Point detection in textured ultrasound images

Area of Science:

• Medical Imaging
• Ultrasound Technology
• Signal Processing

Background:

• Point scatterer detection in ultrasound images is often hindered by complex textures.
• Existing methods may struggle with optimal texture correction, limiting detection accuracy.

Purpose of the Study:

• To investigate the effectiveness of four multilook methods for improving point scatterer detection in textured ultrasound images.
• To identify which multilook methods offer the most significant performance gains under various imaging conditions.

Main Methods:

• Analysis of ultrasound images with known point scatterer locations and random textures.
• Evaluation of normalized matched filter (NMF) and multilook coherence factor (MLCF) methods.
• Assessment of weighted MLCF and NMF (NMFW) methods, including those applied to prewhitened and texture-corrected images.

Main Results:

• Normalized methods like NMF and MLCF do not require prior texture correction, proving beneficial when correction is difficult.
• Significant detection performance improvement was observed when MLCF was weighted with prewhitened and texture-corrected images.
• NMF and NMFW methods are highly effective for images dominated by acoustic noise over speckle background.

Conclusions:

• Multilook methods, particularly MLCF, substantially improve point scatterer detection in challenging ultrasound images.
• Weighted MLCF offers robust performance even without prior knowledge of optimal prewhitening limits.
• NMF and NMFW provide favorable solutions for ultrasound images with dominant acoustic noise.