Sensitivity to point-spread function parameters in medical ultrasound image deconvolution.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Clinical ultrasound images suffer from blurring and speckle, hindering interpretation.
• Image quality can be improved using deconvolution with an estimated point-spread function (PSF).
• Accurate PSF estimation in vivo is challenging due to unknown soft tissue properties affecting sound speed and attenuation.

Purpose of the Study:

• To investigate the sensitivity of a deconvolution algorithm to uncertainties in the point-spread function (PSF).
• To model a PSF for a clinical 128-element 1D linear array, varying lateral and axial dependencies.
• To analyze the impact of parameter variations on deconvolution performance for enhanced ultrasound imaging.

Main Methods:

• Developed a 2D deconvolution algorithm with a shift-variant axial PSF and shift-invariant lateral PSF.
• Conducted simulated and in vitro sensitivity analyses by varying six PSF-dependent parameters.
• Investigated ultrasound machine parameters (foci, element dimensions) and tissue properties (sound speed, attenuation).

Main Results:

• Logarithmic image amplitude effectively indicates perceived restoration improvement.
• Lateral focus and speed of sound are the most critical parameters for deconvolution success, primarily affecting deblurring.
• Similar sensitivity patterns were observed in simulations and in vitro experiments.

Conclusions:

• An exact point-spread function (PSF) is not strictly necessary for enhancing ultrasound image quality via deconvolution.
• Restoration of in vivo ultrasound images is feasible using an assumed PSF.
• This research provides insights into optimizing deconvolution for clinical ultrasound applications.