Artifacts detection-based adaptive filtering to noise reduction of strain imaging

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Strain imaging in medical ultrasound visualizes tissue elasticity but is prone to artifacts.
• Artifacts degrade lesion detectability and can lead to misdiagnosis.
• Effective artifact suppression while preserving tissue structure is crucial for diagnosis.

Purpose of the Study:

• To address the challenge of threshold selection in bilateral filtering for ultrasound strain imaging.
• To develop an adaptive fast bilateral filter for artifact reduction.
• To improve the quality and diagnostic utility of ultrasound strain images.

Main Methods:

• Derived the probability distribution function of amplitude modulation noise from uncompressed speckle statistics.
• Developed a statistical model for artifact formation.
• Designed and implemented an adaptive fast bilateral filter based on the statistical model.

Main Results:

• The proposed method significantly improved the quality of ultrasonic strain imaging in both simulations and phantom tests.
• Elastographic signal-to-noise ratio increased by 129.91% (simulated) and 52.36% (phantom).
• Elastographic contrast-to-noise ratio increased by 521.42% (simulated) and 218.07% (phantom), enhancing visualization.

Conclusions:

• The adaptive fast bilateral filter effectively reduces artifacts in ultrasound strain imaging.
• The method enhances diagnostic accuracy by improving lesion detectability and reducing misdiagnosis likelihood.
• The proposed technique offers superior visualization compared to existing methods.