Accesso libero

Improving Image Quality in Electrical Impedance Tomography (EIT) Using Projection Error Propagation-Based Regularization (PEPR) Technique: A Simulation Study

Tushar Kanti Bera
Tushar Kanti Bera
, 
Samir Kumar Biswas
Samir Kumar Biswas
, 
K. Rajan
K. Rajan
 e 
J. Nagaraju
J. Nagaraju
   | 13 mar 2011
Journal of Electrical Bioimpedance's Cover Image
INFORMAZIONI SU QUESTO ARTICOLO
Articolo precedente
Articolo Successivo
Cita
Article Category: Articles
Pubblicato online: 13 mar 2011
Pagine: 2 - 12
Ricevuto: 31 gen 2011
DOI: https://doi.org/10.5617/jeb.158
Parole chiave
© 2011 Tushar Kanti Bera, Samir Kumar Biswas, K. Rajan and J. Nagaraju, published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Fig.1

An EIT system with electrode array on patient under test.
An EIT system with electrode array on patient under test.

Fig.2

Resistivity imaging for object near electrode 1: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.
Resistivity imaging for object near electrode 1: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.

Fig.3

Resistivity imaging for object near electrode 3: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.
Resistivity imaging for object near electrode 3: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.

Fig.-4

Resistivity imaging for object near electrode 5: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.
Resistivity imaging for object near electrode 5: (a) Original object, (b) with LMR, (c) with PEPR, (d) DRP of the images.

Fig.-5

Resistivity images with noisy (25 %) boundary data (object near electrode 1): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image
Resistivity images with noisy (25 %) boundary data (object near electrode 1): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image

Fig.-6

Resistivity images with noisy (25 %) boundary data (object near electrode 3): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image
Resistivity images with noisy (25 %) boundary data (object near electrode 3): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image

Fig.-7

Resistivity images with noisy (25 %) boundary data (object near electrode 5): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image
Resistivity images with noisy (25 %) boundary data (object near electrode 5): (a) with LMR, (b) with PEPR, (c) DRP of the reconstructed image

Fig.-8

Resistivity images obtained from boundary data with 10 % noise (object near electrode 1): (a) image with LMR, (b) DRP of the reconstructed image shown in Fig.-1a, (c) image with PEPR, (d) DRP of the reconstructed image shown in Fig.-1c.
Resistivity images obtained from boundary data with 10 % noise (object near electrode 1): (a) image with LMR, (b) DRP of the reconstructed image shown in Fig.-1a, (c) image with PEPR, (d) DRP of the reconstructed image shown in Fig.-1c.

Fig.-9

Resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR (λ = 0.01) method (object near electrode 1) for first twelve iterations: (a) to (l) images represents the reconstruction of 1 to 12 iterations respectively.
Resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR (λ = 0.01) method (object near electrode 1) for first twelve iterations: (a) to (l) images represents the reconstruction of 1 to 12 iterations respectively.

Fig.-10

DRP of the resistivity images shown in Fig.-9 (reconstruction with LMR): with noisy data (object near electrode 3): (a) to (l) images represents the DRP of the images shown in Fig.-9a to Fig.-9l respectively.
DRP of the resistivity images shown in Fig.-9 (reconstruction with LMR): with noisy data (object near electrode 3): (a) to (l) images represents the DRP of the images shown in Fig.-9a to Fig.-9l respectively.

Fig.-11

Resistivity images obtained from noisy boundary data (Error added = 10 %) with PEPR (Ψ = 0.01) method (object near electrode 1) for first twelve iterations: (a) to (l) images represents the reconstruction of 1 to 12 iterations respectively.
Resistivity images obtained from noisy boundary data (Error added = 10 %) with PEPR (Ψ = 0.01) method (object near electrode 1) for first twelve iterations: (a) to (l) images represents the reconstruction of 1 to 12 iterations respectively.

Fig.-12

DRP of the resistivity images shown in Fig.-11 (reconstruction with PEPR): with noisy data (object near electrode 1): (a) to (l) images represents the DRP of the images shown in Fig.-11a to Fig.-11l respectively.
DRP of the resistivity images shown in Fig.-11 (reconstruction with PEPR): with noisy data (object near electrode 1): (a) to (l) images represents the DRP of the images shown in Fig.-11a to Fig.-11l respectively.

Fig.-13

Reconstruction parameters and reconstruction errors for the resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR (λ = 0.01) and PEPR (Ψ = 0.01) methods (object near electrode 1): (a) maximum values of the reconstructed inhomogeneity resistivity (IRMax), (b) mean of reconstructed inhomogeneity resistivity (IRMean).
Reconstruction parameters and reconstruction errors for the resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR (λ = 0.01) and PEPR (Ψ = 0.01) methods (object near electrode 1): (a) maximum values of the reconstructed inhomogeneity resistivity (IRMax), (b) mean of reconstructed inhomogeneity resistivity (IRMean).

Fig.-14

Reconstruction parameters and reconstruction errors for the resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR ( λ = 0.01) and PEPR ( Ψ = 0 .01) methods (object near electrode 1): (a) projection error (EV), (b) normalized solution error norm (Eρ).
Reconstruction parameters and reconstruction errors for the resistivity images obtained from noisy boundary data (Error added = 10 %) with LMR ( λ = 0.01) and PEPR ( Ψ = 0 .01) methods (object near electrode 1): (a) projection error (EV), (b) normalized solution error norm (Eρ).

Fig.-15

Projection errors (EV) calculated in the resistivity reconstruction at different iterations (a) with LMR (λ = 1 to 0.0001), (b) with PEPR (Ψ = 1 to 0.0001).
Projection errors (EV) calculated in the resistivity reconstruction at different iterations (a) with LMR (λ = 1 to 0.0001), (b) with PEPR (Ψ = 1 to 0.0001).

CNR, PCR and COC of reconstructed images for noisy data (object near electrode 1)

RegularizationCNRPCRCOC
LMR1.8127.291.85
PEPR3.1763.792.90

CNR, PCR and COC of reconstructed images for noisy data (object near electrode 3)

RegularizationCNRPCRCOC
LMR1.6023.611.73
PEPR1.8935.682.12

CNR, PCR and COC of reconstructed images of the object near electrode 5.

RegularizationCNRPCRCOC
LMR2.9230.321.98
PEPR3.4644.742.40

CNR, PCR and COC of reconstructed images of the object near electrode 3

RegularizationCNRPCRCOC
LMR2.9931.332.01
PEPR3.5146.262.45

CNR, PCR and COC of reconstructed images for noisy data (object near electrode 5)

RegularizationCNRPCRCOC
LMR1.2417.591.54
PEPR1.4825.911.78

CNR, PCR and COC of reconstructed images of the object near electrode 1

RegularizationCNRPCRCOC
LMR3.0832.522.05
PEPR3.5546.572.46

Projection errors (EV) calculated for different values of λ in LMR and Ψ in PEPR.

λ or Ψ at 4th IterationLMRPEPR
1.051.824853.3685
0.151.384451.7703
0.0151.501651.3234
0.00155.947351.3802
0.000170.839456.8654

CNR, PCR, COC, IRMax and IRMean of reconstructed images shown in Fig.-8.

RegularizationCNRPCRCOCIRMaxIRMean
LMR2.9589.152.9077.0341.46
PEPR3.2646.482.4933.9623.70
eISSN:
1891-5469
Lingua:
Inglese
Frequenza di pubblicazione:
Volume Open
Argomenti della rivista:
Engineering, Bioengineering and Biomedical Engineering, Biomedical Electronics, Life Sciences, Biophysics, Medicine, Biomedical Engineering, Physics, Spectroscopy and Metrology
Feed RSS della rivista