A fast 3D adaptive bilateral filter for ultrasound volume visualization

Area of Science:

• Medical imaging
• Image processing
• Ultrasound technology

Background:

• Medical ultrasound volume data often contains significant noise, necessitating filtering.
• Conventional 2D filtering methods fail to leverage inter-slice information.
• Existing 3D filtering methods are computationally intensive and may not suit ultrasound data characteristics.

Purpose of the Study:

• To introduce an effective and fast noise removal method for medical ultrasound volume data.
• To address limitations of existing 2D and 3D filtering techniques.
• To develop an adaptive filtering approach that considers ultrasound sampling characteristics.

Main Methods:

• Developed a parallel bilateral filtering technique utilizing a 3D summed area table.
• Implemented an adaptive filter with a kernel window size adjusted based on distance from the signal transmission point.
• Compared the proposed method against anisotropic diffusion and standard bilateral filtering.

Main Results:

• The adaptive filter effectively accounts for the specific sampling characteristics of ultrasound volumes.
• Demonstrated superior noise removal compared to anisotropic diffusion and bilateral filtering.
• Showcased minimal loss of original data in processed ultrasound images.

Conclusions:

• The proposed parallel, adaptive bilateral filtering method offers enhanced noise reduction for ultrasound data.
• This technique minimizes data distortion more effectively than simple or non-adaptive filters.
• The method provides a faster and more efficient solution for ultrasound image enhancement.