International Journal of Scientific & Engineering Research, Volume 4, Issue 9, Ž™Ž–‹Ž›-2013

ISSN 2229-5518

2566

FPGA based electromagnetic tracking system for fast catheter navigation

Mengfei Li, Tomasz Bien, Georg Rose

Abstract an experimental setup of an electromagnetic tracking system (EMTS) has been developed to perform fast catheter navigation for minimally invasive surgery (MIS). The algorithm for the position and orientation (P&O) calculation is implemented in MATLAB while the whole EMTS is programmed and controlled by LabVIEW. The system utilizes a field programmable gate array (FPGA) for signal generation, acquisition and filtering. With the frequency division multiplexing (FDM) and FPGA infinite impulse response (IIR) filter technology, the developed system is able to track P&O of the catheter tip 35 times per second in five degrees of freedom (DOF). A phantom experiment has been performed to evaluate the performance of the EMTS. After calibration, the positional accuracy of the EMTS is 1.4mm inside the region of interest (ROI).

Keywords Catheter Navigation, Electromagnetic Tracking, Filter, FPGA, Frequency Deviation Multiplexing, LabVIEW, Position and

Orientation calculation.

1 INTRODUCTION

-------------------------------------------------- --------------------------------------------------
H E im age-gu id ed real-tim e (RT) su rgical in stru m en t n avigation can be u tilized for MIS. In stru m en ts su ch as cath eters an d n eed les are target objects of th e n avigation
system s [1]. Th ere are cu rren tly fou r m ajor m od alities of trackin g tech n ologies u sed for MIS: m ech an ic, u ltrason ic, op tic an d electrom agn etic. Th e biggest a d van tage of an electrom agn etic trackin g system is th at th e EMTS d oes n ot n eed lin e-of-sigh t betw een th e n av igation tools an d em itters. Th erefore, EMTS allow s th e p osition of th e su rgical in stru m en t to be tracked flexibly even in sid e th e p atien t’s bod y [2].
EMTS is able to calcu late th e th ree d im en sion al (3D) p osition an d orien tation of an electrom agn etic sen sor w h ich is relative to th e gen erated m agn etic field from a field gen erator [3]. In com p u ter-assisted su rgery, th e EMTS is ap p lied to track th e p osition s of th e su rgical in stru m en t relative to th e p atien t’s bod y [4]. Before th e trackin g p rocess begin s, th e coord in ate system of th e EMTS h as to be registered w ith th e coord in ate system of th e m ed ical im age. Du rin g th e in terven tion , th e tip of th e cath eter is tracked relativ e to th e p atien t’s an atom y [5].
Th ere are tw o com m on tech n ologies for electrom agn etic trackin g: altern atin g cu rren t (AC) EMTS an d d irect cu rren t (DC) EMTS. Th is p ap er focu s on an AC EMTS w ith th e FDM tech n ology. Th e p rin cip le of th e system is: to gen erate m agn etic field s by su p p lyin g m u ltip le em ittin g coils w ith sign als of d ifferen t frequ en cies an d sim u ltan eou sly m easu re th e v oltage in d u ced in th e sen sin g coil w h ich is w ith in th e m agn etic field s, filter ou t th e m easu red sign al at d istin ct frequ en cies an d com p are th e m easu red voltages after filterin g an d th e sim u lated voltages from each oth er in ord er to calcu late th e P&Os of th e sen sin g coil.

————————————————

M engfei Li is currently a research assistant in medical system engineering of Otto-von-Guericke University, Germany, E-mail: mengfei.li@ovgu.de

Tomasz Bien is currently a scientific research assistant in medical system engineering of Otto-von-Guericke University, Germany.

Georg Rose is currently the professor in medical system engineering of

Otto-von-Guericke University, Germany.

2 METHOD

Th e exp erim en tal setu p of EMTS con sists of a PXI system (PX Ie 8133 and PX I 7854R , N ation al In stru m en ts, USA), a field gen erator w ith eigh t em ittin g coils, a sen sin g coil (A urora

5DOF Sensor, N orth ern Digital, Can ad a) w h ich is in tegrated


in sid e a cath eter (Twin -Pass Dual A ccess catheter, Vascu lar Solu tion , USA) an d tw o t yp es of am p lifiers (LT 1210 and LT1168a, Lin ear Tech n olog y , USA). Th e follow in g figu re illu strates th e h ard w are of th e w h ole system .

Fig.1 Th e exp erim en tal setu p of an EMTS con tain in g a sen sin g coil (1), an am p lifier for th e sen sin g coil (2), th e N I PXI system (3), eigh t am p lifiers for th e em ittin g coils (4) an d a field gen erator w ith eigh t em ittin g coils (5).

Th e FPGA in sid e th e N I PXI system is th e core com p on en t for th e exp erim en tal setu p of EMTS. FPGA tech n ology is ap p lied to rep lace th e tr ad ition al d igit al sign al p rocessor (DSP) in ord er to in crease th e system sp eed by p ar allel p r ocessin g. A DSP w ork s sequ en tially w h ile th e FPGA is able to execu te m u ltip le p rocesses sim u ltan eou sly w ith ou t slow in g d ow n its w orkin g sp eed [6]. In th is exp erim en tal setu p , th e FPGA is u tilized for sign al gen er ation , d ata acqu isition an d filter in g. A sen sin g coil w ith a d iam eter of 0.5m m an d a len gth of 8m m is
u sed to m easu re th e v oltages in d u ced in th e m agn etic field s gen erated by th e field gen er ator. Th e sen sin g coil is in tegrated in th e tip of a cath eter w h ich h as a d iam eter of 1.0m m . Th e am p lifiers are u sed to am p lify th e sign als su p p lied in t o th e em ittin g coils an d th e sign als m easu red by th e sen sin g coil.

IJSER © 2013

http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 4, Issue 9, September-2013

ISSN 2229-5518

Th e algorith m of p osition an d orien tation estim ation of th e sen sin g coil is based on th e m agn etic d ip ole m od el w h ich is d escribed in literatu re [7], [8] an d [9]. Th e follow in g equ ation s rep resen t th e algorith m of P &O calcu lation ap p lied for th e d ev elop ed exp erim en tal setu p of EM TS.
(1) (2)
(3)

2567

( 4 )


Equ ation (1) calcu lates th e sim u lated v oltage in d u ced in th e sen sin g coil in th e m agn etic field th at is gen erated by each em ittin g coil resp ectiv ely. In th is equ ation , is th e n u m ber of em ittin g coils (1-8), is th e cross-section al area of t h e sen sin g coil. Th e m agn etic field in th e cross -section al ar ea of th e sen sin g coil can be assu m ed to be h om ogen eou s, becau se of th e sm all d iam eter of th e sen sin g coil (0.5m m ). Th e v ariable is th e an gu lar frequ en cy of th e cu rren ts w h ich are fed in to each em ittin g coil. Th e v ector rep resen ts th e m agn etic flu x d en sity in th e sen sin g coil w h en th e -th em ittin g coil is gen eratin g a m agn etic field . Th e v ariable is th e n orm al v ector of th e sen sin g coil. In Equ ation (2) th e m agn etic flu x d en sity is calcu lated . In th is equ ation , μ is th e m ag n etic p erm eability of th e v acu u m . Vector stan d s for th e electrom agn etic d ip ole m om en t an d is th e p osition v ector of d istin ct em ittin g coils in th e coord in ate system of th e EMTS. is th e p osition v ector of th e sen sin g coil. Th e m agn etic d ip ole m om en t of th e i-th em ittin g coil can be calcu lat ed in Equ ation (3). is th e n u m ber of tu rn s an d is th e rad iu s of each em ittin g coil. With in equ ation s (1), (2) an d (3), th e v oltages in d u ced in th e sen sin g coil w ill be estim ated basin g on th e kn ow n d istrib u tion of th e electrom ag n etic field of th e em ittin g coils. Th e P&O of th e sen sin g coil w ill be estim ated by m in im izin g th e d ifferen ce betw een th e m easu red an d estim ated v olt ages on th e sen sin g coil (4). In th e exp erim en tal setu p of EMTS, th e FPGA is u tilized to gen erate sign als. Th e FPGA based d irect d igital syn th esis (DDS) tech n ology [10] is ap p lied for th e an alog sign al gen erat ion in th is sy stem . By DDS, th e EMTS is able to gen erate m u ltip le ch an n els of sign als w ith d ifferen t ch ar acteristics sim u ltan eou sly.
Th ere are tw o m eth od s to realize electrom a gn etic trackin g: tim e d iv ision m u ltip lexin g (TDM) [11] an d frequ en cy d iv ision m u ltip lexin g (FDM) [12]. Both of th e m eth od s h av e been ap p lied to th e exp erim en tal setu p an d th e resu lts are com p ared . With th e m eth od of TDM, th e gen erated sin u soid al sign als ar e sequ en tially su p p lied to th e eigh t em ittin g coils. Wh en on e of th e em ittin g coils is w orkin g an d th e oth er sev en are stop p ed , th e v oltages in d u ced in th e sen sin g coil is m easu red in st an tan eou sly. Th e ch art below sh ow s th e w orkin g flow of th e EMTS w ith TD M.

Fig.2 Flow d iagram of voltage m easu rem en t w ith TDM

As is sh ow n in Fig. 2, th e gen erated sign als are sequ en tially ou tp u tted from an alog ou tp u t term in als of th e FPGA AO1 to AO8. Th e d igit al ou tp u t sig n als DIO1 to DIO8 are u sed to tu rn on an d off th e am p lifier of t h e em ittin g coils. Wh en on e of th e am p lifiers is tu rn ed on , relativ ely, on e of th e em ittin g coils is fed w ith th e sin u soid al cu rren t. On th e oth er h an d , w h en all of th e am p lifier s are tu rn ed off, th ere are n ot an y an alog ou tp u t sign als. With TD M, th e sp eed of cath eter tr ackin g is h igh ly d ep en d en t on th e sp eed of th e v oltage m easu rem en t . Th e frequ en cy of th e sign als fed to th e em ittin g coils is equ al to 1 kH z. Tw en ty p eriod s of th e v oltage sign als in d u ced in th e sen sin g coil are m easu red w h en on e of th e eigh t em ittin g coil gen erates a m agn etic field , w h ich m ean s 160 p eriod s of th e sign als are m easu red d u rin g on e com p lete v oltage m easu rem en t (eigh t em ittin g coils). Th erefore, th e system requ ires 160m s for a w h ole v oltage m easu rem en t.
FDM tech n ology m akes th e v oltage m easu rem en t eigh t tim es faster. For each m easu rem en t, in stead of 160, on ly 20 p eriod s of th e sign als are req u ired .

Fig.3 Flow diagram of voltage measurement with FDM

IJSER © 2013

http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 4, Issue 9, September-2013

ISSN 2229-5518

2568

Fig. 3 illu strates th e w ork flow of th e EMTS by ap p lyin g FDM m eth od . After th e m easu rem en t begin s, th e gen erated sign als w ith d ifferen t frequ en cies for eigh t em ittin g coils are p ar allel fed in to th e am p lifier. Mean w h ile, th e v oltages on th e sen sin g coil are m easu red . Th e m easu red v oltages con sistin g of th e sign als of d istin ct frequ en cies, are sen t to eigh t ban d -p ass filters to restor e th e v oltages in d u ced in th e m agn etic field gen er ated by each em ittin g coil in d iv id u ally. Th ese filtered v olt ages are com p ared w ith th e sim u lated v oltages for P&O calcu lation . Th e FPGA is u tiliz ed to ru n all of th ese p rocesses in p arallel, in clu d in g th e eigh t IIR ban d -p ass filter s. Com p ared to sequ en tial execu tion of eigh t ban d -p ass filters by a DSP p rocessor (i.e. by MATLAB in PC), p ar allel execu tion s of FP GA m ake th e filterin g p rocesses sign ifican tly faster.
In each step , 2000×8 (eigh t em ittin g coils) sam p les of th e filtered sign als are sen t to th e h ost m ach in e from FPGA p eriod ically by d irect m em ory access (DM A) tech n ology. DMA d ata tran sfer s are accom p lish ed by fir st -in -first-ou t (FIFO ) arch itectu re. Th ere are tw o FIFO s u sed for th e DMA p rocess. On e FIFO u ses th e block RAM is in th e FPGA an d th e oth er FIFO is th e DMA FIFO in th e h ost m ach in e. A DMA en gin e tran sfers th e d at a fr om th e FPGA d ev ice RAM to th e h ost m ach in e m em ory au tom atically in t h e N I PXI system [13]. With th e FIFO tech n ology, th e d ev elop ed EM TS is able to w ork in real-tim e.

3 RESULTS

In th e exp erim en tal setu p of th e EMTS, th e calcu lated p osition s an d orien tation s of th e sen sin g coil (in th e tip of th e cath eter) are d escribed as n u m bers. In th e in terv en tion s, th e P&Os of th e su rgical in stru m en t h av e to be d irectly v isu alized relativ e to th e p atien t’s an at om y im age. Th e EMTS w h ich h as been d ev elop ed is aim ed to realize com p u ter assisted en d ov ascu lar in terv en tion s. In stead of clin ical exp erim en ts, an exp erim en t w ith an an eu ry sm p h an tom h as been p er for m ed in ord er to ev alu ate th e ap p licability of th e d ev elop ed th e electrom agn etic trackin g sy stem in th e clin ical con d ition . In th is exp erim en t, th e P&Os of th e cath eter’s tip in th e an eu rysm p h an tom is v isu alized by 3D Slicer.

Fig.4 An eu rysm p h an tom an d its m od el in 3D Slicer

Th e an eu rysm p h an tom is sh ow n in Fig. 4(a). Before t h e exp erim en t, a set of th e CT scan s of th e an eu rysm p h an tom h as been tr an sform ed in to t h e su rface -m od el in 3D slicer. Th e coord in ate sy stem of th e CT im ages w as registered w ith th e coord in ate sy stem of th e EMTS. Du rin g th e exp erim en t, th e cath eter (w ith th e sen sin g coil in th e tip of it) is p u sh ed in an d p u lled ou t in th e v ascu lar sy stem of th e p h an tom . Th e p ositi on of th e sen sin g coil is p r esen ted as a yellow p oin t. Its orien tation is illu str ated as a blu e lin e in Fig. 4(b). Th e P &Os of
th e cath eter tip are tracked an d v isu alized in 3D slicer.
Th e resu lts of th e p h an tom exp erim en ts w ith th e TDM system an d FDM sy st em are com p ared . Both of th e tw o system s are able to track th e P&Os of th e cath eter tip stably. Th e m ain d ifferen ce betw een th e tw o system s is th e trackin g sp eed . By ap p lyin g TDM, th e EMTS on ly m easu res th e P&Os ap p roxim ately 6 tim es p er secon d . H ow ev er, b y ap p lyin g FDM, th e system en ables th e P&Os of th e cath eter’s tip w ith a m easu rin g rate of 35 tim es p er secon d . It m ean s, w ith th is exp erim en tal setu p , th e FD M system ru n s ap p roxim at ely 6 tim es faster th an th e TDM sy stem .
Th e v isu alization of th e P&O s of t h e cath eter tip relativ e to th e p atien t’s v ascu lar sy stem en ables th e in terv en tion s w ith red u ced d oses of th e con trast agen t an d d ecreased d oses of x-ray r ad iation . Before th e in terv en tion , th e p atien t w ill be scan n ed by CT on ce. After registration , d u rin g th e en tire in terv en tion p rocesses, n o m ore CT scan s are requ ired .
Th e accu racy of th e im age-gu id ed su rgery sy stem is d ep en d en t on th e accu racy of th e trackin g system [14]. For th e accu rate n av igation , th e electrom agn etic track in g sy stem n eed s to be calibrated . Th e algorith m for tracker calibr ation of th e EMTS is d escribed in [9]. An accu racy ev alu ation of th e exp erim en tal setu p w ith FDM h as been p er form ed both before an d after calibr ation .

Fig.5 Th e m easu rem en t setu p of accu racy evalu ation con sists an op tical trackin g system (Polaris Spectra, N orth ern Digital, Can ad a) (1), th e tip of th e cath eter (2), op tical tracker (3), field gen erator (4) an d th e Lego Min d storm s robot system on a w ood en fram e (5).


As is seen in Fig. 5, in th is m easu rem en t setu p , th e cath eter tip (2) w h ich con sists of a sen sin g coil in sid e, is rigid ly fixed at on e p osition an d orien t ation relativ e to th e op tical m arker . Th e robot m ov es th e op tical m arker to 350 d ifferen t p osition s in sid e th e ROI ( ). For each m ov em en t,

IJSER © 2013

http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 4, Issue 9, September-2013

ISSN 2229-5518

2569


th e p osition s of th e sen sin g coil an d th e p osition s of th e op tical m arker are m easu red by EMTS an d th e op tical trackin g system resp ectiv ely. Th e m ean error of th e Polaris Sp ectra op tical tr ackin g sy stem is 0.25m m , w h ich is m u ch sm aller th an th e exp ected m ean error of an electrom agn etic trackin g system [9]. Becau se of its h igh accu r acy, th e op tical sy stem is u tilized as a referen ce p osition m easu rin g sy stem to ev alu ate th e accu racy of th e EMTS. Assu m in g th e p osition errors of th e sen sin g coil are in x, y an d z axis resp ectiv ely, th e
total error is calcu lated as . Th e follow in g figu res in d icate th e d ifferen ce of th e accu racy of th e EMTS
before an d after calibration .

Fig.6 Position errors of EMTS w ith FDM before calibration

As is sh ow n in Fig. 6, before calibr ation th e m axim u m error of th e EMTS is ap p roxim ately 16m m an d th e m ean error for is 3.5m m .Th is figu re clearly d em on str ates th at in creasin g th e d istan ce of th e sen sin g coil from th e cen ter of th e field gen erator d ecreases th e accu racy. After calibr ation , th e EMTS h as con sid er ably h igh er accu racy, w h ich is illu str ated in th e grap h below .

Fig. 7 Position errors of EMTS w ith FDM after calibration

Th e resu lts after calibr ation is sh ow n in Fig. 7. Th e m axim u m error in p osition estim at ion of th e sen sin g coil red u ces from 16m m to 5m m . Mean w h ile, th e root m ean squ are (RMS) error is cor resp on d in gly m u ch sm aller, w h ich d ecreases from 4.4m m to 1.4m m . Th e accu racy of th e sy st em is com p arable to th e com m ercial electrom agn etic trackin g system s, e.g., th e N DI AURORA h as a p osition al accu racy (RMS err or) of 0.7m m , th e accu racy of Ascen sion m icroBird is
1.4m m an d th e accu racy of P olh em u s Fast rak is 0.76m m [15].

4 DISCUSSION

Th is p ap er in trod u ces an exp erim en tal setu p of th e FPGA based electr om agn etic tr ackin g system . Tw o m eth od s for con figu rin g th e system are in trod u ced : tim e d iv ision m u ltip lexin g an d frequ en cy d iv ision m u ltip lexin g. Wit h th e sam e exp erim en tal setu p , th e FDM system tr ack s th e P&Os of th e cath eter ap p roxim ately six tim es faster th an th e TDM system w ith ach iev in g th e sam e p recision . Th e EMTS u tilizes a FPGA as th e core com p on en t to ru n m u ltip le p rocesses in p ar allel, w h ich m akes th e system sign ifican tly faster th an u sin g trad ition al DSP d ev ices. In th e fu tu re, th e exp erim en ts of cath eter trackin g sh ou ld be p erform ed in real p atien ts’ an atom y to ev alu ate th e EMTS for actu al m ed ical ap p lication s. Th e w orkin g v olu m e of th e EMTS cou ld be en larged by in creasin g th e frequ en cies an d am p litu d es of th e sign als w h ich are su p p lied to th e em ittin g coils. Fu rth erm ore, th is exp erim en tal setu p of th e EMTS is able to be u sed as a testin g p latform for n ov el research es in m in im al in v asiv e su rger y.

ACKNOWLEDGMENT

Th e w ork of th is p ap er is p artly fu n d ed by th e Germ an Min istry of Ed u cation an d Research (BMBF) w ith in th e Forsch u n g scam p u s STIM ULATE u n d er gran t n u m ber
‘03FO 16102A’.

REFERENCES

[1] Perrin , D.P., Vasilyev , V.N ., N ovotn y, P., Stoll, J., H ow e, R.D., Du p on t, P.E., Salgo, I.S., d el N id o, P.J., ‘‘Im age gu id ed su rgical in terv en tion s,’’ Card iac Su rger y, 46(9), p p . 730-766. 2009.

[2] Sch ich o, K., Figl, M., Don at, M., Birkfelln er, W., Seem an n , R., Wagn er, A., Bergm an n , H ., Ew ers, R., ‘‘Stability of m in iatu re electrom agn etic trackin g system s,’’ Ph ysics in Med icin e an d Biology, 50(9), p p . 2089-2098,

2005.

[3] Pom ian ow ski S., ‘‘A 3-sp ace electrom agn etic trackin g d ev ice -- a u sefu l m eth od in an in v itro stu d y,’’ Acta of Bioen gin eerin g an d Biom ech an ics,

3(2), p p . 49-53, 2001.

[4] Birkfelln er, W., H u m m el, J., Wilson , E., Cleary, K., ‘‘Im age-gu id ed in terv en tion s, Trackin g d ev ices,’’ Sp rin ger US, 2008.

[5] Wood , B.J., Zh an g, H ., Du rran i, A., Glossop , N ., Ran jan , S., Lin d isch , D.,

Lev y, E., Ban ovac, F., Borgert, J., Kru eger, S., Kru ecker, J., Visw an ath an , A., Cleary, K., ‘‘N av igation w ith electrom agn etic trackin g for in ter v en tion al rad iology p roced u res: a feasibility stu d y,’’ 16(4), p p . 493-505, 2005.

[6] Mah m ou d , M., Moh am ed , A.R., Abd el-Moh aim en S., ‘‘FPGA-based Real-Tim e d igital sim u lation ,’’ Pow er System s Tran sien ts, p p . 19-23, 2005. [7] Raab, F.H ., Blood , E.B., Stein er, T.O., Jon es, H .R., ‘‘Magn etic p osition an d orien tation trackin g system ,’’ Aerosp ace an d Electron ic Syst em s Society, 15(5), p p . 709-718, 1979.

[8] Plotkin , A., Pap ern o, E., ‘‘3-D m agn etic trackin g of a sin gle su bm in iatu re coil w ith a large 2-D array of u n iaxial tran sm itters,’’ Magn etics, 39(5), p p . 3295-3297, 2003.

[9] Bien , T., Rose, G., ‘‘Algorith m for calibration of th e electrom agn etic

trackin g system ,’’ Biom ed ical an d H ealth In form atics (BH I), p p . 85-88,

2012.

[10] N ation al In stru m en t, ‘‘DDS w av eform gen eration referen ce d esign for

La bVIEW FPGA,’’ w h ite p ap er, 2012.

[11] H iggin s F.R., ‘‘AC m agn etic trackin g w ith p h ase d isam bigu ation ,’’ U.S. Paten t, 2009035645 A1, 2009.

IJSER © 2013

http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 4, Issue 9, September-2013

ISSN 2229-5518

[12] Govari, A., ‘‘Electrom agn etic p osition sin gle axis system ,’’ U.S. Paten t,

6484118 B1, 2002.

[13] N ation al In stru m en t, ‘‘Usin g DMA FIFO to d ev elop h igh -sp eed d ata acq u isition ap p lication s for recon figu rable I/ O Dev ices ,’’ w h ite p ap er,

2012.

[14] Kw artow itz, D.M., Rettm an n , M.E., H olm es D.R., Robb R.A., ‘‘A n ov el tech n iq u e for an alysis of accu racy of m ag n etic trackin g system s u sed in im age gu id ed su rgery,’’ Med ica l Im agin g, 7625(1), p p . 7625-56, 2010.

[15] Wilson , E., ‘‘Accu racy an alysis of electrom agn etic trackin g w ith in

m ed ical en v iron m en ts,’’ 2006.

2570

IJSER © 2013

http://www.ijser.org