Image reconstruction of moderate contrast targets using the distorted born iterative method

Roberto Lavarello, Michael Oelze

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Ultrasonic tomography using inverse scattering, i.e., the distorted Born iterative method (DBIM), allows for the quantitative image reconstruction of mechanical properties of materials. The use of multi-frequency information has been proposed to avoid convergence issues for targets with moderate speed of sound contrasts Δc (i.e., targets for which the excess phase Δφ when propagating through the object is larger than π) but not validated experimentally. Furthermore, DBIM has to be regularized due to its ill-conditioning. To experimentally validate DBIM for use in ultrasonic tomography, a systematic procedure to choose the regularization parameter based on the Rayleigh quotient iteration was developed and images of objects with moderate Ac were reconstructed. The performance of DBIM using the developed regularization scheme was studied through: 1) Simulations of a two dimensional (2D) phantom (Δφ = 2.14π) with inclusions smaller than a wavelength, 2) Experiments with a balloon phantom with high Ac with scattered data collected at 0.64 MHz (Δφ = 0.84π) and 1.2 MHz (Δφ = 1.6π), and 3) Three dimensional (3D) reconstruction of a moderate contrast sphere (Δφ = 1.32π). The 2D and 3D multi-frequency DBIM simulations were successfully stabilized by the proposed regularization scheme as evidenced by the low reconstruction mean square errors (MSEs) (Ac MSE = 9% for 2D and 15% for 3D) and the detection of the sub-wavelength inclusions in the 2D reconstruction. In experiments, the measured scattered fields agreed well with the predicted scattered fields from the phantom model (MSE = 4% for 0.64 MHz and MSE = 7.8% for 1.2 MHz). The MSE of the reconstructed image using only the experimental data at 0.64 MHz was 19%. Using the experimental data at 1.2 MHz to refine the 0.64 MHz reconstruction allowed the MSE to be reduced to 12%, and improved the spatial resolution as evidenced by the reduced edge blurring.

Original languageEnglish (US)
Title of host publication2007 IEEE Ultrasonics Symposium Proceedings, IUS
Pages1993-1996
Number of pages4
DOIs
StatePublished - Dec 1 2007
Event2007 IEEE Ultrasonics Symposium, IUS - New York, NY, United States
Duration: Oct 28 2007Oct 31 2007

Publication series

NameProceedings - IEEE Ultrasonics Symposium
ISSN (Print)1051-0117

Other

Other2007 IEEE Ultrasonics Symposium, IUS
CountryUnited States
CityNew York, NY
Period10/28/0710/31/07

Fingerprint

image reconstruction
tomography
ultrasonics
inclusions
quotients
blurring
inverse scattering
balloons
conditioning
wavelengths
iteration
simulation
spatial resolution
mechanical properties
acoustics

ASJC Scopus subject areas

  • Acoustics and Ultrasonics

Cite this

Lavarello, R., & Oelze, M. (2007). Image reconstruction of moderate contrast targets using the distorted born iterative method. In 2007 IEEE Ultrasonics Symposium Proceedings, IUS (pp. 1993-1996). [4410074] (Proceedings - IEEE Ultrasonics Symposium). https://doi.org/10.1109/ULTSYM.2007.501

Image reconstruction of moderate contrast targets using the distorted born iterative method. / Lavarello, Roberto; Oelze, Michael.

2007 IEEE Ultrasonics Symposium Proceedings, IUS. 2007. p. 1993-1996 4410074 (Proceedings - IEEE Ultrasonics Symposium).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Lavarello, R & Oelze, M 2007, Image reconstruction of moderate contrast targets using the distorted born iterative method. in 2007 IEEE Ultrasonics Symposium Proceedings, IUS., 4410074, Proceedings - IEEE Ultrasonics Symposium, pp. 1993-1996, 2007 IEEE Ultrasonics Symposium, IUS, New York, NY, United States, 10/28/07. https://doi.org/10.1109/ULTSYM.2007.501
Lavarello R, Oelze M. Image reconstruction of moderate contrast targets using the distorted born iterative method. In 2007 IEEE Ultrasonics Symposium Proceedings, IUS. 2007. p. 1993-1996. 4410074. (Proceedings - IEEE Ultrasonics Symposium). https://doi.org/10.1109/ULTSYM.2007.501
Lavarello, Roberto ; Oelze, Michael. / Image reconstruction of moderate contrast targets using the distorted born iterative method. 2007 IEEE Ultrasonics Symposium Proceedings, IUS. 2007. pp. 1993-1996 (Proceedings - IEEE Ultrasonics Symposium).
@inproceedings{73fd66fa70404a57a88fa82088887916,
title = "Image reconstruction of moderate contrast targets using the distorted born iterative method",
abstract = "Ultrasonic tomography using inverse scattering, i.e., the distorted Born iterative method (DBIM), allows for the quantitative image reconstruction of mechanical properties of materials. The use of multi-frequency information has been proposed to avoid convergence issues for targets with moderate speed of sound contrasts Δc (i.e., targets for which the excess phase Δφ when propagating through the object is larger than π) but not validated experimentally. Furthermore, DBIM has to be regularized due to its ill-conditioning. To experimentally validate DBIM for use in ultrasonic tomography, a systematic procedure to choose the regularization parameter based on the Rayleigh quotient iteration was developed and images of objects with moderate Ac were reconstructed. The performance of DBIM using the developed regularization scheme was studied through: 1) Simulations of a two dimensional (2D) phantom (Δφ = 2.14π) with inclusions smaller than a wavelength, 2) Experiments with a balloon phantom with high Ac with scattered data collected at 0.64 MHz (Δφ = 0.84π) and 1.2 MHz (Δφ = 1.6π), and 3) Three dimensional (3D) reconstruction of a moderate contrast sphere (Δφ = 1.32π). The 2D and 3D multi-frequency DBIM simulations were successfully stabilized by the proposed regularization scheme as evidenced by the low reconstruction mean square errors (MSEs) (Ac MSE = 9{\%} for 2D and 15{\%} for 3D) and the detection of the sub-wavelength inclusions in the 2D reconstruction. In experiments, the measured scattered fields agreed well with the predicted scattered fields from the phantom model (MSE = 4{\%} for 0.64 MHz and MSE = 7.8{\%} for 1.2 MHz). The MSE of the reconstructed image using only the experimental data at 0.64 MHz was 19{\%}. Using the experimental data at 1.2 MHz to refine the 0.64 MHz reconstruction allowed the MSE to be reduced to 12{\%}, and improved the spatial resolution as evidenced by the reduced edge blurring.",
author = "Roberto Lavarello and Michael Oelze",
year = "2007",
month = "12",
day = "1",
doi = "10.1109/ULTSYM.2007.501",
language = "English (US)",
isbn = "1424413834",
series = "Proceedings - IEEE Ultrasonics Symposium",
pages = "1993--1996",
booktitle = "2007 IEEE Ultrasonics Symposium Proceedings, IUS",

}

TY - GEN

T1 - Image reconstruction of moderate contrast targets using the distorted born iterative method

AU - Lavarello, Roberto

AU - Oelze, Michael

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Ultrasonic tomography using inverse scattering, i.e., the distorted Born iterative method (DBIM), allows for the quantitative image reconstruction of mechanical properties of materials. The use of multi-frequency information has been proposed to avoid convergence issues for targets with moderate speed of sound contrasts Δc (i.e., targets for which the excess phase Δφ when propagating through the object is larger than π) but not validated experimentally. Furthermore, DBIM has to be regularized due to its ill-conditioning. To experimentally validate DBIM for use in ultrasonic tomography, a systematic procedure to choose the regularization parameter based on the Rayleigh quotient iteration was developed and images of objects with moderate Ac were reconstructed. The performance of DBIM using the developed regularization scheme was studied through: 1) Simulations of a two dimensional (2D) phantom (Δφ = 2.14π) with inclusions smaller than a wavelength, 2) Experiments with a balloon phantom with high Ac with scattered data collected at 0.64 MHz (Δφ = 0.84π) and 1.2 MHz (Δφ = 1.6π), and 3) Three dimensional (3D) reconstruction of a moderate contrast sphere (Δφ = 1.32π). The 2D and 3D multi-frequency DBIM simulations were successfully stabilized by the proposed regularization scheme as evidenced by the low reconstruction mean square errors (MSEs) (Ac MSE = 9% for 2D and 15% for 3D) and the detection of the sub-wavelength inclusions in the 2D reconstruction. In experiments, the measured scattered fields agreed well with the predicted scattered fields from the phantom model (MSE = 4% for 0.64 MHz and MSE = 7.8% for 1.2 MHz). The MSE of the reconstructed image using only the experimental data at 0.64 MHz was 19%. Using the experimental data at 1.2 MHz to refine the 0.64 MHz reconstruction allowed the MSE to be reduced to 12%, and improved the spatial resolution as evidenced by the reduced edge blurring.

AB - Ultrasonic tomography using inverse scattering, i.e., the distorted Born iterative method (DBIM), allows for the quantitative image reconstruction of mechanical properties of materials. The use of multi-frequency information has been proposed to avoid convergence issues for targets with moderate speed of sound contrasts Δc (i.e., targets for which the excess phase Δφ when propagating through the object is larger than π) but not validated experimentally. Furthermore, DBIM has to be regularized due to its ill-conditioning. To experimentally validate DBIM for use in ultrasonic tomography, a systematic procedure to choose the regularization parameter based on the Rayleigh quotient iteration was developed and images of objects with moderate Ac were reconstructed. The performance of DBIM using the developed regularization scheme was studied through: 1) Simulations of a two dimensional (2D) phantom (Δφ = 2.14π) with inclusions smaller than a wavelength, 2) Experiments with a balloon phantom with high Ac with scattered data collected at 0.64 MHz (Δφ = 0.84π) and 1.2 MHz (Δφ = 1.6π), and 3) Three dimensional (3D) reconstruction of a moderate contrast sphere (Δφ = 1.32π). The 2D and 3D multi-frequency DBIM simulations were successfully stabilized by the proposed regularization scheme as evidenced by the low reconstruction mean square errors (MSEs) (Ac MSE = 9% for 2D and 15% for 3D) and the detection of the sub-wavelength inclusions in the 2D reconstruction. In experiments, the measured scattered fields agreed well with the predicted scattered fields from the phantom model (MSE = 4% for 0.64 MHz and MSE = 7.8% for 1.2 MHz). The MSE of the reconstructed image using only the experimental data at 0.64 MHz was 19%. Using the experimental data at 1.2 MHz to refine the 0.64 MHz reconstruction allowed the MSE to be reduced to 12%, and improved the spatial resolution as evidenced by the reduced edge blurring.

UR - http://www.scopus.com/inward/record.url?scp=48149104361&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=48149104361&partnerID=8YFLogxK

U2 - 10.1109/ULTSYM.2007.501

DO - 10.1109/ULTSYM.2007.501

M3 - Conference contribution

AN - SCOPUS:48149104361

SN - 1424413834

SN - 9781424413836

T3 - Proceedings - IEEE Ultrasonics Symposium

SP - 1993

EP - 1996

BT - 2007 IEEE Ultrasonics Symposium Proceedings, IUS

ER -