Constrained least squares filtering for ultrasound image deconvolution.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Ultrasound imaging quality is often degraded by speckle noise and low contrast.
• Accurate estimation of tissue signals is crucial for diagnostic clarity.

Purpose of the Study:

• To develop an ultrasound image deconvolution algorithm for enhanced image quality.
• To introduce a new tissue model incorporating random signal fluctuations.
• To improve speckle reduction and contrast ratio in ultrasound images.

Main Methods:

• A novel tissue model simulating random fluctuations in ultrasound radio frequency (RF) echo signals was developed.
• A modified regularization method combining optimal Wiener filtering and constrained least squares (LS) filtering was proposed for tissue signal estimation.
• Algorithm performance was evaluated through simulations.

Main Results:

• The proposed deconvolution algorithm demonstrated significant speckle reduction.
• A notable improvement in the contrast ratio of deconvolved ultrasound images was observed.
• The new tissue model effectively incorporated signal fluctuations for improved estimation.

Conclusions:

• The developed ultrasound deconvolution algorithm effectively enhances image quality.
• The novel tissue model and modified regularization method offer a promising approach for improving ultrasound diagnostics.
• This technique has the potential to improve the accuracy and reliability of ultrasound-based medical assessments.