Optimización de la interpretación de ecocardiogramas con la utilización de Machine Learning en el estudio WASE / Optimizing Echocardiograms Interpretation with the Application of Machine Learning in the WASE Study

RESUMEN Introducción: El número creciente de estudios ecocardiográficos y la necesidad de cumplir rigurosamente con las recomendaciones de guías internacionales de cuantificación, ha llevado a que los cardiólogos deban realizar tareas sumamente extensas y repetitivas, como parte de la interpretación y análisis de cantidades de información cada vez más abrumadoras. Novedosas técnicas de machine learning (ML), diseñadas para reconocer imágenes y realizar mediciones en las vistas adecuadas, están siendo cada vez más utilizadas para responder a esta necesidad evidente de automatización de procesos. Objetivos: Nuestro objetivo fue evaluar un modelo alternativo de interpretación y análisis de estudios ecocardiográficos, basado fundamentalmente en la utilización de software de ML, capaz de identificar y clasificar vistas y realizar mediciones estandarizadas de forma automática. Material y métodos: Se utilizaron imágenes obtenidas en 2000 sujetos normales, libres de enfermedad, de los cuales 1800 fueron utilizados para desarrollar los algoritmos de ML y 200 para su validación posterior. Primero, una red neuronal convolucional fue desarrollada para reconocer 18 vistas ecocardiográficas estándar y clasificarlas de acuerdo con 8 grupos (stacks) temáticos. Los resultados de la identificación automática fueron comparados con la clasificación realizada por expertos. Luego, algoritmos de ML fueron desarrollados para medir automáticamente 16 parámetros de eco Doppler de evaluación clínica habitual, los cuales fueron comparados con las mediciones realizadas por un lector experto. Finalmente, comparamos el tiempo necesario para completar el análisis de un estudio ecocardiográfico con la utilización de métodos manuales convencionales, con el tiempo necesario con el empleo del modelo que incorpora ML en la clasificación de imágenes y mediciones ecocardiográficas iniciales. La variabilidad inter e intraobservador también fue analizada. Resultados: La clasificación automática de vistas fue posible en menos de 1 segundo por estudio, con una precisión de 90 % en imágenes 2D y de 94 % en imágenes Doppler. La agrupación de imágenes en stacks tuvo una precisión de 91 %, y fue posible completar dichos grupos con las imágenes necesarias en 99% de los casos. La concordancia con expertos fue excelente, con diferencias similares a las observadas entre dos lectores humanos. La incorporación de ML en la clasificación y medición de imágenes ecocardiográficas redujo un 41 % el tiempo de análisis y demostró menor variabilidad que la metodología de interpretación convencional. Conclusión: La incorporación de técnicas de ML puede mejorar significativamente la reproducibilidad y eficiencia de las interpretaciones y mediciones ecocardiográficas. La implementación de este tipo de tecnologías en la práctica clínica podría resultar en reducción de costos y aumento en la satisfacción del personal médico.

ABSTRACT Background: The growing number of echocardiographic tests and the need for strict adherence to international quantification guidelines have forced cardiologists to perform highly extended and repetitive tasks when interpreting and analyzing increasingly overwhelming amounts of data. Novel machine learning (ML) techniques, designed to identify images and perform measurements at relevant visits, are becoming more common to meet this obvious need for process automation. Objectives: Our objective was to evaluate an alternative model for the interpretation and analysis of echocardiographic tests mostly based on the use of ML software in order to identify and classify views and perform standardized measurements automatically. Methods: Images came from 2000 healthy subjects, 1800 of whom were used to develop ML algorithms and 200 for subsequent validation. First, a convolutional neural network was developed in order to identify 18 standard echocardiographic views and classify them based on 8 thematic groups (stacks). The results of automatic identification were compared to classification by experts. Later, ML algorithms were developed to automatically measure 16 Doppler scan parameters for regular clinical evaluation, which were compared to measurements by an expert reader. Finally, we compared the time required to complete the analysis of an echocardiographic test using conventional manual methods with the time needed when using the ML model to classify images and perform initial echocardiographic measurements. Inter- and intra-observer variability was also analyzed. Results: Automatic view classification was possible in less than 1 second per test, with a 90% accuracy for 2D images and a 94% accuracy for Doppler scan images. Stacking images had a 91% accuracy, and it was possible to complete the groups with any necessary images in 99% of cases. Expert agreement was outstanding, with discrepancies similar to those found between two human readers. Applying ML to echocardiographic imaging classification and measurement reduced time of analysis by 41% and showed lower variability than conventional reading methods. Conclusion: Application of ML techniques may significantly improve reproducibility and efficiency of echocardiographic interpretations and measurements. Using this type of technologies in clinical practice may lead to reduced costs and increased medical staff satisfaction.

Deep learning assisted measurement of echocardiographic left heart parameters: improvement in interobserver variability and workflow efficiency.

Machine learning techniques designed to recognize views and perform measurements are increasingly used to address the need for automation of the interpretation of echocardiographic images. The current study was designed to determine whether a recently developed and validated deep learning (DL) algorithm for automated measurements of echocardiographic parameters of left heart chamber size and function can improve the reproducibility and shorten the analysis time, compared to the conventional methodology. The DL algorithm trained to identify standard views and provide automated measurements of 20 standard parameters, was applied to images obtained in 12 randomly selected echocardiographic studies. The resultant measurements were reviewed and revised as necessary by 10 independent expert readers. The same readers also performed conventional manual measurements, which were averaged and used as the reference standard for the DL-assisted approach with and without the manual revisions. Inter-reader variability was quantified using coefficients of variation, which together with analysis times, were compared between the conventional reads and the DL-assisted approach. The fully automated DL measurements showed good agreement with the reference technique: Bland-Altman biases 0-14% of the measured values. Manual revisions resulted in only minor improvement in accuracy: biases 0-11%. This DL-assisted approach resulted in a 43% decrease in analysis time and less inter-reader variability than the conventional methodology: 2-3 times smaller coefficients of variation. In conclusion, DL-assisted approach to analysis of echocardiographic images can provide accurate left heart measurements with the added benefits of improved reproducibility and time savings, compared to conventional methodology.

Left Ventricular Diastolic Function in Healthy Adult Individuals: Results of the World Alliance Societies of Echocardiography Normal Values Study.

BACKGROUND: The World Alliance Societies of Echocardiography (WASE) study was conducted to describe echocardiographic normal values in adults and to compare races and nationalities using a uniform acquisition and measurement protocol. This report focuses on left ventricular (LV) diastolic function. METHODS: WASE is an international, cross-sectional study. Participants were enrolled with equal distribution according to age and gender. Echocardiograms were analyzed in a core laboratory based on the latest American Society of Echocardiography/European Association of Cardiovascular Imaging guidelines. Left ventricular diastolic function was assessed by E, E/A, e' velocities, E/e', left atrial volume index (LAVI), and tricuspid regurgitation velocity. Determination of LV diastolic function was made using the algorithm proposed by the guidelines. RESULTS: A total of 2,008 subjects from 15 countries were enrolled. The majority were of white or Asian race (42.8%, 41.8%, respectively). E and E/e' were higher in female patients, while LAVI was similar in both genders. Consistent increase in E/e' and decrease in E/A, E, and e' were found as age increased. The upper limit of normal for LAVI was higher in WASE compared with the guidelines. The lower limits of normal for e' were smaller in elder groups than those in the guidelines, while the upper limits of normal for E/e' were below the guideline values. These findings suggest that the cutoff value for LAVI should be shifted upward and age-specific cutoff values for e' should be considered. In WASE, <93.6% of patients were classified as normal LV diastolic function using the guidelines' algorithm, and the proportion increased to 97.4% when applying the revised cutoff values for LAVI obtained in our study. CONCLUSIONS: Guideline-recommended normal values for e' velocities and LAVI should be reconsidered. The algorithm for the determination of LV diastolic function proposed by the guidelines is useful, but adjustments to LAVI could further improve it.

Similarities and Differences in Left Ventricular Size and Function among Races and Nationalities: Results of the World Alliance Societies of Echocardiography Normal Values Study.

BACKGROUND: The World Alliance Societies of Echocardiography (WASE) Normal Values Study evaluates individuals from multiple countries and races with the aim of describing normative values that could be applied to the global community worldwide and to determine differences and similarities among people from different countries and races. The present report focuses specifically on two-dimensional (2D) left ventricular (LV) dimensions, volumes, and systolic function. METHODS: The WASE Normal Values Study is a multicenter international, observational, prospective, cross-sectional study of healthy adult individuals. Participants recruited in each country were evenly distributed among six predetermined subgroups according to age and gender. Comprehensive 2D transthoracic echocardiograms were acquired and analyzed following strict protocols based on recent American Society of Echocardiography and European Association of Cardiovascular Imaging guidelines. Analysis was performed at the WASE 2D core laboratory and included 2D LV dimensions, LV volumes, and LV ejection fraction (LVEF) by the biplane Simpson method and global longitudinal strain (GLS). RESULTS: Two thousand eight subjects were enrolled in 15 countries. The median age was 45 years (interquartile range, 32-65 years), 42.8% were white, 41.8% were Asian, and 9.7% were black. LV dimensions and volumes were larger in male subjects, while LVEF and GLS were higher in female subjects. Global WASE normal ranges for LV dimensions were smaller than those in the guidelines, but the upper limits of normal for LV volumes and the lower limits of normal for LVEF were higher in the WASE study. Significant intercountry variation was identified for all LV parameters reflecting LV size (dimensions, mass, and volumes) even after indexing to body surface area, with LV end-diastolic and end-systolic volumes having the highest variation. The largest volumes were noted in Australia, while the smallest were measured in India for both genders. This finding suggests that in addition to gender and body surface area, specific country should be considered when evaluating LV volumes. Intercountry variation for LVEF and GLS was smaller but still statistically significant (P < .05 for all). CONCLUSIONS: LV dimensions and volumes are larger in men, while LVEF and GLS are higher in women. Current guideline-recommended normal ranges for LV volumes and LVEF should be adjusted. Intercountry variability is significant for LV volumes, and therefore nationality should be considered for defining ranges of normality.