TY - GEN
T1 - Plug-and-Play Enhanced Compressive Sensing for Limited Sample PAT Image Reconstruction
AU - John, Mary
AU - Barhumi, Imad
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Medical imaging technologies are crucial for precise disease diagnosis, encompassing X-rays, Magnetic Resonance Imaging (MRI), and Ultrasound scans. However, these methods have limitations such as ionizing radiation, cost, time consumption, or compromised image quality. Photoacoustic Tomography (PAT) is a promising noninvasive technique providing high-resolution, high-contrast images in deep tissues. In this paper we integrate the plug-and-play framework to enhance compressive sensing algorithms for PAT imaging with reduced samples as it offers faster image reconstruction, optimized memory, and expedited data transmission in real-world PAT applications. Our approach formulates and evaluates novel Split Bregman Total Variation (SBTV) and Relaxed Basis Pursuit ADMM (rBP-ADMM) algorithms fortified with Plug-and-Play (PnP) denoising techniques. These algorithms are assessed under diminishing sample conditions with various PnP denoisers. Notably, SBTV coupled with Bilateral Filtering as a PnP denoiser yields significant SSIM enhancements (11-17%). These findings highlight the potential of PnP-enhanced SBTV for robust image reconstruction in low-sample scenarios, addressing a critical need for efficient PAT applications.
AB - Medical imaging technologies are crucial for precise disease diagnosis, encompassing X-rays, Magnetic Resonance Imaging (MRI), and Ultrasound scans. However, these methods have limitations such as ionizing radiation, cost, time consumption, or compromised image quality. Photoacoustic Tomography (PAT) is a promising noninvasive technique providing high-resolution, high-contrast images in deep tissues. In this paper we integrate the plug-and-play framework to enhance compressive sensing algorithms for PAT imaging with reduced samples as it offers faster image reconstruction, optimized memory, and expedited data transmission in real-world PAT applications. Our approach formulates and evaluates novel Split Bregman Total Variation (SBTV) and Relaxed Basis Pursuit ADMM (rBP-ADMM) algorithms fortified with Plug-and-Play (PnP) denoising techniques. These algorithms are assessed under diminishing sample conditions with various PnP denoisers. Notably, SBTV coupled with Bilateral Filtering as a PnP denoiser yields significant SSIM enhancements (11-17%). These findings highlight the potential of PnP-enhanced SBTV for robust image reconstruction in low-sample scenarios, addressing a critical need for efficient PAT applications.
KW - Bilateral Filtering
KW - Compressive Sensing
KW - Photoacoustic Tomography
KW - Plug-and-Play Framework
KW - Split Bregman Total Variation
UR - http://www.scopus.com/inward/record.url?scp=85182275174&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85182275174&partnerID=8YFLogxK
U2 - 10.1109/ICSPIS60075.2023.10343788
DO - 10.1109/ICSPIS60075.2023.10343788
M3 - Conference contribution
AN - SCOPUS:85182275174
T3 - 2023 6th International Conference on Signal Processing and Information Security, ICSPIS 2023
SP - 110
EP - 115
BT - 2023 6th International Conference on Signal Processing and Information Security, ICSPIS 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 6th International Conference on Signal Processing and Information Security, ICSPIS 2023
Y2 - 8 November 2023 through 9 November 2023
ER -