac myofibroma causing severe right ventricular outflow tract obstruction in an infant
Cardiac tumours are relatively uncommon, particularly in children. Myofibroma is an extremely rare variety of cardiac tumour, which nearly always arises in the context of infantile myofibromatosis. Herein, we present a case of a solitary cardiac myofibroma causing right ventricular outflow tract obstruction in a 2-month-old male infant.
Left Ventricular Mass Quantification by Two-Dimensional Echocardiography in a Pediatric Population: Correlation with Cardiac Magnetic Resonance Imaging
Quantification of left ventricular (LV) mass by echocardiography has not been validated against the gold standard of cardiac magnetic resonance imaging (CMR) in the pediatric population. The purpose of this study was to compare LV mass by two-dimensional and conventional M-mode echocardiography versus CMR in children. Consecutive CMR studies were paired with echocardiograms and retrospectively analyzed in children ageâ€‰â‰¤â€‰16Â years (3Â days old to 16Â years old). Studies performedâ€‰>â€‰3Â months between modalities and single ventricle anatomy were excluded. Unindexed LV mass was calculated using M-mode, area-length (AL), and truncated ellipsoid (TE) methods via echocardiography, and compared to cine stack CMR images. There were 46 patients included in the study (both MRI and echocardiography). Good correlations were observed for LV mass measured by CMR and all echocardiographic methods: M-mode (Râ€‰=â€‰0.965), AL (Râ€‰=â€‰0.975), and TE (Râ€‰=â€‰0.975). There was a significant overestimation using TE echocardiography, by a mean of 10.5Â g (95% confidence interval 5.7-15.2Â g, pâ€‰<â€‰0.05). There was no significant over- or underestimation of LV mass observed by M-mode or AL echocardiographic measurements, with tight limits of agreement when compared to CMR (95% confidence interval -â€‰5.2 to 4.4Â g and -â€‰1.5 to 6.7Â g, respectively). Interobserver agreement was good for each of the echocardiographic measurements, but inferior with M-mode (ICC, 0.89) compared to two-dimensional methods (ICC, 0.97). Echocardiographic estimates of LV mass have good correlation with CMR in children. Performance comparison showed AL echocardiographic method provides the most accurate measurement of LV mass with the best reproducibility compared to other methods.
Does artificial intelligence and deep learning architecture system accurately measure left ventricular systolic function? [Meeting Abstract]
Background: Left ventricular ejection fraction (LVEF) is the most used measurement in the assessment of cardiac function. However, human error and image quality are among several factors contributing to inaccuracies of this measurement. The purpose of this study was to establish the correlation between LVEF measured by expert ultrasound cardiac sonographers and by a system based on a deep learning architecture and geometry processing algorithms. Methods: We developed a system based on a novel deep learning architecture and geometry processing algorithms for building a computational pipeline. Input was from a stream of cardiac ultrasound images that automatically calculated the ejection fraction of the left ventricle (LV). The system analyzes each image, identifies the viewing direction, segments the left ventricle, identifies systolic/diastolic frames and reconstructs 3D models of the LV at different times (see Figure 1). It then computes robust estimation of the ejection fraction. Echocardiographic images from of 114 patients were used. Each case contains cardiac ultrasound clips from two and four-chamber views fed to the system as a single stream with no additional data. We created the ground truth ejection fraction (EF) value for each case by taking the median of 8 estimators: 4 expert cardiologists eye-ball estimations, 3 computations using Simpson formula and 3D reconstructions based on marked LV segmentations of an expert technician, and the original medical report. Results: Figure 2 depicts error histograms comparing EF values and the ground-truth EF values. We compared the error histogram of our algorithm (right) and that of one of the experts (left ). The mean and variance of histograms is comparable indicating our algorithm is comparable in accuracy to an expert cardiologist. Conclusions: An excellent correlation was found between LVEF measured by expert Ultrasound cardiac sonographer and the LVEF measured by a system based on a deep learning architecture and geometry processing algorithms (Figure presented)
Right ventricular systolic function in patients with hypoplastic left heart syndrome following norwood operation: A novel method using doppler dP/dT across the sano shunt [Meeting Abstract]
Background: Hypoplastic left heart syndrome (HLHS) is of the highest risk lesions in congenital heart disease. The initial surgery typically involves the Norwood procedure, frequently with Sano modification. Despite surgical advances, interstage mortality remains approximately 10%, with diminished right ventricular (RV) systolic function a risk factor for death. Tricuspid regurgitation (TR) derived dP/dT has been shown to correlate with catheter derived dP/dT in patients with HLHS, and is frequently used in assessment of RV systolic function. However, in patients without an adequate TR Doppler signal, this modality cannot be used. The purpose of this study was to evaluate the correlation between a Sano derived RV dP/dT and TR derived RV dP/dT in the assessment of RV systolic function following Norwood operation with Sano modification. Methods: An echocardiographic retrospective review was performed in patients following a Norwood/Sano operation, and assessed for the presence of a continuous wave Doppler tracing across the tricuspid valve and Sano shunt. 46 studies met this criteria. A dP/dT for both methods was measured between 1 and 3 m/s, and a qualitative assessment of function was also assigned to each study. Results: The mean value of TR derived dP/ dT and Sano derived dP/dT was 979.9 mmHg/s and 1102.89 mmHg/s respectively. A linear relationship exists between a Sano and TR derived RV dP/dT, where the TR derived RV dP/dT = 387+0.73x, (p=0.001). There was also a statistically significant relationship between the Sano derived dP/dT and the qualitative analysis of RV systolic function, particularly in the higher values. Moderately diminished function was found to have a Sano derived dP/dT of 860 to 1093, mildly diminished 732 to 1325, and normal function 866 to 2048. Conclusions: Sano derived RV dP/dT can be effectively used as a surrogate for TR derived RV dP/dT, which may be of use in patients with inadequate TR. There is a statistically significant correlation between the Sano dP/dT and the qualitative assessment of function (Figure presented)
A prospective case series evaluating use of an in-line air detection and purging system to reduce air burden during major surgery
Background/UNASSIGNED:Intravascular air embolism (AE) is a preventable but potentially catastrophic complication caused by intravenous tubing, trauma, and diagnostic and surgical procedures. The potentially fatal risks of arterial AE are well-known, and emerging evidence demonstrates impact of venous AEs on inflammatory response and coagulation factors. A novel FDA-approved in-line air detection and purging system was used to detect and remove air caused by administering a rapid fluid bolus during surgery. Methods/UNASSIGNED:A prospective, randomized, case series was conducted. Subjects were observed using standard monitors, including transesophageal echocardiography (TEE) in the operating room. After general anesthesia was induced, an introducer and pulmonary artery catheter was inserted in the right internal jugular to administer fluids and monitor cardiac pressures. Six patients undergoing cardiac surgery were studied. Each patient received four randomized fluid boluses: two with the in-line air purging device, two without. For each bolus, a bulb infuser was squeezed three times (10-15 mL) over 5 s. The TEE was positioned in the mid-esophageal right atrium (RA) to quantify peak air clearance, and images were video recorded throughout each bolus. Air was quantified using optical densitometry (OD) from images demonstrating maximal air in the RA. Results/UNASSIGNED:â€‰<â€‰0.001). Conclusion/UNASSIGNED:An air purging system reduced air burden from bolus administration and could consequently reduce the risk of harmful or fatal AEs during surgery.
Intracardiac echogenic foci revisited [Meeting Abstract]
Background/Hypothesis: The clinical importance of intracardiac echogenic foci(ICEF)on prenatal ultrasound remains debatable amongst pediatric cardiologists and obstetricians. It has previously been discredited as a marker of triploidy and no clear association with congenital heart disease (CHD) has been established. Yet it continues to be a frequent reason for referral for fetal echocardiogram. The aim of this study was to assess the presence of CHD in patients with ICEF and to determine if it represents echocardiographic artifact. We hypothesized that there is no significant association between ICEF and CHD and that ICEF represents echocardiographic artifact. Materials and Methods: The institutional fetal echocardiography database was retrospectively searched for reports containing the diagnosis of echogenic focus from November 2014-November 2016. Each study was evaluated by a single reviewer. The presence of ICEF in the 3 standard views analogous to the apical 4 chamber, parasternal short axis, and parasternal long axis views was recorded. In addition, the presence of CHD, pericardial effusion, and/or fetal arrhythmia for each study was documented. Statistical analysis was performed using cross-tabulation with chi-square analysis. Results: 55 patients, yielding 145 studies were reviewed. 6 patients (10.9%) had CHD, which included VSD(n= 3), PA/VSD(n =2), and vascular ring(n =1). ICEF was noted in all 3 views in 33 studies(23%), in 2 views in 45 studies(31%), in 1 view in 22 studies (15%), and resolved in 45 studies (31%)(see table). Of those studies with ICEF noted in all 3 views, 19%(n=6) were associated with CHD, indicating a significant association in this highly selective population (p =0.039). Conclusions: There is a higher incidence of CHD in fetuses with ICEF compared to the general population. The presence of ICEF was more likely to be identified in multiple views, suggesting it is unlikely echocardiographic artifact
Utility and Scope of Rapid Prototyping in Patients with Complex Muscular Ventricular Septal Defects or Double-Outlet Right Ventricle: Does it Alter Management Decisions?
Rapid prototyping facilitates comprehension of complex cardiac anatomy. However, determining when this additional information proves instrumental in patient management remains a challenge. We describe our experience with patient-specific anatomic models created using rapid prototyping from various imaging modalities, suggesting their utility in surgical and interventional planning in congenital heart disease (CHD). Virtual and physical 3-dimensional (3D) models were generated from CT or MRI data, using commercially available software for patients with complex muscular ventricular septal defects (CMVSD) and double-outlet right ventricle (DORV). Six patients with complex anatomy and uncertainty of the optimal management strategy were included in this study. The models were subsequently used to guide management decisions, and the outcomes reviewed. 3D models clearly demonstrated the complex intra-cardiac anatomy in all six patients and were utilized to guide management decisions. In the three patients with CMVSD, one underwent successful endovascular device closure following a prior failed attempt at transcatheter closure, and the other two underwent successful primary surgical closure with the aid of 3D models. In all three cases of DORV, the models provided better anatomic delineation and additional information that altered or confirmed the surgical plan. Patient-specific 3D heart models show promise in accurately defining intra-cardiac anatomy in CHD, specifically CMVSD and DORV. We believe these models improve understanding of the complex anatomical spatial relationships in these defects and provide additional insight for pre/intra-interventional management and surgical planning.
A Case of Prenatally Diagnosed Aortico-Left Ventricular Tunnel [Meeting Abstract]
Prenatal Diagnosis of a Persistent Fifth Aortic Arch, Pulmonary-to-Systemic type: An Unusual Association with Evolving Aortic Coarctation
Persistent fifth arch (PFA) is a rare anomaly that is often underdiagnosed and missed. Different PFA types that have been reported in the literature are systemic-to-systemic type (most common), systemic-to-pulmonary artery (PA), and PA-to-systemic types. The designations of systemic-to-PA or PA-to-systemic are based on if the PFA is a source of PA or systemic blood flow, respectively, in the setting of critical proximal obstruction (pulmonary atresia or aortic atresia). This case describes an unusual PFA, which connects the distal PA to distal ascending aorta; however, it is not associated with critical proximal obstruction, and while it appeared to be an incidental finding in early gestation, progressive serial distal obstruction of the left fourth arch was seen to develop. This case highlights that prenatal diagnosis of PFA is possible and that once a diagnosis is made, serial fetal echocardiograms should be performed to evaluate for evolving lesions.