TY - JOUR

T1 - Experimental study of one step linear Gauss-Newton algorithm for improving the quality of image reconstruction in high-speed Electrical Impedance Tomography (EIT)

AU - Widodo, A.

AU - Endarko,

N1 - Publisher Copyright:
© Published under licence by IOP Publishing Ltd.

PY - 2018/12/23

Y1 - 2018/12/23

N2 - One step linear Gauss-Newton was well known as one of image reconstruction algorithm in Electrical Impedance Tomography (EIT). In the present study, we proposed the mathematical model of one step linear Gauss-Newton for processing the experimental data for achieving the best image reconstruction. Two ways were conducted to solve this study. The first one, we build EIT hardware that mainly controlled by Single Board Computer (SBC) to produce a constant current at 1 mA 20 kHz that will be injected to the 16 electrodes of the practical phantom with speed of 1 ms/data for acquiring the data of voltage boundary from that practical phantom when using aluminium as anomaly. The second one, we processed the data of voltage boundary using one-step linear Gauss-Newton to achieve the best image reconstruction with modified the exponent (p) of Regularization matrix that varied at 0 - 1 by the increment of 0.1 and hyperparameter value (λ). The result showed that the best image reconstruction obtained at p = 0.7 and λ = 8 for this algorithm.

AB - One step linear Gauss-Newton was well known as one of image reconstruction algorithm in Electrical Impedance Tomography (EIT). In the present study, we proposed the mathematical model of one step linear Gauss-Newton for processing the experimental data for achieving the best image reconstruction. Two ways were conducted to solve this study. The first one, we build EIT hardware that mainly controlled by Single Board Computer (SBC) to produce a constant current at 1 mA 20 kHz that will be injected to the 16 electrodes of the practical phantom with speed of 1 ms/data for acquiring the data of voltage boundary from that practical phantom when using aluminium as anomaly. The second one, we processed the data of voltage boundary using one-step linear Gauss-Newton to achieve the best image reconstruction with modified the exponent (p) of Regularization matrix that varied at 0 - 1 by the increment of 0.1 and hyperparameter value (λ). The result showed that the best image reconstruction obtained at p = 0.7 and λ = 8 for this algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85059419867&partnerID=8YFLogxK

U2 - 10.1088/1742-6596/1120/1/012067

DO - 10.1088/1742-6596/1120/1/012067

M3 - Conference article

AN - SCOPUS:85059419867

SN - 1742-6588

VL - 1120

JO - Journal of Physics: Conference Series

JF - Journal of Physics: Conference Series

IS - 1

M1 - 012067

T2 - 8th International Conference on Theoretical and Applied Physics, ICTAP 2018

Y2 - 20 September 2018 through 21 September 2018

ER -