A Topology-Enhanced Dual-stream Attention Model for Parasternal Long-Axis View Identification in Echocardiography

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Cardiology

Background:

• The echocardiographic parasternal long-axis (PLAX) view is crucial for cardiac evaluation.
• Accurate PLAX view identification is challenging due to complex cardiac anatomy and speckle noise.
• Existing methods struggle with subtle inter-view variations and noise interference.

Purpose of the Study:

• To develop an advanced model for accurate echocardiographic PLAX view identification.
• To enhance diagnostic reliability in cardiac ultrasound imaging.
• To leverage anatomical topology for improved view recognition.

Main Methods:

• Proposed a Topology-Enhanced Dual-stream Attention (TEDA) model.
• Constructed an anatomy-guided graph representing cardiac regions and their dependencies.
• Integrated spatial, structural, semantic, and topological features using a dual-stream attention mechanism.

Main Results:

• The TEDA model achieved a high accuracy of 90.30% on an in-house echocardiographic dataset.
• Demonstrated superior performance compared to state-of-the-art methods.
• Effectively captured discriminative representations of PLAX view characteristics.

Conclusions:

• The TEDA model offers a robust solution for accurate PLAX view identification.
• This approach has significant potential for improving echocardiographic navigation and view localization.
• Exploiting topological relationships enhances recognition accuracy and diagnostic confidence.