DEEP NEURAL NETWORKS FOR NON-LINEAR MODEL-BASED ULTRASOUND RECONSTRUCTION
Abstract
Ultrasound reflection tomography is widely used to image large complex specimens that are only accessible from a single side, such as well systems and nuclear power plant containment walls. Typical methods for inverting the measurement rely on delay-and-sum algorithms that rapidly produce reconstructions but with significant artifacts. Recently, model-based reconstruction approaches using a linear forward model have been shown to significantly improve image quality compared to the conventional approach. Nonetheless, even these techniques result in artifacts for complex objects because of the inherent non-linearity of the ultrasound forward model.In this paper, we propose a non-iterative model-based reconstruction method for inverting measurements that are based on non-linear forward models for ultrasound imaging. Our approach involves obtaining an approximate estimate of the reconstruction using a simple linear back-projection and training a deep neural network to refine this to the actual reconstruction. As a result, we apply our method to simulated and experimental ultrasound data to demonstrate dramatic improvements in image quality compared to the delay-and-sum approach and the linear model-based reconstruction approach.
- Authors:
-
- Purdue Univ., West Lafayette, IN (United States)
- Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
- Publication Date:
- Research Org.:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Org.:
- USDOE Office of Energy Efficiency and Renewable Energy (EERE); USDOE Office of Fossil Energy (FE)
- OSTI Identifier:
- 1502581
- Grant/Contract Number:
- AC05-00OR22725
- Resource Type:
- Accepted Manuscript
- Journal Name:
- IEEE Global Conference on Signal and Information Processing
- Additional Journal Information:
- Journal Volume: 2018; Journal Issue: 1
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; Image reconstruction; Ultrasonic imaging; Mathematical model; Training; Neural networks; Imaging; Transducers
Citation Formats
Almansouri, H., Venkatakrishnan, S.V., Buzzard, G.T., Bouman, C.A., and Santos-Villalobos, H. DEEP NEURAL NETWORKS FOR NON-LINEAR MODEL-BASED ULTRASOUND RECONSTRUCTION. United States: N. p., 2019.
Web. doi:10.1109/GlobalSIP.2018.8646704.
Almansouri, H., Venkatakrishnan, S.V., Buzzard, G.T., Bouman, C.A., & Santos-Villalobos, H. DEEP NEURAL NETWORKS FOR NON-LINEAR MODEL-BASED ULTRASOUND RECONSTRUCTION. United States. https://doi.org/10.1109/GlobalSIP.2018.8646704
Almansouri, H., Venkatakrishnan, S.V., Buzzard, G.T., Bouman, C.A., and Santos-Villalobos, H. Thu .
"DEEP NEURAL NETWORKS FOR NON-LINEAR MODEL-BASED ULTRASOUND RECONSTRUCTION". United States. https://doi.org/10.1109/GlobalSIP.2018.8646704. https://www.osti.gov/servlets/purl/1502581.
@article{osti_1502581,
title = {DEEP NEURAL NETWORKS FOR NON-LINEAR MODEL-BASED ULTRASOUND RECONSTRUCTION},
author = {Almansouri, H. and Venkatakrishnan, S.V. and Buzzard, G.T. and Bouman, C.A. and Santos-Villalobos, H.},
abstractNote = {Ultrasound reflection tomography is widely used to image large complex specimens that are only accessible from a single side, such as well systems and nuclear power plant containment walls. Typical methods for inverting the measurement rely on delay-and-sum algorithms that rapidly produce reconstructions but with significant artifacts. Recently, model-based reconstruction approaches using a linear forward model have been shown to significantly improve image quality compared to the conventional approach. Nonetheless, even these techniques result in artifacts for complex objects because of the inherent non-linearity of the ultrasound forward model.In this paper, we propose a non-iterative model-based reconstruction method for inverting measurements that are based on non-linear forward models for ultrasound imaging. Our approach involves obtaining an approximate estimate of the reconstruction using a simple linear back-projection and training a deep neural network to refine this to the actual reconstruction. As a result, we apply our method to simulated and experimental ultrasound data to demonstrate dramatic improvements in image quality compared to the delay-and-sum approach and the linear model-based reconstruction approach.},
doi = {10.1109/GlobalSIP.2018.8646704},
journal = {IEEE Global Conference on Signal and Information Processing},
number = 1,
volume = 2018,
place = {United States},
year = {Thu Feb 21 00:00:00 EST 2019},
month = {Thu Feb 21 00:00:00 EST 2019}
}