Ultrasound synthetic aperture imaging with SpecularNet for enhanced needle visualization

Area of Science:

• Medical Imaging
• Ultrasound Technology
• Artificial Intelligence in Medicine

Background:

• Ultrasound is vital for real-time, cost-effective needle guidance.
• Conventional ultrasound struggles with visualizing oblique needles due to echo reflection away from the transducer.

Purpose of the Study:

• To enhance ultrasound visualization of needles, particularly oblique ones.
• To improve image contrast and reduce artifacts in needle detection.

Main Methods:

• Developed specular synthetic transmit aperture (STA) beamforming with delay and standard deviation (DASD).
• Integrated convolutional neural networks (CNN) via SpecularNet to predict needle orientation and specular probability.
• Utilized pixel-wise STA channel data for specular information estimation.

Main Results:

• Specular STA-DASD improved needle-to-speckle ratio (NSR) by 11.0 dB and needle-to-artifact ratio (NAR) by 18.9 dB for 10°-40° needle orientations compared to conventional STA-DASD.
• SpecularNet efficiently inferred specular information, leading to further NSR and NAR improvements of 25-30 dB over standard specular beamforming.

Conclusions:

• Demonstrated the efficacy of specular STA-DASD beamforming for enhanced needle visualization.
• Showcased SpecularNet's computational efficiency in predicting specular information.
• Highlighted the potential for improved needle guidance through noise and artifact suppression.