Publication detail

Algebraic Reconstruction Technique for Ultrasound Transmission Tomography

Igor Peterlík, Radovan Jiřík, Nicole Ruiter, Rainer Stotzka, Jiří Jan Radim Kolář

Original Title

Algebraic Reconstruction Technique for Ultrasound Transmission Tomography

Czech Title

Algebraická rekonstrukční technika pro průzvučnou ultrazvukovou tomografii

English Title

Algebraic Reconstruction Technique for Ultrasound Transmission Tomography

Type

conference paper

Language

en

Original Abstract

Ultrasound transmission tomography is potentially promising alternative to standard X-Ray imaging in medical diagnosis, especially in mammography. The reconstruction of the local attenuation coefficient from the measured signals can be formulated as a large overdetermined system of linear equations based on a simplified ultrasound transmission model. It can be solved by means of the Kaczmarz algebraic reconstruction technique. The algorithm successively iterates through the equations and computes the corrections of the initial solution estimates. Because the original version of the algorithm does not guarantee convergence to the optimum, an extended version of the method is used here. It has been shown previously to converge to the least-mean-squares optimum. Both the original and extended algorithms are strictly sequential since the computation in the particular iteration depends on the corrections from the previous step. To enable parallelization of the method, thus speeding up the computation, a partitioning scheme is proposed and analyzed. The sequential as well as the partitioning-scheme algorithms are tested on both synthetic and real radiofrequency data (acquired using an experimental tomograph).

Czech abstract

Průzvučná ultrazvuková tomografie může být v oblasti mamografie alternativou k používaným rentgenovým zobrazovacím přístrojům a klasickým ultrazvukovým zobrazovacím systémům. Problém rekonstrukce útlumových map je zde formulován jako přeurčená soustava lineárních rovnic. Článek prezentuje některé přístupy a jejich modifikace k řešení tohoto problému, založené na Kaczmarzově metodě. Algoritmy jsou testovány na syntetický datech a datech snímaných na umělém objektu pomocí experimentálního ultrazvukového tomografu.

English abstract

Ultrasound transmission tomography is potentially promising alternative to standard X-Ray imaging in medical diagnosis, especially in mammography. The reconstruction of the local attenuation coefficient from the measured signals can be formulated as a large overdetermined system of linear equations based on a simplified ultrasound transmission model. It can be solved by means of the Kaczmarz algebraic reconstruction technique. The algorithm successively iterates through the equations and computes the corrections of the initial solution estimates. Because the original version of the algorithm does not guarantee convergence to the optimum, an extended version of the method is used here. It has been shown previously to converge to the least-mean-squares optimum. Both the original and extended algorithms are strictly sequential since the computation in the particular iteration depends on the corrections from the previous step. To enable parallelization of the method, thus speeding up the computation, a partitioning scheme is proposed and analyzed. The sequential as well as the partitioning-scheme algorithms are tested on both synthetic and real radiofrequency data (acquired using an experimental tomograph).

Keywords

Algebraic reconstruction techniques, inverse Radon transform

RIV year

2006

Released

01.01.2006

Publisher

The University of Ioannina

Location

Ioannina, Řecko

Pages from

1

Pages to

6

Pages count

6

BibTex


@inproceedings{BUT24212,
  author="Igor {Peterlík} and Radovan {Jiřík} and Jiří {Jan} and Radim {Kolář}",
  title="Algebraic Reconstruction Technique for Ultrasound Transmission Tomography",
  annote="Ultrasound transmission tomography is potentially promising alternative to standard X-Ray imaging in medical diagnosis, especially in mammography. The reconstruction of the local attenuation coefficient from the measured signals can be formulated as a large overdetermined system of linear equations based on a simplified ultrasound transmission model. It can be solved by means of the Kaczmarz algebraic reconstruction technique. The algorithm successively iterates through the equations and computes the corrections of the initial solution estimates. Because the original version of the algorithm does not guarantee convergence to the optimum, an extended version of the method is used here. It has been shown previously to converge to the least-mean-squares optimum. Both the original and extended algorithms are strictly sequential since the computation in the particular iteration depends on the corrections from the previous step. To enable parallelization of the method, thus speeding up the computation, a partitioning scheme is proposed and analyzed. The sequential as well as the partitioning-scheme algorithms are tested on both synthetic and real radiofrequency data (acquired using an experimental tomograph). 
",
  address="The University of Ioannina",
  booktitle="Proceedings of International Conference ITAB 2006",
  chapter="24212",
  institution="The University of Ioannina",
  year="2006",
  month="january",
  pages="1",
  publisher="The University of Ioannina",
  type="conference paper"
}