Pediatric Echocardiogram Classification

This study developed a convolutional neural network to automatically classify 27 standard pediatric echocardiographic views using over 12,000 training images from patients aged 0–19 years. The model achieved 90.3% overall accuracy across age groups and performed well across different view types, with especially high accuracy for Doppler tracings and color Doppler sweeps. These results demonstrate reliable pediatric-specific view classification and establish a foundation for fully automated interpretation and quantitative analysis of pediatric echocardiograms.