Temporal Feature Weaving for Neonatal Echocardiographic Viewpoint Video Classification

TL;DR

This paper introduces a CNN-GRU based method with temporal feature weaving for classifying neonatal echocardiogram viewpoints, improving accuracy with minimal computational cost, and provides a new annotated dataset for future research.

Contribution

It presents a novel temporal feature weaving technique within a CNN-GRU architecture for echocardiogram viewpoint classification and releases a new dataset to support further studies.

Findings

4.33% accuracy improvement over baseline

Effective use of only four frames

Minimal additional computational overhead

Abstract

Automated viewpoint classification in echocardiograms can help under-resourced clinics and hospitals in providing faster diagnosis and screening when expert technicians may not be available. We propose a novel approach towards echocardiographic viewpoint classification. We show that treating viewpoint classification as video classification rather than image classification yields advantage. We propose a CNN-GRU architecture with a novel temporal feature weaving method, which leverages both spatial and temporal information to yield a 4.33\% increase in accuracy over baseline image classification while using only four consecutive frames. The proposed approach incurs minimal computational overhead. Additionally, we publish the Neonatal Echocardiogram Dataset (NED), a professionally-annotated dataset providing sixteen viewpoints and associated echocardipgraphy videos to encourage future work and development in this field. Code available at: https://github.com/satchelfrench/NED