Wavelet-based edge detection in ultrasound images

Area of Science:

• Medical Imaging
• Signal Processing
• Biomedical Engineering

Background:

• Ultrasound (US) imaging is crucial for non-invasive diagnostics.
• Accurate edge detection in US images is vital for quantitative analysis and feature identification.
• Existing edge detection methods may struggle with US image noise and artifact characteristics.

Purpose of the Study:

• To present a novel wavelet-based method for robust edge detection in ultrasound images.
• To demonstrate the effectiveness of the continuous wavelet transform (CWT) for analyzing US signal scales.
• To provide a threshold-independent approach for localizing edges in complex US data.

Main Methods:

• Ultrasound image data beams were transformed into the wavelet domain using the continuous wavelet transform (CWT).
• Signal decomposition into multiple scales was performed, leveraging CWT's preservation of scale and time information.
• Edge localization was achieved by identifying the scale of maximum power spectrum and the peak position of the squared CWT at that scale.

Main Results:

• The proposed method successfully detected edges in various ultrasound images, including straight and jagged edges of submerged structures and an abdominal aorta aneurysm phantom.
• The technique demonstrated effective edge localization without relying on arbitrary thresholding after scale range determination.
• A general range of scales suitable for ultrasound images was suggested, enhancing method applicability.

Conclusions:

• The developed wavelet-based method offers a robust and accurate solution for edge detection in ultrasound imaging.
• The continuous wavelet transform provides a powerful tool for analyzing scale-dependent features in ultrasound signals.
• This threshold-independent approach holds promise for improving the analysis and interpretation of ultrasound imaging data.