Real-Time 3D Ultrasound Reconstruction Using U2-Net Segmentation for Cardiovascular Interventional Surgery.

Area of Science:

• Medical Imaging
• Interventional Cardiology
• Biomedical Engineering

Background:

• Current cardiovascular interventions rely on 2D X-ray imaging, leading to radiation exposure.
• This method lacks intuitive 3D visualization crucial for complex surgical navigation.

Purpose of the Study:

• To develop a real-time 3D ultrasound reconstruction technique for cardiovascular interventional surgery.
• To provide radiation-free, intuitive 3D surgical guidance, reducing reliance on X-rays.

Main Methods:

• Utilized U2-Net segmentation for blood vessel identification.
• Implemented real-time 3D ultrasound reconstruction at 60 frames per second using volume rendering.
• Integrated electromagnetic tracking and fused ultrasound data with preoperative CT scans.

Main Results:

• Achieved real-time 3D reconstruction of segmented blood vessels.
• Demonstrated accurate ultrasound-to-navigation coordinate registration with an RMS error of 1.375±0.369 mm.
• Provided physicians with intuitive, radiation-free 3D surgical visualization.

Conclusions:

• The proposed 3D ultrasound technique offers a viable, radiation-free alternative to X-ray guidance in cardiovascular interventions.
• This innovation enhances surgical precision and patient safety by providing clear 3D anatomical information.