Microwave TomographySystemforMethodical …downloads.hindawi.com/journals/ijap/2019/4074862.pdfIn this paper, we designed and tested a laboratory microwave imaging system suitable

Research ArticleMicrowave Tomography System for Methodical Testing of HumanBrain Stroke Detection Approaches

Ilja Merunka 1 Andrea Massa2 David Vrba 3 Ondrej Fiser3 Marco Salucci2

and Jan Vrba 3

1Department of Electromagnetic Field Czech Technical University in Prague Prague Czech Republic2ELEDIA Research Center (ELEDIAUniTN-University of Trento) Via Sommarive 9 I-38123 Trento Italy3Department of Biomedical Technology Czech Technical University in Prague Prague Czech Republic

Correspondence should be addressed to Ilja Merunka meruniljfelcvutcz

Received 19 October 2018 Revised 16 January 2019 Accepted 23 January 2019 Published 31 March 2019

Guest Editor Sandra Costanzo

Copyright copy 2019 Ilja Merunka et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

In this work a prototype of a laboratory microwave imaging system suitable to methodically test the ability to image detect andclassify human brain strokes using microwave technology is presented It consists of an antenna array holder equipped with tennewly developed slot bowtie antennas a 25 D reconfigurable and replaceable human head phantom stroke phantoms andrelated measuring technology and software This prototype was designed to allow measurement of a complete S-matrix of theantenna array The reconfigurable and replaceable phantom has currently 23 different predefined positions for stroke phantomplacement This setting allows repeated measurements for the stroke phantoms of different types sizesshapes and at differentpositions It is therefore suitable for large-scale measurements with high variability of measured data for stroke detection andclassification based on machine learning methods In order to verify the functionality of the measuring system S-parameterswere measured for a hemorrhagic phantom sequentially placed on 23 different positions and distributions of dielectricparameters were reconstructed using the Gauss-Newton iterative reconstruction algorithm The results correlate well with theactual position of the stroke phantom and its type

1 Introduction

Microwave tomography (MWT) is an emerging imagingtechnique for retrieving the spatial distribution of dielectricparameters of biological tissues [1]

In microwave imaging the imaged area is typicallyexposed to electromagnetic (EM) waves radiated by anten-na(s) positioned around the imaged area The spatial distri-bution of the dielectric parameters in the imaged areainfluences the way of propagation of these waves and theresulting EM field is subsequently measured by the sameantennas The measured data is used to estimate the distribu-tion of dielectric properties within the imaged area

While the use of nonionizing radiation in MWT systemsis a good argument when looking for benefits over a com-puted tomography and mammography (nonionizing

radiation would allow frequent screenings) portability andcost effectiveness are good points when comparing MWTwith magnetic resonance imaging [2]

Although one of the first reference of microwave hemor-rhagic stroke detector can be found in the US patent byinventors Haddad and Trebes [3] there are nowadays severalresearch groups working towards early stroke detection anddifferentiation systems Just to name few leaders of thegroups active in this fieldmdashMikael Persson from MedfieldDiagnostics AB Sweden [4] Serguei Semenov from EMTen-sor GmbH Austria [5] Amin Abbosch from The Universityof Queensland Australia [6] and Lorenzo Crocco from theNational Research Council of Italy Italy [7ndash9] Thorougherreview can be found in [10]

In general two different approaches for detection andclassification of strokes can be found in literature methods

HindawiInternational Journal of Antennas and PropagationVolume 2019 Article ID 4074862 9 pageshttpsdoiorg10115520194074862

based on visualization (image reconstruction) of dielectricparameters distribution in the region of interests [6 11] andmethods based on machine learning algorithms [4 12 13]

Most of the published systems (both hardware andreconstructionclassification algorithms) have not been sys-tematically tested Typically one or two stroke phantompositions are selected and the results obtained are evaluatedand presented

In this paper we designed and tested a laboratorymicrowave imaging system suitable for measuring headand stroke phantoms of various types shapes and sizesThe stroke phantoms could be placed at any of 23 prede-fined positions covering almost all the volume of the headphantom Such a system is especially suited for testingmethods for stroke detection and classification based onmachine learning algorithms but it can be used for testingimage reconstruction methods as well Additionally theproposed system allows testing of the two approaches forphantoms of different complexities and matching liquidsof various dielectric properties The system was tested usinga simple 25 D reconfigurable head phantom with hemor-rhagic stroke phantom which was sequentially placed atsome of the predefined positions The system phantomsand measurements are described in Section 2 Section 3includes an analysis of the measured data as a function ofstroke phantom position Finally a deterministic iterativealgorithm for the reconstruction of the dielectric propertiesdistribution was applied in Section 4 to the measured datain order to verify functionality of the presented system

2 Microwave Imaging System Description

21 Phantoms of Head and Brain Strokes Antenna ArrayThe shape of the head phantom was designed with respectto the main objective of the article namely to systematicallytest the use of microwave imaging for the purpose of detec-tion and classification of brain strokes occurring anywherein the volume of human brain For this it is necessary to beable to place a stroke phantom on a predefined position inthe entire head phantom Towards this end a liquid headphantom has been chosen to fill an elliptical container witha human head-shaped cross section in a transverse planeThe head phantom vessel is placed into the antenna elementholder The space between the vessel and the holder is filledwith a matching liquid see Figure 1 Both the antenna holderand the head phantom vessel have a height of 200mm(including a base of 4mm) and a wall thickness of 2mmThe numerical model of the proposed system and its physicalform are shown in Figure 2 23 octagonal pins are designed atthe bottom of this container and are used to define the posi-tions of phantoms of strokes These latter are made by hollowcylinders filled by liquids of desired dielectric parametershaving an octagonal opening in their bottom This ensuresaccurate and repeatable placement of phantoms of strokesin the head phantom The containers for stroke phantomshave an internal diameter of 40mm a height of 196mmand a wall thickness of 1mm All the parts were made by3D printing using Prusa i3 MK2 (Prusa Research CzechRepublic) and PETG material

22 Antenna Elements Ten identical slot bowtie antennas(antenna design is described in [14]) were placed on the wallsof the antenna array holder The antennas were made of twolayers of 15mm RF substrates Rogers R04003C with thedimensions of 59 times 59 mm2 and realized by Pragoboardsro Prague Czech Republic The main advantages of thisantenna element design are suppressed radiation outsidethe imaging area symmetry of radiated field mechanicalstiffness and easy inexpensive and repeatable productionThe front side of the realized antenna and the modules ofmeasured reflection coefficients of all 10 antenna elementslaid on the surface of a liquid head phantom are illustratedin Figure 3 The reflection coefficients of all ten antennaelements show virtually identical values and a low ampli-tude (less than -10 dB) in the frequency range fromnearly 09 to 13MHz The system introduced in thiswork can therefore be used at any frequency in this fre-quency band The frequency band was chosen in viewof the fact that it fits into frequency bands already con-sidered the most suitable for this application [7] consider-ing attenuation of EM waves inside the human head andresolution of the imaging method Although the authorsbelieve that the higher number of antennas would leadto better results the maximal number of antennas isdetermined by the actual dimensions of the containerand the used antenna elements

23 Measurement Setup and Settings The measuring instru-ments used in the presented MWI system are the microwaveswitching matrix ZN-Z84-B42 (Rohde amp Schwarz Germany)and the vector network analyzer (VNA) ZNB4-B32 (Rohdeamp Schwarz Germany) The antennas were connected to theswitching matrix test ports using semirigid coaxial cables(Rohde amp Schwarz Czech Republic) Although the isolationbetween all tested matrix ports should be greater than90 dB 10 out of 24 ports of the switching matrix test portsinvolved in the measurement were selected for the highestpossible isolation between channels The fact that 90 dB iso-lation should be sufficient is supported by another group thatused an 80dB [4] custom-made switching matrix All high-frequency connections were tightened by a torque wrench09Nm Setting and triggering of measurements togetherwith reading out of the measured data were done withMATLAB scripts written in-house (MathWorks MAUSA) The photograph of the MWI system is shown inFigure 2 The following parameters of the measurementinstruments were used operating frequency 1GHz interme-diate frequency bandwidth 30Hz and output power of theVNA 13dBm The working frequency 1GHz was selectedwith respect to penetration depth spatial resolution and lastbut not least the fact that it fits into the frequency intervalpreferred for the same application in [7]

24 Calibration A full-port calibration was performed justbefore the measurement using the automatic calibration unitZN-Z153 (Rohde amp Schwarz Germany)

Throughout the calibration period the antenna elementswere disconnected from the coaxial cables leading into theswitching matrix To these open ends of the coaxial cables

2 International Journal of Antennas and Propagation

the calibration unit was three times reconnected (due to thelower number of ports of the calibration unit than the num-ber of used antenna elementssystem channels) following

the calibration instructions displayed on the VNA In thisway the switching matrix was included in the full-portVNA calibration

A1

A2

A4A5

A6

A7

A8

A9

A10

P1

P2

P3P4 P5

P6

P7

P8

P9

P10P11P12

P13

P14

P15

P16 P17 P18

P19

P20P21P22

P23-720

-6326

-4144-1252 1855

4748

6627

730

66-27

47-4818-55-12-52

-41-44

-63-26

-2424 327 3126

370

31-263-27-24-24

00

220

19018

0

157

735

40

Head phantomHead phantom vesselMatching liquidAntenna holder

Stroke phantom vesselStroke phantom

A3

-370

Figure 1 Bottom view of measuring container with marked positions of the stroke phantoms Black walls of the antenna holder yellowantennas with ports blue matching liquid white head phantom vessel green liquid phantom of human head and red liquid phantomof stroke at position P7 (all dimensions are in mm)

z (mm)200

150

100

50

0100

500

-50-100 -100

-50 050

100y (mm)

x (mm)

(a) (b)

Figure 2 3D model of measuring container with antennas and cylindrical phantom of stroke marked blue (a) and the photograph of themeasuring setup (b)

3International Journal of Antennas and Propagation

The VNA switching matrix and automatic calibrationunit were turned on 90 minutes before the calibration onsetCalibration data was saved and reloaded before eachmeasurement

25 Liquid Phantom Composition and Dielectric PropertiesTwo liquid phantoms substituting healthy head and brainhemorrhagic stroke tissue were prepared The dielectricproperties of the head phantom are equal to average dielec-tric parameters of a human head (εrHH = 39 8 σHH = 0 94Sm) [15] The dielectric properties of the hemorrhagicstroke phantom are chosen based on the knowledge of thedielectric properties of blood [16] where the contrast tobrain tissues is about 50 - 60 A conservative change atthe level of 30 is chosen here (εrHR = 51 4 σHR = 1 22Sm) as in [5] Dielectric properties of the phantoms weremeasured using DAK (Schmid amp Partner Engineering AGSwitzerland) just before the measurement in theMWI systemprototype Compositions in weight percent and comparisonof target and measured dielectric properties of the phantomsare listed in Table 1 In this table the HEAD marks thecolumn devoted to the phantom of the head and theHEM marks the column devoted to the hemorrhagicstroke phantom The dielectric parameters listed in rowsfollowing the compositions were measured at a frequencyof 1GHz and 25degC before the measurement in the MWIsystem In order to eliminate the reflections from theinterface matching liquid-head the head phantom materialwas also used as the matching liquid between the antennasand the head vessel

26 Measuring Procedure The antenna array holder and thehead phantom vessel were firstly filled with homogeneoushead phantom liquid Special attention was paid to elimina-tion of bubbles which appeared on the surface of antennasduring the filling procedure of the container Finally the con-tainer was enclosed by microwave pyramid absorbers Themeasurement procedure is composed from two main steps(1) measurement of the MWI system filled with the headphantom only and (2) measurement with the head phantomincluding the hemorrhagic stroke phantom The phantom ofstroke was subsequently positioned and measured in all 23positions (Figure 1) Whole S-matrix was measured 10 times

(a)

minus12

minus10

minus8

minus6

minus4

-2

0

997903S 1

1997903 (d

B)06 08 10 12 14

Frequency (GHz)A1A2A3A4

A4A5A6A7

A8A9A10

(b)

Figure 3 Photograph of the antenna element from the side facing to the phantom (a) (dimensions in cm) and measured magnitude of S11for all antennas in frequency range 06 - 14GHz heading to the liquid homogeneous phantom of human head (b) Central operatingfrequency is marked by a dashed line

Table 1 Phantom composition in weight percent target andmeasured dielectric parameters at 1GHz and 25degC immediatelybefore the measurement in the MWI system

HEAD HEM

Isopropyl alcohol (wt) 5048 3400

Deionized water (wt) 4835 6467

NaCl (wt) 116 133

Target ϵr (-) 3960 5148

Measured ϵr (-) 3980 5137

U (k = 2) () 17 21

Target σ (Sm) 094 1220

Measured σ (Sm) 094 1224

U (k = 2) () 27 27

U expanded uncertainty of type B HEAD head phantom HEMhemorrhagic phantom


in series to cover fluctuations that could appear during themeasurement for every position of the stroke One measure-ment of the full S-matrix took about 12 s

3 Measured Data

The response of the system to a presence of the phantomof hemorrhagic stroke at all 23 positions is visualized inFigure 4 It is calculated as the relative difference in themodules (equation (1)) and as the absolute difference inthe arguments (equation (2)) of the S-parameters mea-sured with (SSTROKEmn ) and without (SHOMO

mn ) the strokephantom respectively

Δ Smn =SSTROKEmn minus SHOMO

mn

SHOMOmn

1

ΔangSmn = angSSTROKEmn minusangSHOMOmn 2

The most responding S-parameters to the stroke at thespecific position can be deduced from those images Thecharacteristic patterns in both amplitude and phase imagesindicate the strongest responses in transmission coefficients

4 Application of Image ReconstructionAlgorithm to Measured Data

A reconstruction algorithm based on Gauss-Newton algo-rithm with Tikhonov regularization [17] was used to recon-struct the dielectric image in the central plane of the

antennas from the measured S-parameters This algorithmis based on deterministic optimization procedure assuminglinear behavior of outcome in every iteration when only asmall change in dielectric properties is introduced In orderto reduce the building time of the Jacobian matrix the so-called adjoint method was used [18] The value of theTikhonov parameter was determined using the proceduredescribed in [17]

41 Image Reconstruction Procedure The image reconstruc-tion procedure includes solution of forward problem (com-putation of electromagnetic field distribution and S-matrix)extraction of E-fields and S-matrix computation of Jacobianand computation of step in dielectric properties of thematerial domains To model the propagation of E-field inthe forward step the 3D numerical version of the realmeasuring system was modeled in COMSOL Multiphysics(Figure 2(a)) Models of the real antennas together withSMA ports were used in the numerical model Specialattention was payed to correctly discretize all the model com-ponents Finally EM field was computed using the finite ele-ment method in 3D All three components of E-field wereextracted and used in the inverse step In order to preciselyrepresent the real-world scenario any special assumptioneither simplification in the computational procedure of EMfield was not made Therefore the framework presented inthis paper can be easily adapted to the other realistic scenarioby simply replacing the numerical model Computation ofthe forward problem is the most computationally extensiveprocedure and takes about 150 minutes for all 10 antennaports on the current PC (i7-6700 - 34GHz 64GB DDR4

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

10

20

30

40

50

60

Δ|S m

n| (

)

(a)

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

1020304050607080

S mn (ordm)

(b)

Figure 4 Response of the system ((a) relative change in the modulus of S-matrix and (b) change in the phase of S-matrix in degrees) to apresence of the hemorrhagic stroke phantom at 23 different positions (y axes) in the phantom of human head


RAM - 2133MHz) The reconstruction algorithm was imple-mented in MATLAB

In an attempt to achieve the actual values of complex per-mittivity in the region of interest 10 iterations of the recon-struction algorithm were performed Since the imagereconstruction takes more than 25 hours for 10 iterationsthe maximal number of iterations allowed was set to 10 Bet-ter results could be possibly achieved with higher number ofiterations However as it can be seen from Figure 5 there isonly a small change in the dielectric parameters over the lastfew iterations Results of the reconstruction in a plane goingthrough the center of the antennas for hemorrhagic stroke at

position P7 are shown in Figure 5 Even though values of rel-ative permittivity and conductivity in the reconstructedimages did not reach the actual values and shape of thestroke the biggest change in those quantities can be observedin the area of the stroke phantom (white circle)

Nine different unique positions of the stroke phantominside the head phantom were selected and the measureddata were processed with the image reconstruction algo-rithm In order to show the results of the reconstructionfor all the positions of the stroke in one image 350 valuesof relative permittivity and conductivity differing the mostfrom the background matching liquid were detected in all

40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46

minus80 minus60 minus40 0minus20 20minus94

є (-)

(a)

20 40 60 80 94

x (mm)

minus76minus94 minus80 minus60 minus40 minus20 0

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)

minus94 minus80 minus60 minus40 minus20 20 40 60 80 94x (mm)

32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

minus94 minus80 minus60 minus40 minus20 0 20 40 60 80 94

120590 (S

m)

15

14

13

12

11

1

09

08

(d)

Figure 5 Results (a b) of iterative reconstruction procedure for hemorrhagic stroke phantom at position P7 (marked by white circle) andevolution (c d) of relative permittivity and conductivity over the iterations (i = 1 for the first iteration i = 10 for the tenth iteration) onthe line going through the center of the stroke phantom (marked by dashed white lines) Actual values of both relative permittivity andconductivity are marked by solid black lines in graphs (c) and (d)


80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)

minus100 minus80 minus60 minus40 minus20 0 20 40 60 80 100x (mm)

Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)

Figure 6 Detection of position of the strokes from reconstructed images Circles mark the actual position of the stroke dots of adequate colorshow the detected positions

Table 2 Maximal values of dielectric parameters for the selected positions of the stroke phantom after the reconstruction

Position P1 P3 P7 P8 P11 P18 P19 P22 P23

max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124


9 images and its positions plotted together for every strokephantom position see Figures 6(a) and 6(b) In this figurethe threshold used to allow the binarization is based on theassumption that the size of the stroke phantom is knownAnd so the size of the area covered by the dots of the specificcolor in Figure 6 is equaled to the size of the area of thecross-section of the stroke phantom Positions of the strokephantom can be better recognized from images depictingconductivity Since the quantitative information is lost inFigure 6 the maximal values of dielectric parameters arelisted in Table 2

5 Discussion

The MWI systems are very sensitive to the quality of con-ductedmicrowavemeasurement There aremany cofoundersincluding temperature drift of measuring devices changes incomplex permittivity of phantoms due to the changes in tem-perature evaporation and possible leak of liquids duringstroke phantom replacement procedure appearance of bub-bles in the liquid phantoms and quality of multiport calibra-tion procedure that the experimenter has to deal with duringthe measuring procedure All those sources of noise nega-tively impact the quality of reconstructed images

Conservative estimates of the differences between thedielectric properties of brain tissues and areas affected bystrokes were considered It can be assumed that a higher con-trast in the dielectric properties of the head phantom andstroke phantoms would yield in better reconstruction results

The number of antennaschannels of the presentedsystem was determined by the dimensions of the humanheadhead phantom and the antenna element For the usedantenna elements only limited size reductions can beachieved Metamaterial-based structures such as describedin [19ndash21] could be suitable for systems with a higher num-ber of the antenna elements

The forward problem in every iteration is solved for everyantenna element This can be done in parallel and thusreduce the total computational time Further reduction ofthe computational time could be achieved by using the FiniteDifferences in Time-Domain Method (FDTD) which can beaccelerated using GPU If a broadband formulation of FDTDis used it is possible to obtain results for multiple frequenciesin a single simulation without a significant computationaltime increase

In this work imaging results for only 9 stroke phantompositions are presented although there are in total of 23 posi-tions available This step is only motivated by the clarity ofthe imaging results presented The 9 positions consideredwere carefully selected based on the geometric symmetry ofthe head phantom and antenna array In other words theimaging results for each of the remaining 14 positions wouldnot differ much from the results corresponding to one of the9 positions

6 Conclusion

The prototype of the MWT system allowing systematic eval-uation of human brain stroke detection and classification

approaches has been presented in this paper In order toprove the capabilities of the presented system hemorrhagicphantom of human brain stroke was prepared and the exper-imental measurements were conducted with this strokesequentially placed at 23 different positions in the headphantom Achieved results support the statements ofanother groups working in this field that the detection anddifferentiation of the strokes by means of microwave tech-nique should be possible at least in the laboratory conditionsIn our future work we are going to improve our system inseveral ways to deal with some major known issues such aslow sensitivity spatial resolution and sensitivity to noise A25 D anatomically realistic head phantom consisting ofthree solid layers with shape and dielectric properties corre-sponding to skin skull and cerebrospinal fluid is currentlyunder development We also plan to combine the recon-structed images with detection and classification by algo-rithms of machine learning

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The authors declare that there is no conflict of interestregarding the publication of this paper

Acknowledgments

This work has been supported by a grant from the CzechScience Foundation number 17-00477Y

References

[1] A Fhager P Hashemzadeh and M Persson ldquoReconstructionquality and spectral content of an electromagnetic time-domain inversion algorithmrdquo IEEE Transactions on Biomedi-cal Engineering vol 53 no 8 pp 1594ndash1604 2006

[2] P M Meaney ldquoMicrowave imaging and emerging applica-tionsrdquo International Journal of Biomedical Imaging vol 2012Article ID 252093 2 pages 2012

[3] W Haddad and J Trebes ldquoMicrowave hemorrhagic strokedetectorrdquo US Patent US20 030 018 244 A1 2003 US Classifi-cation 600371 International Classification A61B505 Coop-erative Classification A61B54076 A61B50507 A61B505European Classification A61B505M A61B505 httpswwwgooglechpatentsUS20030018244

[4] M Persson A Fhager H D Trefna et al ldquoMicrowave-basedstroke diagnosis making global prehospital thrombolytic treat-ment possiblerdquo IEEE Transactions on Biomedical Engineeringvol 61 no 11 pp 2806ndash2817 2014

[5] M Hopfer R Planas A Hamidipour T Henriksson andS Semenov ldquoElectromagnetic tomography for detection dif-ferentiation and monitoring of brain stroke a virtual dataand human head phantom studyrdquo IEEE Antennas and Propa-gation Magazine vol 59 no 5 pp 86ndash97 2017

[6] D Ireland K Bialkowski and A Abbosh ldquoMicrowave imag-ing for brain stroke detection using Born iterative methodrdquo


IET Microwaves Antennas amp Propagation vol 7 no 11pp 909ndash915 2013

[7] R Scapaticci L Di Donato I Catapano and L Crocco ldquoAfeasibility study on microwave imaging for brain stroke mon-itoringrdquo Progress In Electromagnetics Research B vol 40pp 305ndash324 2012

[8] R Scapaticci O M Bucci I Catapano and L Crocco ldquoDiffer-ential microwave imaging for brain stroke followuprdquo Interna-tional Journal of Antennas and Propagation vol 2014 ArticleID 312528 11 pages 2014

[9] R Scapaticci J Tobon G Bellizzi F Vipiana andL Crocco ldquoDesign and numerical characterization of a low-complexity microwave device for brain stroke monitoringrdquoIEEE Transactions on Antennas and Propagation vol 66no 12 pp 7328ndash7338 2018

[10] L Crocco I Karanasiou M L James and R C ConceiccedilatildeoEds Emerging Electromagnetic Technologies for Brain DiseasesDiagnostics Monitoring and Therapy Springer InternationalPublishing 2018 httpswwwspringercomlabook9783319750064

[11] A Abbosh ldquoMicrowave systems for head imaging challengesand recent developmentsrdquo in 2013 IEEE MTT-S InternationalMicrowave Workshop Series on RF and Wireless Technologiesfor Biomedical and Healthcare Applications (IMWS-BIO)pp 1ndash3 Singapore December 2013

[12] M Salucci J Vrba I Merunka and A Massa ldquoReal-timebrain stroke detection through a learning-by-examples tech-nique - an experimental assessmentrdquo Microwave and OpticalTechnology Letters vol 59 no 11 pp 2796ndash2799 2017

[13] M Salucci A Gelmini J Vrba I Merunka G Oliveri andP Rocca ldquoInstantaneous brain stroke classification and locali-zation from real scattering datardquoMicrowave and Optical Tech-nology Letters vol 61 no 3 pp 805ndash808 2019

[14] I Merunka J Vrba O Fiser and D Vrba ldquoComparison ofbowtie slot and rectangular waveguide-based antennas formicrowave medical imagingrdquo in 12th European Conferenceon Antennas and Propagation (EuCAP 2018) pp 1ndash5 LondonUK April 2018

[15] IEEE Std 1528-2013 (Revision of IEEE Std 1528-2003) IEEERecommended Practice for Determining the Peak Spatial-Average Specific Absorption Rate (SAR) in the Human Headfrom Wireless Communications Devices Measurement Tech-niques IEEE 2013 httpssearchlibrarywisceducatalog9912350582702121

[16] P A Hasgall F Di Gennaro C Baumgartner et alldquoITrsquoIS database for thermal and electromagnetic parame-ters of biological tissuesrdquo January 2015 httpwwwitisethzchdatabase

[17] N Joachimowicz C Pichot and J P Hugonin ldquoInverse scat-tering an iterative numerical method for electromagneticimagingrdquo IEEE Transactions on Antennas and Propagationvol 39 no 12 pp 1742ndash1753 1991

[18] Q Fang P M Meaney S D Geimer A V Streltsov and K DPaulsen ldquoMicrowave image reconstruction from 3-D fieldscoupled to 2-D parameter estimationrdquo IEEE Transactions onMedical Imaging vol 23 no 4 pp 475ndash484 2004

[19] M Poliacutevka and D Vrba ldquoShielded micro-coplanar CRLH TLzeroth-order resonator antenna critical performance evalua-tionrdquo Radioengineering vol 18 no 1 pp 1592ndash1595 2009

[20] J Vrba and D Vrba ldquoAmicrowave metamaterial inspired sen-sor for non-invasive blood glucose monitoringrdquo Radioengi-neering vol 24 no 4 pp 877ndash884 2015

[21] D Vrba J Vrba D B Rodrigues and P Stauffer ldquoNumericalinvestigation of novel microwave applicators based on zero-order mode resonance for hyperthermia treatment of cancerrdquoJournal of the Franklin Institute vol 354 no 18 pp 8734ndash8746 2017


International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

based on visualization (image reconstruction) of dielectricparameters distribution in the region of interests [6 11] andmethods based on machine learning algorithms [4 12 13]

Most of the published systems (both hardware andreconstructionclassification algorithms) have not been sys-tematically tested Typically one or two stroke phantompositions are selected and the results obtained are evaluatedand presented

In this paper we designed and tested a laboratorymicrowave imaging system suitable for measuring headand stroke phantoms of various types shapes and sizesThe stroke phantoms could be placed at any of 23 prede-fined positions covering almost all the volume of the headphantom Such a system is especially suited for testingmethods for stroke detection and classification based onmachine learning algorithms but it can be used for testingimage reconstruction methods as well Additionally theproposed system allows testing of the two approaches forphantoms of different complexities and matching liquidsof various dielectric properties The system was tested usinga simple 25 D reconfigurable head phantom with hemor-rhagic stroke phantom which was sequentially placed atsome of the predefined positions The system phantomsand measurements are described in Section 2 Section 3includes an analysis of the measured data as a function ofstroke phantom position Finally a deterministic iterativealgorithm for the reconstruction of the dielectric propertiesdistribution was applied in Section 4 to the measured datain order to verify functionality of the presented system

2 Microwave Imaging System Description

21 Phantoms of Head and Brain Strokes Antenna ArrayThe shape of the head phantom was designed with respectto the main objective of the article namely to systematicallytest the use of microwave imaging for the purpose of detec-tion and classification of brain strokes occurring anywherein the volume of human brain For this it is necessary to beable to place a stroke phantom on a predefined position inthe entire head phantom Towards this end a liquid headphantom has been chosen to fill an elliptical container witha human head-shaped cross section in a transverse planeThe head phantom vessel is placed into the antenna elementholder The space between the vessel and the holder is filledwith a matching liquid see Figure 1 Both the antenna holderand the head phantom vessel have a height of 200mm(including a base of 4mm) and a wall thickness of 2mmThe numerical model of the proposed system and its physicalform are shown in Figure 2 23 octagonal pins are designed atthe bottom of this container and are used to define the posi-tions of phantoms of strokes These latter are made by hollowcylinders filled by liquids of desired dielectric parametershaving an octagonal opening in their bottom This ensuresaccurate and repeatable placement of phantoms of strokesin the head phantom The containers for stroke phantomshave an internal diameter of 40mm a height of 196mmand a wall thickness of 1mm All the parts were made by3D printing using Prusa i3 MK2 (Prusa Research CzechRepublic) and PETG material

22 Antenna Elements Ten identical slot bowtie antennas(antenna design is described in [14]) were placed on the wallsof the antenna array holder The antennas were made of twolayers of 15mm RF substrates Rogers R04003C with thedimensions of 59 times 59 mm2 and realized by Pragoboardsro Prague Czech Republic The main advantages of thisantenna element design are suppressed radiation outsidethe imaging area symmetry of radiated field mechanicalstiffness and easy inexpensive and repeatable productionThe front side of the realized antenna and the modules ofmeasured reflection coefficients of all 10 antenna elementslaid on the surface of a liquid head phantom are illustratedin Figure 3 The reflection coefficients of all ten antennaelements show virtually identical values and a low ampli-tude (less than -10 dB) in the frequency range fromnearly 09 to 13MHz The system introduced in thiswork can therefore be used at any frequency in this fre-quency band The frequency band was chosen in viewof the fact that it fits into frequency bands already con-sidered the most suitable for this application [7] consider-ing attenuation of EM waves inside the human head andresolution of the imaging method Although the authorsbelieve that the higher number of antennas would leadto better results the maximal number of antennas isdetermined by the actual dimensions of the containerand the used antenna elements

23 Measurement Setup and Settings The measuring instru-ments used in the presented MWI system are the microwaveswitching matrix ZN-Z84-B42 (Rohde amp Schwarz Germany)and the vector network analyzer (VNA) ZNB4-B32 (Rohdeamp Schwarz Germany) The antennas were connected to theswitching matrix test ports using semirigid coaxial cables(Rohde amp Schwarz Czech Republic) Although the isolationbetween all tested matrix ports should be greater than90 dB 10 out of 24 ports of the switching matrix test portsinvolved in the measurement were selected for the highestpossible isolation between channels The fact that 90 dB iso-lation should be sufficient is supported by another group thatused an 80dB [4] custom-made switching matrix All high-frequency connections were tightened by a torque wrench09Nm Setting and triggering of measurements togetherwith reading out of the measured data were done withMATLAB scripts written in-house (MathWorks MAUSA) The photograph of the MWI system is shown inFigure 2 The following parameters of the measurementinstruments were used operating frequency 1GHz interme-diate frequency bandwidth 30Hz and output power of theVNA 13dBm The working frequency 1GHz was selectedwith respect to penetration depth spatial resolution and lastbut not least the fact that it fits into the frequency intervalpreferred for the same application in [7]

24 Calibration A full-port calibration was performed justbefore the measurement using the automatic calibration unitZN-Z153 (Rohde amp Schwarz Germany)

Throughout the calibration period the antenna elementswere disconnected from the coaxial cables leading into theswitching matrix To these open ends of the coaxial cables




A1

A2

A4A5

A6

A7

A8

A9

A10

P1

P2

P3P4 P5

P6

P7

P8

P9

P10P11P12

P13

P14

P15

P16 P17 P18

P19

P20P21P22

P23-720

-6326

-4144-1252 1855

4748

6627

730

66-27

47-4818-55-12-52

-41-44

-63-26

-2424 327 3126

370

31-263-27-24-24

00

220

19018

0

157

735

40



A3

-370


z (mm)200

150

100

50

0100

500

-50-100 -100

-50 050

100y (mm)

x (mm)

(a) (b)






(a)

minus12

minus10

minus8

minus6

minus4

-2

0

997903S 1

1997903 (d

B)06 08 10 12 14


A4A5A6A7

A8A9A10

(b)



HEAD HEM



NaCl (wt) 116 133

Target ϵr (-) 3960 5148


U (k = 2) () 17 21



U (k = 2) () 27 27




3 Measured Data




mn

SHOMOmn

1







11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

10

20

30

40

50

60

Δ|S m

n| (

)

(a)

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

1020304050607080

S mn (ordm)

(b)







40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46


є (-)

(a)

20 40 60 80 94

x (mm)


minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)


32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10


120590 (S

m)

15

14

13

12

11

1

09

08

(d)



80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




A1

A2

A4A5

A6

A7

A8

A9

A10

P1

P2

P3P4 P5

P6

P7

P8

P9

P10P11P12

P13

P14

P15

P16 P17 P18

P19

P20P21P22

P23-720

-6326

-4144-1252 1855

4748

6627

730

66-27

47-4818-55-12-52

-41-44

-63-26

-2424 327 3126

370

31-263-27-24-24

00

220

19018

0

157

735

40



A3

-370


z (mm)200

150

100

50

0100

500

-50-100 -100

-50 050

100y (mm)

x (mm)

(a) (b)






(a)

minus12

minus10

minus8

minus6

minus4

-2

0

997903S 1

1997903 (d

B)06 08 10 12 14


A4A5A6A7

A8A9A10

(b)



HEAD HEM



NaCl (wt) 116 133

Target ϵr (-) 3960 5148


U (k = 2) () 17 21



U (k = 2) () 27 27




3 Measured Data




mn

SHOMOmn

1







11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

10

20

30

40

50

60

Δ|S m

n| (

)

(a)

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

1020304050607080

S mn (ordm)

(b)







40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46


є (-)

(a)

20 40 60 80 94

x (mm)


minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)


32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10


120590 (S

m)

15

14

13

12

11

1

09

08

(d)



80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





(a)

minus12

minus10

minus8

minus6

minus4

-2

0

997903S 1

1997903 (d

B)06 08 10 12 14


A4A5A6A7

A8A9A10

(b)



HEAD HEM



NaCl (wt) 116 133

Target ϵr (-) 3960 5148


U (k = 2) () 17 21



U (k = 2) () 27 27




3 Measured Data




mn

SHOMOmn

1







11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

10

20

30

40

50

60

Δ|S m

n| (

)

(a)

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

1020304050607080

S mn (ordm)

(b)







40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46


є (-)

(a)

20 40 60 80 94

x (mm)


minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)


32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10


120590 (S

m)

15

14

13

12

11

1

09

08

(d)



80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



3 Measured Data




mn

SHOMOmn

1







11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

10

20

30

40

50

60

Δ|S m

n| (

)

(a)

11 22 33 44 55 66 77 88 99 1010mn

1

5

10

15

20

23

Posit

ion

1020304050607080

S mn (ordm)

(b)







40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46


є (-)

(a)

20 40 60 80 94

x (mm)


minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)


32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10


120590 (S

m)

15

14

13

12

11

1

09

08

(d)



80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






40 60 80 94x (mm)

minus76

minus60

minus40

minus20

0

20

40

60

76

y (m

m)

Permittivity

34

36

38

40

42

44

46


є (-)

(a)

20 40 60 80 94

x (mm)


minus60

minus40

minus20

0

20

40

60

76

y (m

m)

07

08

09

1

11

12

13

120590 (S

m)

Conductivity

(b)


32

34

36

38

40

42

44

46

48

50

52Permittivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10

0

120598

(c)

x (mm)

Conductivity

i = 1i = 2i = 3i = 4i = 5

i = 6i = 7i = 8i = 9i = 10


120590 (S

m)

15

14

13

12

11

1

09

08

(d)



80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Permittivity

(a)

80

60

40

20

0

minus20

minus40

minus60

minus80

y (m

m)


Conductivity

(b)




max ϵr (-) 4813 5134 4630 4760 4487 4471 4471 4523 4448

max σ (Sm) 142 132 139 138 130 123 124 122 124



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



5 Discussion






6 Conclusion



Data Availability




Acknowledgments


References



























RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


Documents

Microwave TomographySystemforMethodical …downloads.hindawi.com/journals/ijap/2019/4074862.pdfIn this paper, we designed and tested a laboratory microwave imaging system suitable