FPGA based electromagnetic tracking system for

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

T

H E image-guided real-tim e (RT) surgical instrument navigation can be utilized for MIS. Instruments such as catheters and needles are target objects of the navigation system s [1]. There are currently four major mod alities of tracking technologies u sed for MIS: mechanic, ultrasonic, optic and electrom agnetic. The biggest advantage of an electrom agnetic tracking system is that the EMTS does not need line-of-sight between the navigation tools and emitters. Therefore, EMTS allow s the position of the surgical instrument to be tracked flexibly even inside the patient’s body [2]. EMTS is able to calculate the three d imensional (3D) position and orientation of an electromagnetic sensor w hich is relative to the generated magnetic field from a field generator [3]. In com puter-assisted surgery, the EMTS is applied to track the positions of the surgical instrument relative to the patient’s body [4]. Before the tracking process begins, the coord inate system of the EMTS has to be registered w ith the coord inate system of the medical im age. During the intervention, the tip of the catheter is tracked relative to the patient’s anatomy [5]. There are two common technologies for electrom agnetic tracking: alternating cu rrent (AC) EMTS and d irect cu rrent (DC) EMTS. This paper focu s on an AC EMTS w ith the FDM technology. The principle of the system is: to generate magnetic fields by su pplying multiple emitting coils w ith signals of different frequencies and simultaneously measure the voltage induced in the sensing coil w hich is w ithin the m agnetic field s, filter out the measured signal at distinct frequencies and com pare the measured voltages after filtering and the simu lated voltages from each other in order to calcu late the P&Os of the sensing coil.

2 METHOD The experim ental setu p of EMTS consists of a PXI system (PX Ie 8133 and PX I 7854R, N ational Instru m ents, USA), a field generator w ith eight em itting coils, a sensing coil (A urora 5DOF Sensor, N orthern Digital, Canad a) w hich is integrated insid e a catheter (Twin-Pass Dual A ccess catheter, Vascu lar Solu tion, USA) and tw o types of am plifiers (LT1210 and LT1168a, Linear Technology, USA). The follow ing figu re illu strates the hard w are of the w hole system .

IJSER

————————————————

 M engfei Li is currently a research assistant in medical system engineering of Otto-von-Guericke University, Germany, E-mail: [email protected]  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.

Fig.1 The exp erim ental setu p of an EMTS containing a sensing coil (1), an am p lifier for the sensing coil (2), the NI PXI system (3), eight am p lifiers for the em itting coils (4) and a field generator w ith eight em itting coils (5).

The FPGA insid e the N I PXI system is the core com ponent for the experim ental setu p of EMTS. FPGA technology is ap plied to replace the trad itional d igital signal p rocessor (DSP) in ord er to increase the system speed by parallel processing. A DSP w orks sequ entially w hile the FPGA is able to execu te m u ltiple processes sim u ltaneou sly w ithou t slow ing d ow n its w orking speed [6]. In this experim ental setu p, the FPGA is u tilized for signal generation, d ata acqu isition and filtering. A sensing coil w ith a d iam eter of 0.5m m and a length of 8m m is u sed to m easu re the voltages ind u ced in the m agnetic field s generated by the field generator. The sensing coil is integrated in the tip of a catheter w hich has a d iam eter of 1.0m m . The am plifiers are u sed to am plify the signals su p plied into the em itting coils and the signals m easu red by the sensing coil.

IJSER © 2013 http://www.ijser.org

2567

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

ISSN 2229-5518

The algorithm of p osition and orientation estim ation of the sensing coil is based on the m agnetic d ipole m od el w hich is d escribed in literatu re [7], [8] and [9]. The follow ing equ ations rep resent the algorithm of P&O calcu lation ap plied for the d eveloped experim ental setu p of EMTS. (1) (2)

(3) (4) Equ ation (1) calcu lates the sim u lated voltage ind u ced in the sensing coil in the m agnetic field that is generated by each em itting coil respectively. In this equ ation, is the nu m ber of em itting coils (1-8), is the cross-sectional area of the sensing coil. The m agnetic field in the cross-sectional area of the sensing coil can be assu m ed to be hom ogeneou s, becau se of the sm all d iam eter of the sensing coil (0.5m m ). The variable is the angu lar frequ ency of the cu rrents w hich are fed into each em itting coil. The vector represents the m agnetic flu x d ensity in the sensing coil w hen the -th em itting coil is generating a m agnetic field . The variable is the norm al vector of the sensing coil. In Equ ation (2) the m agnetic flu x d ensity is calcu lated . In this equ ation, μ is the m agnetic perm eability of the vacu um . Vector stand s for the electrom agnetic d ipole m om ent and is the position vector of d istinct em itting coils in the coord inate system of the EMTS. is the position vector of the sensing coil. The m agnetic d ipole m om ent of the i-th em itting coil can be calcu lated in Equ ation (3). is the nu m ber of tu rns and is the rad iu s of each em itting coil. Within equ ations (1), (2) and (3), the voltages ind u ced in the sensing coil w ill be estim ated basing on the know n d istribu tion of the electrom agnetic field of the em itting coils. The P&O of the sensing coil w ill be estim ated by m inim izing the d ifference betw een the m easu red and estim ated voltages on the sensing coil (4). In the experim ental setu p of EMTS, the FPGA is u tilized to generate signals. The FPGA based d irect d igital synthesis (DDS) technology [10] is applied for the analog signal generation in this system . By DDS, the EMTS is able to generate m u ltiple channels of signals w ith d ifferent characteristics sim u ltaneou sly. There are tw o m ethod s to realize electrom agnetic tracking: tim e d ivision m u ltiplexing (TDM) [11] and frequ ency d ivision m u ltiplexing (FDM) [12]. Both of the m ethod s have been ap plied to the experim ental setu p and the resu lts are com pared . With the m ethod of TDM, the generated sinu soid al signals are sequ entially su p plied to the eight em itting coils. When one of the em itting coils is w orking and the other seven are stopped , the voltages ind u ced in the sensing coil is m easu red instantaneou sly. The chart below show s the w orking flow of the EMTS w ith TDM.

Fig.2 Flow d iagram of voltage measu rem ent w ith TDM

As is show n in Fig. 2, the generated signals are sequ entially ou tpu tted from analog ou tp u t term inals of the FPGA AO1 to AO8. The d igital ou tpu t signals DIO1 to DIO8 are u sed to tu rn on and off the am plifier of the em itting coils. When one of the am plifiers is tu rned on, relatively, one of the em itting coils is fed w ith the sinu soid al cu rrent. On the other hand , w hen all of the am plifiers are tu rned off, there are not any analog ou tpu t signals. With TDM, the speed of catheter tracking is highly d epend ent on the speed of the voltage m easu rem ent. The frequ ency of the signals fed to the em itting coils is equ al to 1 kH z. Tw enty period s of the voltage signals ind u ced in the sensing coil are m easu red w hen one of the eight em itting coil generates a m agnetic field , w hich m eans 160 period s of the signals are m easu red d u ring one com plete voltage m easu rem ent (eight em itting coils). Therefore, the system requ ires 160m s for a w hole voltage m easu rem ent. FDM technology m akes the voltage m easu rem ent eight tim es faster. For each m easu rem ent, instead of 160, only 20 period s of the signals are requ ired .

IJSER

Fig.3 Flow diagram of voltage measurement with FDM IJSER © 2013 http://www.ijser.org

2568


ISSN 2229-5518

Fig. 3 illu strates the w ork flow of the EMTS by ap plying FDM m ethod . After the m easu rem ent begins, the generated signals w ith d ifferent frequ encies for eight em itting coils are parallel fed into the am plifier. Meanw hile, the voltages on the sensing coil are m easu red . The m easu red voltages consisting of the signals of d istin ct frequ encies, are sent to eight band -pass filters to restore the voltages ind u ced in the m agnetic field generated by each em itting coil ind ivid u ally. These filtered voltages are com pared w ith the sim u lated voltages for P&O calcu lation. The FPGA is u tilized to ru n all of these processes in parallel, inclu d ing the eight IIR band -pass filters. Com p ared to sequ ential execu tion of eight band -pass filters by a DSP p rocessor (i.e. by MATLAB in PC), parallel execu tions of FPGA m ake the filtering p rocesses significantly faster. In each step, 2000×8 (eight em itting coils) sam ples of the filtered signals are sent to the host m achine from FPGA period ically by d irect m em ory access (DMA) technology. DMA d ata transfers are accom plished by first-in-first-ou t (FIFO) architectu re. There are tw o FIFOs u sed for the DMA process. One FIFO u ses the block RAM is in the FPGA and the other FIFO is the DMA FIFO in the host m achine. A DMA engine transfers the d ata from the FPGA d evice RAM to the host m achine m em ory au tom atically in the N I PXI system [13]. With the FIFO technology, the d eveloped EMTS is able to w ork in real-tim e.

3 RESULTS

the catheter tip are tracked and visu alized in 3D slicer. The resu lts of the phantom experim ents w ith the TDM system and FDM system are com p ared . Both of the tw o system s are able to track the P&Os of the catheter tip stably. The m ain d ifference betw een the tw o system s is the tracking speed . By ap plying TDM, the EMTS only m easu res the P&Os ap proxim ately 6 tim es per second . H ow ever, b y app lying FDM, the system enables the P&Os of the catheter’s tip with a m easu ring rate of 35 tim es per second . It m eans, w ith this experim ental setu p, the FDM system ru ns app roxim ately 6 tim es faster than the TDM system . The visu alization of the P&Os of the catheter tip relative to the patient’s vascu lar system enables the interventions w ith red u ced d oses of the contrast agent and d ecreased d oses of x-ray rad iation. Before the intervention, the patient w ill be scanned by CT once. After registration, d u ring the entire intervention processes, no m ore CT scans are requ ired . The accu racy of the im age-gu id ed su rgery system is d epend ent on the accu racy of the tracking system [14]. For the accu rate navigation, the electrom agnetic tracking system need s to be calibrated . The algorithm for tracker calibration of the EMTS is d escribed in [9]. An accu racy evalu ation of the experim ental setu p w ith FDM has been perform ed both before and after calibration.

IJSER

In the experim ental setu p of the EMTS, the calcu lated positions and orientations of the sensing coil (in the tip of the catheter) are d escribed as nu m bers. In the interventions, the P&Os of the su rgical instru m ent have to be d irectly visu alized relative to the patient’s anatom y im age. The EMTS w hich has been d eveloped is aim ed to realize com pu ter assisted end ovascu lar interventions. Instead of clinical experim ents, an experim ent w ith an aneu rysm phantom has been perform ed in ord er to evalu ate the app licability of the d eveloped the electrom agnetic tracking system in the clinical cond ition. In this experim ent, the P&Os of the catheter’s tip in the aneu rysm phantom is visu alized by 3D Slicer.

Fig.4 Aneu rysm p hantom and its m od el in 3D Slicer

The aneu rysm phantom is show n in Fig. 4(a). Before the experim ent, a set of the CT scans of the aneu rysm phantom has been transform ed into the su rface-m od el in 3D slicer. The coord inate system of the CT im ages w as registered w ith the coord inate system of the EMTS. Du ring the experim ent, the catheter (w ith the sensing coil in the tip of it) is pu shed in and pu lled ou t in the vascu lar system of the phantom . The position of the sensing coil is presented as a yellow point. Its orientation is illu strated as a blu e line in Fig. 4(b). The P&Os of

Fig.5 The m easu rem ent setup of accu racy evalu ation consists an op tical tracking system (Polaris Spectra, N orthern Digital, Canad a) (1), the tip of the catheter (2), op tical tracker (3), field generator (4) and the Lego Mind storm s robot system on a w ood en fram e (5).

As is seen in Fig. 5, in this m easu rem ent setu p, the catheter tip (2) w hich consists of a sensing coil insid e, is rigid ly fixed at one position and orientation relative to the optical m arker. The robot m oves the optical m arker to 350 d ifferent positions insid e the ROI ( ). For each m ovem ent,


2569


ISSN 2229-5518

the positions of the sensing coil and the positions of the optical m arker are m easu red by EMTS and the optical tracking system respectively. The m ean error of the Polaris Sp ectra optical tracking system is 0.25m m , w hich is m u ch sm aller than the expected m ean error of an electrom agnetic tracking system [9]. Becau se of its high accu racy, the optical system is u tilized as a reference position m easu ring system to evalu ate the accu racy of the EMTS. Assu m ing the position errors of the sensing coil are in x, y and z axis respectively, the total error is calcu lated as

. The follow ing

figu res ind icate the d ifference of the accu racy of the EMTS before and after calibration.

4 DISCUSSION This p aper introd u ces an experim ental setu p of the FPGA based electrom agnetic tracking system . Tw o m ethod s for configu ring the system are introd u ced : tim e d ivision m u ltiplexing and frequ ency d ivision m u ltiplexing. With the sam e experim ental setu p, the FDM system tracks the P&Os of the catheter approxim ately six tim es faster than the TDM system w ith achieving the sam e precision. The EMTS u tilizes a FPGA as the core com ponent to ru n mu ltiple processes in parallel, w hich m akes the system significantly faster than u sing trad itional DSP d evices. In the fu tu re, the experim ents of catheter tracking shou ld be perform ed in real p atients’ anatom y to evalu ate the EMTS for actu al m ed ical applications. The w orking volu m e of the EMTS cou ld be enlarged by increasing the frequ encies and am plitu d es of the signals w hich are su p plied to the emitting coils. Fu rtherm ore, this experim ental setu p of the EMTS is able to be u sed as a testing platform for novel researches in m inim al invasive su rgery.

ACKNOWLEDGMENT The w ork of this p aper is partly fu nd ed by the Germ an Ministry of Ed u cation and Research (BMBF) w ithin the Forschu ngscam pu s STIMULATE u nd er grant nu m ber ‘03FO16102A’.

IJSER REFERENCES

Fig.6 Position errors of EMTS with FDM before calibration

As is show n in Fig. 6, before calibration the m axim u m error of the EMTS is ap proxim ately 16m m and the m ean error for is 3.5m m .This figu re clearly d em onstrates that increasing the d istance of the sensing coil from the center of the field generator d ecreases the accu racy. After calibration, the EMTS has consid erably higher accu racy, w hich is illu strated in the grap h below .

Fig. 7 Position errors of EMTS with FDM after calibration

The resu lts after calibration is show n in Fig. 7. The m axim u m error in position estim ation of the sensing coil red u ces from 16m m to 5m m . Meanw hile, the root m ean squ are (RMS) error is correspond ingly m u ch sm aller, w hich d ecreases from 4.4m m to 1.4m m . The accu racy of the system is com parable to the com mercial electrom agnetic tracking system s, e.g., the N DI AURORA has a positional accu racy (RMS error) of 0.7m m , the accu racy of Ascension m icroBird is 1.4m m and the accu racy of Polhem u s Fastrak is 0.76m m [15].

[1] Perrin, D.P., Vasilyev, V.N ., Novotny, P., Stoll, J., How e, R.D., Dupont, P.E., Salgo, I.S., d el N ido, P.J., ‘‘Image guid ed surgical interventions,’’ Card iac Surgery, 46(9), pp. 730-766. 2009. [2] Schicho, K., Figl, M., Donat, M., Birkfellner, W., Seem ann, R., Wagner, A., Bergmann, H., Ew ers, R., ‘‘Stability of m iniature electromagnetic tracking system s,’’ Physics in Med icine and Biology, 50(9), pp. 2089-2098, 2005. [3] Pom ianow ski S., ‘‘A 3-space electromagnetic tracking d evice --- a useful m ethod in an in vitro stud y,’’ Acta of Bioengineering an d Biom echanics, 3(2), pp. 49-53, 2001. [4] Birkfellner, W., H um m el, J., Wilson, E., Cleary, K., ‘‘Image-guid ed interventions, Tracking d evices,’’ Springer US, 2008. [5] Wood, B.J., Zhang, H ., Durrani, A., Glossop, N ., Ranjan, S., Lind isch, D., Levy, E., Banovac, F., Borgert, J., Krueger, S., Kruecker, J., Viswanathan, A., Cleary, K., ‘‘Navigation w ith electrom agnetic tracking for inter ventional rad iology proced ures: a feasibility stud y,’’ 16(4), pp. 493-505, 2005. [6] Mahm oud, M., Moham ed, A.R., Abd el-Mohaim en S., ‘‘FPGA-based Real-Tim e d igital sim ulation,’’ Pow er System s Transients, pp. 19-23, 2005. [7] Raab, F.H ., Blood, E.B., Steiner, T.O., Jones, H .R., ‘‘Magnetic position and orientation tracking system,’’ Aerospace and Electronic System s Society, 15(5), pp. 709-718, 1979. [8] Plotkin, A., Paperno, E., ‘‘3-D m agnetic tracking of a single subm iniature coil w ith a large 2-D array of uniaxial transm itters,’’ Magnetics, 39(5), pp. 3295-3297, 2003. [9] Bien, T., Rose, G., ‘‘Algorithm for calibration of the electromagnetic tracking system ,’’ Biom ed ical and H ealth Inform atics (BH I), pp. 85-88, 2012. [10] National Instrum ent, ‘‘DDS waveform generation reference d esign for LabVIEW FPGA,’’ w hite paper, 2012. [11] H iggins F.R., ‘‘AC magnetic tracking w ith phase d isam biguation,’’ U.S. Patent, 2009035645 A1, 2009.


2570


ISSN 2229-5518 [12] Govari, A., ‘‘Electromagnetic position single axis system ,’’ U.S. Patent, 6484118 B1, 2002. [13] National Instrum ent, ‘‘Using DMA FIFO to d evelop high -speed d ata acquisition applications for reconfigurable I/ O Devices,’’ white paper, 2012. [14] Kwartow itz, D.M., Rettmann, M.E., H olm es D.R., Robb R.A., ‘‘A novel technique for analysis of accuracy of m agnetic tracking system s used in im age guid ed surgery,’’ Med ical Im aging, 7625(1), pp. 7625-56, 2010. [15] Wilson, E., ‘‘Accuracy analysis of electrom agnetic tracking w ithin m ed ical environm ents,’’ 2006.

IJSER
