Application of Compressive Sensing to Ultrasound Images: A Review

Area of Science:

• Medical Imaging
• Signal Processing
• Ultrasound Technology

Background:

• Compressive sensing (CS) enables data acquisition below the Nyquist rate, offering significant reductions in sensing, transmission, and storage for medical imaging.
• The efficacy of CS relies on data sparsity, either in its original form or a transformed domain, making it suitable for applications like ultrasound (US) imaging.
• Numerous algorithms exist for accurate data recovery from undersampled data using CS, though often with trade-offs in computational efficiency.

Purpose of the Study:

• To review significant compressive sensing algorithms for recovering undersampled ultrasound images.
• To discuss the application and challenges of CS in 3D ultrasound imaging.
• To provide a classification and summary of sparse recovery algorithms used in US.

Main Methods:

• Classification of sparse recovery algorithms applied to ultrasound into five distinct groups.
• Discussion and summarization of algorithms based on technique, compression ratio, and sparsifying transform.
• Inclusion of considerations for 3D ultrasound and deep learning applications within the CS framework.

Main Results:

• A comprehensive review of various CS algorithms for US image recovery from undersampled data.
• Categorization of algorithms highlights differences in techniques, compression capabilities, and sparsifying transforms.
• The study addresses the specific challenges and potential of CS in 3D US imaging.

Conclusions:

• Identifies research gaps and outlines future directions in compressive sensing for ultrasound applications.
• The review aims to guide early-career researchers interested in CS and its specific applications in US.
• Highlights the growing importance of deep learning in advancing CS techniques for medical imaging.