GPU-accelerated 3D mipmap for real-time visualization of ultrasound volume data

Area of Science:

• Medical Imaging
• Computer Graphics
• Signal Processing

Background:

• Ultrasound volume rendering is crucial for fetal visualization in obstetrics and gynecology.
• High-quality rendering requires effective noise removal and coordinate conversion.
• Existing 2D and 3D noise filtering methods have limitations in information utilization and processing time.

Purpose of the Study:

• To develop an efficient 3D noise reduction method for ultrasound volume rendering.
• To improve the quality of fetal visualizations in obstetrics and gynecology.
• To enable real-time 3D ultrasound data processing.

Main Methods:

• A 3D-mipmap-based noise reduction technique utilizing graphics hardware is proposed.
• Noise detection is achieved by comparing density values across consecutive mipmap levels.
• An adaptive noise detector selects optimal mipmap levels based on density differences.

Main Results:

• The proposed method significantly reduces noise and artifacts in ultrasound data.
• It processes 3D ultrasound data in real-time, outperforming traditional methods.
• The technique efficiently utilizes storage and generation time compared to standard 3D mipmaps.

Conclusions:

• The 3D-mipmap noise reduction method offers an efficient solution for high-quality ultrasound volume rendering.
• This approach enhances fetal imaging in obstetrics and gynecology by enabling real-time noise filtering.
• The method provides a practical advancement for medical imaging visualization.