Compressive Deconvolution in Medical Ultrasound Imaging

Area of Science:

• Medical Imaging
• Signal Processing
• Biomedical Engineering

Background:

• Compressive sampling (CS) has shown promise in ultrasound (US) imaging, enabling data reconstruction from fewer measurements and reduced pulse emissions.
• However, limitations in transducer bandwidth and wave propagation phenomena affect RF image spatial resolution, contrast, and signal-to-noise ratio (SNR).
• Existing deconvolution techniques aim to improve US image quality but are often applied post-reconstruction.

Purpose of the Study:

• To introduce a novel framework, compressive deconvolution, for reconstructing enhanced RF images directly from compressed measurements.
• To integrate data compression and image quality improvement into a unified acquisition and reconstruction model.
• To address the limitations of conventional ultrasound imaging by improving spatial resolution, contrast, and SNR.

Main Methods:

• Developed a unified formulation of the direct acquisition model combining random projections (compressive sensing) with 2D convolution using a spatially invariant point spread function.
• Implemented an optimization method based on the Alternating Direction Method of Multipliers (ADMM) for joint data volume reduction and image enhancement.
• Evaluated the compressive deconvolution framework using both simulated and in vivo ultrasound data.

Main Results:

• The compressive deconvolution framework successfully reconstructs enhanced RF images from compressed measurements, demonstrating joint data volume reduction and improved image quality.
• The proposed method shows potential in overcoming limitations related to transducer bandwidth and US wave propagation.
• Evaluation on simulated and in vivo data validates the effectiveness of the ADMM-based optimization for the proposed framework.

Conclusions:

• Compressive deconvolution offers a novel approach to improve ultrasound image quality by integrating compressed sensing and deconvolution.
• The unified framework provides a powerful tool for enhancing RF image resolution, contrast, and SNR in ultrasound imaging.
• This method holds significant potential for advancing ultrasound diagnostic capabilities through improved image reconstruction.