Clutter filter design for ultrasound color flow imaging

Area of Science:

• Medical imaging
• Ultrasound technology
• Signal processing

Background:

• High-quality ultrasound color flow imaging necessitates sufficient suppression of clutter signals from stationary or slowly moving tissues.
• Inadequate clutter rejection hinders accurate measurement of low-velocity blood flow and introduces bias in higher velocity estimates.
• The limited number of available samples (8 to 16) presents a significant challenge for effective clutter filtering in color flow imaging.

Purpose of the Study:

• To review and analyze three classes of filters: finite impulse response (FIR), infinite impulse response (IIR), and regression filters.
• To assess filter quality based on frequency response and the bias and variance of mean blood velocity estimation.
• To identify optimal filter configurations for improved clutter suppression and velocity estimation in ultrasound color flow imaging.

Main Methods:

• Analysis of finite impulse response (FIR) filters, including non-linear phase response for improved frequency response.
• Evaluation of infinite impulse response (IIR) filters with different initialization methods (zero, step, projection) for short signal processing.
• Assessment of regression filters utilizing polynomial basis functions for effective clutter suppression.
• Comparison of filter performance using frequency response, bias, and variance of mean blood velocity estimates derived from an autocorrelation technique.

Main Results:

• Allowing non-linear phase response improved the frequency response of FIR filters.
• Projection initialization yielded superior IIR filters for short signal applications.
• Polynomial basis functions in regression filters provided effective clutter suppression.
• The best filters from each class (FIR, IIR, regression) demonstrated comparable bias and variance in mean blood velocity estimates.
• Polynomial regression and projection-initialized IIR filters exhibited slightly better frequency response compared to FIR filters.

Conclusions:

• Polynomial regression filters and projection-initialized IIR filters offer superior performance in clutter suppression and frequency response for ultrasound color flow imaging compared to traditional FIR filters.
• Effective clutter rejection is crucial for accurate low-velocity blood flow measurement and reducing bias in velocity estimation.
• The choice of filter initialization and basis functions significantly impacts the performance of clutter filtering in limited-sample ultrasound data.