×

Classification of stroke using neural networks in electrical impedance tomography. (English) Zbl 1451.92178

Summary: Electrical impedance tomography (EIT) is an emerging non-invasive medical imaging modality. It is based on feeding electrical currents into the patient, measuring the resulting voltages at the skin, and recovering the internal conductivity distribution. The mathematical task of EIT image reconstruction is a nonlinear and ill-posed inverse problem. Therefore any EIT image reconstruction method needs to be regularized, typically resulting in blurred images. One promising application is stroke-EIT, or classification of stroke into either ischemic or hemorrhagic. Ischemic stroke involves a blood clot, preventing blood flow to a part of the brain causing a low-conductivity region. Hemorrhagic stroke means bleeding in the brain causing a high-conductivity region. In both cases the symptoms are identical, so a cost-effective and portable classification device is needed. Typical EIT images are not optimal for stroke-EIT because of blurriness. This paper explores the possibilities of machine learning in improving the classification results. Two paradigms are compared: (a) learning from the EIT data, that is Dirichlet-to-Neumann maps and (b) extracting robust features from data and learning from them. The features of choice are virtual hybrid edge detection (VHED) functions [A. Greenleaf et al., Anal. PDE 11, No. 8, 1901–1943 (2018; Zbl 1388.35225)] that have a geometric interpretation and whose computation from EIT data does not involve calculating a full image of the conductivity. We report the measures of accuracy, sensitivity and specificity of the networks trained with EIT data and VHED functions separately. Computational evidence based on simulated noisy EIT data suggests that the regularized grey-box paradigm (b) leads to significantly better classification results than the black-box paradigm (a).

MSC:

92C55 Biomedical imaging and signal processing
92B20 Neural networks for/in biological studies, artificial life and related topics

Citations:

Zbl 1388.35225
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Alsaker M and Mueller J L 2019 EIT images of human inspiration and expiration using a D-bar method with spatial priors J. Appl. Comput. Electromagn. Soc.34 325-30
[2] Alsaker M, Mueller J L and Murthy R 2019 Dynamic optimized priors for D-bar reconstructions of human ventilation using electrical impedance tomography J. Comput. Appl. Math.362 276-94 · Zbl 1418.92067 · doi:10.1016/j.cam.2018.07.039
[3] Simon A, Maass P, Öktem O and Schönlieb C-B 2019 Solving inverse problems using data-driven models Acta Numer.28 1-174 · Zbl 1429.65116 · doi:10.1017/s0962492919000059
[4] Astala K and Païvärinta L 2006 Calderón’s inverse conductivity problem in the plane Ann. Math.163 265-99 · Zbl 1111.35004 · doi:10.4007/annals.2006.163.265
[5] Bayford R and Tizzard A 2012 Bioimpedance imaging: an overview of potential clinical applications Analyst137 4635-43 · doi:10.1039/c2an35874c
[6] Beals R and Coifman R R 1984 Scattering and inverse scattering for first order systems Commun. Pure Appl. Math.37 39-90 · Zbl 0514.34021 · doi:10.1002/cpa.3160370105
[7] Bishop C M 1995 Training with noise is equivalent to Tikhonov regularization Neural Comput.7 108 · doi:10.1162/neco.1995.7.1.108
[8] Boverman G, Kao T-J, Wang X, Ashe J M, Davenport D M and Amm B C 2016 Detection of small bleeds in the brain with electrical impedance tomography Physiol. Meas.37 727 · doi:10.1088/0967-3334/37/6/727
[9] Calderón A-P 1980 On an inverse boundary value problem Seminar on Numerical Analysis and its Applications to Continuum Physics(Rio de Janeiro, 1980) (Rio de Janeiro: Sociedade Brasileira de Matematica) pp 65-73
[10] Doshi-Velez F and Kim B 2017 Towards a rigorous science of interpretable machine learning (arXiv:1702.08608)
[11] Edic P M, Isaacson D, Saulnier G J, Jain H and Newell J C 1998 An iterative Newton-Raphson method to solve the inverse admittivity problem IEEE Trans. Biomed. Eng.45 899-908 · doi:10.1109/10.686798
[12] Faddeev L D 2016 Increasing solutions of the Schrödinger equation Fifty Years of Mathematical Physics: Selected Works of Ludwig Faddeev (Singapore: World Scientific) pp 34-6 · Zbl 1347.81016 · doi:10.1142/9789814340960_0003
[13] Greenleaf A, Lassas M, Santacesaria M, Siltanen S and Uhlmann G 2018 Propagation and recovery of singularities in the inverse conductivity problem Anal. PDE11 1901-43 · Zbl 1388.35225 · doi:10.2140/apde.2018.11.1901
[14] Hamilton S J, Hänninen A, Hauptmann A and Kolehmainen V 2019 Beltrami-net: domain-independent deep d-bar learning for absolute imaging with electrical impedance tomography (a-EIT) Physiol. Meas.40 074002 · doi:10.1088/1361-6579/ab21b2
[15] Huhtanen M and Perämäki A 2012 Numerical solution of the R-linear Beltrami equation Math. Comput.81 387-97 · Zbl 1255.65219 · doi:10.1090/s0025-5718-2011-02541-x
[16] Hyvönen N 2004 Complete electrode model of electrical impedance tomography: approximation properties and characterization of inclusions SIAM J. Appl. Math.64 902-31 · Zbl 1059.35168 · doi:10.1137/s0036139903423303
[17] Isaacson D, Mueller J L, Newell J C and Siltanen S 2006 Imaging cardiac activity by the D-bar method for electrical impedance tomography Physiol. Meas.27 S43-50 · doi:10.1088/0967-3334/27/5/s04
[18] Kingma D and Ba J 2014 Adam: a method for stochastic optimization Int. Conf. on Learning Representationsvol 12
[19] Kim K, Lassas M, Mueller J L and Siltanen S 2009 Regularized D-bar method for the inverse conductivity problem Inverse Problems Imaging3 599 · Zbl 1184.35314 · doi:10.3934/ipi.2009.3.599
[20] Lucas A, Iliadis M, Molina R and Katsaggelos A K 2018 Using deep neural networks for inverse problems in imaging: beyond analytical methods IEEE Signal Process. Mag.35 20-36 · doi:10.1109/msp.2017.2760358
[21] Lunz S, Öktem O and Schönlieb C-B 2018 Adversarial regularizers in inverse problems Advances in Neural Information Processing Systems pp 8507-16
[22] Malone E, Jehl M, Arridge S, Betcke T and Holder D 2014 Stroke type differentiation using spectrally constrained multifrequency EIT: evaluation of feasibility in a realistic head model Physiol. Meas.35 1051 · doi:10.1088/0967-3334/35/6/1051
[23] McCann M T, Jin K H and Unser M 2017 Convolutional neural networks for inverse problems in imaging: a review IEEE Signal Process. Mag.34 85-95 · doi:10.1109/msp.2017.2739299
[24] McDermott B, O’Halloran M, Porter E and Santorelli A 2018 Brain haemorrhage detection using a SVM classifier with electrical impedance tomography measurement frames PloS One13 e0200469 · doi:10.1371/journal.pone.0200469
[25] McEwan A, Romsauerova A, Yerworth R, Horesh L, Bayford R and Holder D 2006 Design and calibration of a compact multi-frequency EIT system for acute stroke imaging Physiol. Meas.27 S199 · doi:10.1088/0967-3334/27/5/s17
[26] Mueller J L and Siltanen S 2012 Linear and Nonlinear Inverse Problems with Practical Applications vol 10 (Philadelphia, PA: SIAM) · Zbl 1262.65124 · doi:10.1137/1.9781611972344
[27] Mueller J L, Muller P, Mellenthin M, Murthy R, Capps M, Alsaker M, Deterding R, Sagel S D and DeBoer E 2018 Estimating regions of air trapping from electrical impedance tomography data Physiol. Meas.39 05NT01 · doi:10.1088/1361-6579/aac295
[28] Sacco S, Marini C and Carolei A 2004 Medical treatment of intracerebral hemorrhage Neurol. Sci.25 s6-9 · doi:10.1007/s10072-004-0206-7
[29] Saver J L 2006 Time is brain—quantified Stroke37 263-6 · doi:10.1161/01.str.0000196957.55928.ab
[30] Seo J K, Kim K C, Jargal A, Lee K and Harrach B 2019 A learning-based method for solving ill-posed nonlinear inverse problems: a simulation study of lung EIT SIAM J. Imaging Sci.12 1275-95 · doi:10.1137/18m1222600
[31] Shi X, You F, Xu C, Wang L, Fu F, Liu R and Dong X 2009 Experimental study on early detection of acute cerebral ischemic stroke using electrical impedance tomography method World Congress on Medical Physics and Biomedical Engineering(Munich, Germany, September 7-12, 2009) (Berlin: Springer) pp 510-3
[32] Somersalo E, Cheney M and Isaacson D 1992 Existence and uniqueness for electrode models for electric current computed tomography SIAM J. Appl. Math.52 1023-40 · Zbl 0759.35055 · doi:10.1137/0152060
[33] Sylvester J and Uhlmann G 1987 A global uniqueness theorem for an inverse boundary value problem Ann. Math.125 153-69 · Zbl 0625.35078 · doi:10.2307/1971291
[34] Zhou Z, dos Santos G S, Dowrick T, Avery J, Sun Z, Xu H and Holder D S 2015 Comparison of total variation algorithms for electrical impedance tomography Physiol. Meas.36 1193 · doi:10.1088/0967-3334/36/6/1193
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.