Nonlocal means filter-based speckle tracking.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Sensorless freehand 3-D ultrasound imaging aims to reduce hardware costs and complexity.
• Accurate out-of-plane pose estimation remains a significant challenge for full 6-degree-of-freedom (DoF) tracking in ultrasound.

Purpose of the Study:

• To develop a novel filter-based speckle tracking framework to enhance the accuracy of out-of-plane displacement estimation.
• To improve the precision of 6-DoF tracking in freehand 3-D ultrasound imaging.

Main Methods:

• A nonlocal means (NLM) filter utilizing a Rician-inverse Gaussian (RiIG) probabilistic model for ultrasound speckle was developed.
• Stein's unbiased risk estimate (SURE) was derived and used as a weighting factor to aggregate local displacement estimations.
• The framework was evaluated using three noise models, including RiIG, and compared against a previous model-based algorithm.

Main Results:

• The proposed RiIG filter-based speckle tracking method demonstrated higher accuracy and reduced tissue dependency compared to existing methods in ex vivo experiments.
• In vivo evaluation on human subjects showed improved tracking accuracy for small lateral displacements and tilt rotations.
• The method achieved higher tracking accuracy compared to electromagnetic tracker measurements.

Conclusions:

• The developed filter-based speckle tracking framework significantly improves out-of-plane displacement estimation accuracy for sensorless 3-D ultrasound.
• This method offers a more accurate and less tissue-dependent solution for 6-DoF tracking, showing potential for clinical applications.