Wireless Resonant Circuits Printed Using Aerosol Jet Deposition for MRI Catheter Tracking

Abstract

Interventional magnetic resonance imaging could allow for diagnosis and immediate treatment of ischemic stroke; however, such endovascular catheter-based procedures under MRI guidance are inherently difficult. One major challenge is tracking the tip of the catheter, as standard fabrication methods for building inductively coupled coil markers are rigid and bulky. Here, we report a new approach that uses aerosol jet deposition to 3D print an inductively coupled RF coil marker on a polymer catheter. Our approach enables lightweight, conforming markers on polymer catheters and these low-profile markers allow the catheter to be more safely navigated in small caliber vessels. Prototype markers with an inductor with the geometry of a double helix are incorporated on catheters for in vitro studies, and we show that these markers exhibit good signal amplification. We report temperature measurements, and finally demonstrate feasibility in a preliminary in vivo experiment. We provide material properties and electromagnetic simulation performance analysis. This work presents fully aerosol jet-deposited and functional wireless resonant markers on polymer catheters for use in 3T clinical scanners.

I. Introduction

STROKE is a leading cause of long-term disability and the fifth leading cause of death, with approximately 690,000 people suffering from an acute ischemic stroke each year in the United States [1]. While catheter-based endovascular embolectomy under real-time X-ray angiographic guidance is an effective treatment for patients suffering a large vessel occlusion acute ischemic stroke, about 10% or fewer of potentially eligible patients receive such treatments [2]. This is in part due to the narrow recommended time window for therapy within 3–6 hours of stroke onset; however, there is a growing paradigm shift that stroke treatment could potentially follow an expanded time window, if brain tissue viability was known [3]. Recent clinical trials based on perfusion imaging are now allowing stroke patients with viable tissue to be treated up to 24 hours after the onset of symptoms in some cases [4], [5]. Interventional magnetic resonance imaging (MRI) could allow for accurate diagnosis and immediate treatment, since diffusion weighted imaging in combination with perfusion imaging is the gold standard in determining if brain tissue is still viable. However, the magnetic fields in the MRI scanner create a challenging environment for the standard guidewires and catheters used for clot removal, which contain metal components to achieve optimal mechanical properties. During an MRI scan, long conductive structures can give rise to large electric fields at the tip of the device, which can cause unsafe tissue heating [6]. Furthermore, tracking markers are needed to localize the tip of the catheter and must have both a low side profile and be flexible enough to navigate safely in small vessels. If there was a marker that allowed for reliable, safe, and straightforward tracking and visualization of endovascular catheters, this could overcome one of the barriers to clinical adoption by interventional radiologists to use MRI in combination with embolectomy treatment [7].

A. Tracking Markers

Methods for MRI markers have been broadly categorized in the interventional MRI literature as passive or active, or as a hybrid technique, sometimes referred to as semi-active. Passive approaches display the catheter in the image, often using signal void or differences in magnetic susceptibility between the catheter and surroundings, causing local B₀ field inhomogeneities [8]. Active approaches typically use a micro-coil at the tip of catheter, and have wires running down the length of the catheter to receive information about the location of the catheter tip [8]. The hybrid technique has referred to resonant markers, implemented as an inductively coupled coil, or more specifically, a circuit with a capacitor (C) connected to a coil, which has inductance (L) and intrinsic resistance (R). This RLC circuit can be built on the tip of the catheter to provide local signal amplification of the B₁+ field [9]. The circuit inductively couples with the magnetic flux from the radiofrequency (RF) pulse, resulting in induced current in the circuit, and the production of a local magnetic field [10]. The locally amplified B₁+ field produces higher flip angles in the surrounding region, which in turn can enhance signal or alter contrast in MR images. One of the earliest implementations of an inductively coupled coil marker used a solenoid inductor coil geometry surrounding a small solution, which was used as a fiducial marker for localization [9].

Such wireless resonant circuits [11], or inductively coupled RF coils [12], have previously been built on top of large diameter guide catheters using a variety of fabrication methods and geometries. More recently, wireless resonant circuits have been built on catheters for interventional MRI applications, using inductor coil geometries such as a rectangular surface loop coil [13], two opposing solenoids oriented at 45° relative to the axis of the catheter [14], and flat spiral coils [15]. Recent work proposed an inductor geometry of a double helix, each arm tilted 45° with respect to the z-axis [16], [17].

Often, these resonant markers have been built by hand using discrete elements, which introduces a number of challenges, including reproducibility, flexibility, and increased profile thickness of the catheter. In particular, the increased size and decreased flexibility of such a circuit at the catheter tip makes it impractical for neurointerventional use. A number of microfabrication methods for building markers on catheters have been explored beyond manual wire winding, including flexible printed circuit boards, hot embossing, thin-film techniques, and aerosol deposition processes [10]. While thin-film technology can deposit micrometer scale features, it requires a significant investment for each new application, and is a 2-dimensional process, traditionally requiring a rigid, planar framework, and is not ideal for manufacturing 3-dimensional (3D) conformal electronics, such as those required for catheters [15]. Aerosol jet deposition, a new additive manufacturing process, is ideal for building micrometer size electronics, is CAD-file driven [18], and can print conformal electronics on various surfaces. We use aerosol jet deposition to print a complete, distributed LC circuit, using the double helix inductor geometry on a polymer catheter for interventional MRI use at 3.0T.

The wireless resonant circuit can be modeled as an LC circuit, where the inductor is connected in parallel to the capacitor [19]. A current is induced in the circuit, due to inductive coupling with both the RF transmit coil and the local MR signal in the immediate vicinity. This generates an associated magnetic field and flux, enhancing the local B₁. There is local amplification of the flip angle achieved during transmit, and then enhanced local sensitivity during receive. During receive, the marker is loosely coupled to the receive coil, so the B₁ of each combine to enhance the local sensitivity at the marker. The coupling is also dependent on the orientation of the marker with respect to the MR signal phase [19]. The magnetic flux through a surface is calculated by the inner product of the surface normal vector and the magnetic field. Given the double helix design, the surface normal vector cannot be uniquely defined for the whole coil as a single analytical quantity. The current in the inductor is amplified depending on the quality factor, Q, of the resonant circuit. A long capacitor with parallel plates running side-by-side can be effectively modeled as a twin-lead transmission line, which has a distributed capacitance and inductance.

II. Methods

A. Catheter fabrication

Polymer catheters were constructed in a catheter fabrication facility (Penumbra, Alameda, CA). The catheters consisted of a polytetrafluoroethylene (PTFE) catheter substrate, braided with polyether ether ketone (PEEK) fibers, and sealed with a thermoplastic elastomer, polyether block amide (Pebax 72D, Zeus Inc. Orangeburg, SC). The polymer catheter had an outer diameter of approximately 2.248 mm (Table I).

TABLE I.

DIMENSIONS OF THE POLYMER CATHETER AND THE MATERIAL PROPERTIES OF THE INKS ARE REPORTED FOR THE AEROSOL JET-DEPOSITED MARKERS

Material	Dimensions
Catheter
Inner diameter of Polytetrafluoroethylene (PTFE) liner	1.770–1.783 mm
PTFE liner wall thickness	0.019–0.025 mm
Polyether ether ketone (PEEK) fiber diameter	0.127 mm
Polyether block amide wall thickness (Pebax® 72D)	0.076 mm
Catheter outer diameter	2.248 mm (6.75 Fr)
Overall catheter length	140 cm
Silver nano-flake ink (NovaCentrix HPS 108AE1)
Particle Size	400–700 nm
Solid content	60% wt
Sintering condition	60 mins or longer at ≥120°C
Resistivity	484 μΩ-cm (140°C cure temp)
Printing Nozzle diameter	200 μm
Polyimide ink (Sigma Aldrich)
Solid content	11%
Sintering Condition	30 minutes between steps at
	≥120°C, 2 hours for final cure
Dieletric Constant	3.6–3.7
Printing Nozzle diameter	300 μm

B. Resonant marker microfabrication

All conductive prints used a water-based silver nanoparticle flake ink (NovaCentrix HPS-108AE1, NovaCentrix, Austin, TX) and the dielectric ink was polyimide ink (ε_r = 3.6–3.7) (Sigma Aldrich, St. Louis, MO), with the material properties given in Table I. The double helix included 10 turns, tilted at a helix angle of 22.5°, for a 3.53 mm pitch (πr), for a total length of 35.31 mm (10πr). The conductive traces were printed with approximately a 250 μm trace width and a 4 μm thickness. The capacitor plates were printed measuring 180 mm long, with a 280–300 μm width using an approximately 50 μm printed trace width, a 4 μm trace thickness, and 200 μm separation of the plates (Fig. 1). An aerosol jet deposition system (Aerosol Jet® 300P system, Optomec, Albuquerque, NM) was used for printing (Quest Integrated LLC, Kent, WA), with a tolerance of ±60 μm (Fig. 2a). First, the single helical arm and capacitor plates were printed with the silver ink and cured for 1 hour. Second, the dielectric ink for the capacitor and intermediate encapsulation for the first coil were printed and cured for 2 hours. The first printed coil was left exposed at the tip for electrical connection. Third, the top helical arm was printed with the silver ink and cured for 30–60 minutes. Finally, the dielectric ink, was printed to insulate the entire double helix coil and capacitor assembly, and the completed print was cured for 2 hours.

Fig. 1.

Schematic of the resonant marker design: an inductively coupled RF coil marker can be used for tracking during MRI. The side profile of a single helical arm is shown, as a single arm was printed first. The top-down view of the 10-turn double helix geometry of the inductor, along with the twin-lead transmission line capacitor, is shown with dielectric ink (red) and conductive ink (blue) traces. Solid lines indicate traces in front of the catheter, and dashed lines indicate traces behind the catheter (green).

Fig. 2.

Fabrication method and realization example of aerosol jet deposited circuits. a) A simplified schematic of aerosol jet deposition is shown. Nitrogen gas is used for pneumatic aerosolization and propulsion with excess gas being stripped off by a virtual impactor and outer sheath gas collimates the beam and keeps the aerosol from contacting the nozzle. b) An example of a test printing process is shown. c) A realization example of the 10-turn double helix prints are shown. d) This marker print is shown in comparison with a hand-wound marker, demonstrating its extremely low profile.

C. In vitro image acquisition

The catheter was placed in a rectangular water phantom doped with CuSO₄ and the phantom was oriented either parallel or perpendicular to B₀. Images were acquired in both orientations at 3.0T (Discovery MR 750w, GE Healthcare, Chicago, IL) using the body coil for transmit (B₁+) and an 8-channel cardiac receive-only coil. A balanced steady state free precession (bSSFP) sequence was acquired for visualization, and a Bloch-Siegert B₁+ map [20], [21] was acquired for flip angle quantification, using a coronal scan plane, a 10 mm slice thickness, and the imaging parameters in Table II. An ROI was manually drawn around the entire marker to measure the mean signal, and then the ratio with the nearby background water signal was calculated to measure the relative amplification factor. All image processing was performed in MATLAB 2016a (MathWorks Inc., Natick, MA).

TABLE II.

MRI Acquisition Parameters for the Datasets in this Study

Sequence	TE (MS)	TR (MS)	FA (°)	BW (kHz)	FOV (CM2)	Matrix Size
In vitro
bSSFP	1.7	4.63	5/15/90	125	30, 36	384×384
B1+ map	13.4	28	5	15.63	36	128×128
Temperature
GRE	3.7	100	5/20/45	15.63	24, 30	384×384
Thermometry
fSPGR	16.1	120	30	7.81	32	160×128
FSE	14	425	90/180	15.63	32	256×256
In vivo
GRE	3.7	100	5/20/45	15.63	24, 30	384×384
B1+ map	13.2	28	5	15.63	24	128×128

D. Simulations

All simulations were performed using finite element method software (COMSOL 5.3a, Burlington, MA) on a 64-bit Windows 7 computer with 32.0 GB RAM, and an Intel® Xeon® CPU E5–1620 0 @ 3.60 GHz. The simulated double helix was modeled as a perfect electric conductor with no resistance, in parallel with a 14 pF capacitor, and had a radius of 2.2 mm, and a pitch distance of 3.455 mm for an angle of 22.5°. A frequency domain analysis of the electromagnetic field was performed at 130 MHz.

E. Temperature measurements

A hydroxyethyl cellulose gel was prepared [22], and four temperature probes (Luxtron, LumaSense Technologies, Inc. Santa Clara, CA) were placed in the gel. Temperature was recorded every second during a fast spin echo (FSE) sequence, as well as two minutes before and after the scan, for a total scan time of 15:02 min (Table II). The baseline temperature of each probe at room temperature was subtracted for comparison. We then acquired a set of MR thermometry images in an oblique scan plane with a 10 mm slice thickness, parallel to the catheter using a fast spoiled gradient echo sequence (FSPGR) with T_acq = 16 sec, interleaved with the periods of FSE acquisition with the number of averages of 1, 2, and 3 (Table II) for a total scan time of 4:37 min. Temperature maps were generated offline with proton resonance frequency shift (PRF) technique.

F. In vivo experiment

The study protocol was approved by the university’s Institutional Animal Care and Use Committee. A catheter was placed into each common carotid artery of a single swine (43.2 kg, female), using a 30 cm × 30 cm flat panel C-arm X-ray system guidance (Cios Alpha, Siemens Healthineers, Munich, Germany). A gradient echo (GRE) sequence was acquired for visualization, and a Bloch-Siegert B₁+ map [20], [21] was acquired for flip angle quantification, using a sagittal scan plane, a 10 mm slice thickness, and the imaging parameters in Table II. The same scanner and coils were used as in the in vitro experiments, and either a coronal or sagittal scan plane was used, with a 10 mm slice thickness. An ROI was manually drawn around the entire marker to measure the mean signal, and then the ratio with the nearby background water signal was calculated to measure the relative amplification factor

III. Results

A. Resonant marker

The microscope images of the 10-turn helical trace are shown in different orientations, showing the double helix, the cross-over points, and the capacitor (Fig. 2b). The estimated maximal thickness is 2.264 mm due to an estimated 16 μm z-height thickness at the cross-over points, compared to approximately 3.2 mm maximal thickness for a hand-built resonant marker (Fig. 2c, d), including heat shrink.

B. In vitro image acquisition

The low flip angle bSSFP sequence (5° and 15°) shows the signal amplification of the markers relative to the background signal, in comparison with the high flip angle (90°) sequence (Fig. 3a). For the B₁+ map sequence, the relative amplification factor measured 1.67 in the parallel orientation, and in the perpendicular orientation, 1.63 (Fig. 3b). Two line plots of the bSSFP signal show the alternating signal between the helices (Fig. 3c).

Fig. 3.

In vitro results in phantoms at 3T show good signal. The diagram, aerosol jet deposition prints, and bSSFP sequences of the double helix resonant marker are shown oriented a) parallel to and perpendicular to B₀ using a coronal scan plane, and a bSSFP sequence with a 10 mm slice thickness. The sequences with low flip angles of 5° and 15° show the surrounding signal amplification of the circuits, which can be compared with a 90° flip angle, showing the background and over-flipping of the signal. b) The bSSFP sequence, and B₁+ map are shown for the full-length marker. c) A line plot of the signal shows the alternating signal between the helices. d) Simulation of the electromagnetic fields confirms the magnetic field pattern.

C. Simulations

The magnetic flux density is shown for the double helix wireless resonant circuit and the surrounding region, showing patterns of enhancement and cancellation, and confirming the experimental results (Fig. 3d).

D. Temperature measurements

No appreciable heating was measured by the temperature probes (Fig. 4 a, b). MR thermometry images confirmed this by demonstrating no detectable heating (Fig. 4c).

Fig. 4.

Temperature measurements and MR thermometry in a phantom at 3T demonstrate minimal heating. a) The setup of the temperature probes in the gel is shown. b) Compared to the reference probe, no substantial heating was observed. c) This was confirmed by MR thermometry maps after three FSE sequences, increasing in time.

E. In vivo experiment

The coronal GRE sequence shows good signal and contrast for the catheter on the left side (Fig. 5a), and the two catheter placements can be seen on X-ray (Fig. 5b, c). The Bloch-Siegert B₁+ map sequence demonstrates quantification of the flip angles (Fig. 5d). The mean scaled signal of the marker was 9.07°, and the mean scaled signal of the background tissue was 5.0°, for a relative amplification factor of 1.81. Using a lower flip angle of 5° demonstrates the low background signal, and the signal amplification of the marker (Fig. 5e). As the flip angle increases to 20°, there is more signal in the surrounding tissue, and the contrast decreases (Fig. 5f). With a higher flip angle of 45°, there is over-flipping within the marker, causing a decreased signal, while the background tissue signal increases (Fig. 5g).

Fig. 5.

The wireless resonant markers demonstrate good signal in an in vivo preclinical scenario at 3T. The markers placed in the right (dashed arrow) and left (solid arrow) carotid arteries of a single swine are shown in a coronal scan plane under a) MRI and b) X-ray, which are more easily seen under c) magnification (left artery with contrast, and the right artery without contrast). d) In a sagittal plane, the B₁+ map quantifies signal amplification. e) The GRE sequence with a lower flip angle (5°) demonstrates low background signal and marker signal amplification. f) As the flip angle increases (20°), there is more signal in the surrounding tissue. g) With a higher flip angle (45°), there is over-flipping near the marker, causing a decreased signal, while the background signal increases.

IV. DISCUSSION

We developed a fully printed and functional wireless resonant marker on a catheter for MRI tracking using aerosol jet deposition. A double helical inductor and a long capacitor were printed using silver nanoparticle ink and polyimide ink on a polymer catheter. The method consisted of a resonant circuit, in the form of a distributed LC circuit, which inductively coupled with the MR signal, as well as the RF receiver coils, to make the system wireless. Preliminary feasibility of using the marker in an interventional MRI setting was demonstrated in vitro and in vivo, and no appreciable heating was measured. This method introduces a complete, low profile resonant marker that is possible to print on a polymer catheter, and exhibits good tracking characteristics at 3T.

A resonant marker and catheter system of this size may allow further navigation into more distal vessels than prior approaches to endovascular interventional MRI allowed. The average internal carotid artery diameter is 4.66 ± 0.78 mm for women, and 5.11 ± 0.87 mm for men [23]. Given that our final PEEK catheter with the aerosol jet deposited circuits had an outer diameter of approximately 2.264 mm (6.78 Fr) and the hand-wound prototypes had an OD of approximately 3.2 mm (9.6 Fr), the aerosol jet-deposited catheter can more readily and safely navigate most cervical internal carotid arteries, which is the location where guide catheters generally terminate in the setting of contemporary bi-axial or tri-axial stroke treatments.

Other groups have used aerosol deposition to print the inductor of a resonant marker for MRI tracking. A prior group used aerosol deposition to print a gold plated copper solenoid inductor on a catheter, and connected it to a 100 pF discrete capacitor, with dimensions 1 mm × 0.5 mm × 0.6 mm, inside the catheter lumen [24], and electromagnetic simulations concurred with experimental validation at 1T [19]. Moving towards printing the marker entirely on the surface, another group demonstrated feasibility in depositing a single layer of a dielectric, lead zirconate titanate, at room temperature using aerosol deposition, in order to be compatible with the low melting temperature substrate of the polymer catheter, and measured the permittivity (ε) to be about 11 at 1 MHz [25]. The authors detected a resonant response at 43 MHz (1T) based on benchtop tests, but image enhancement for tracking purposes was not demonstrated [25]. We have extended these principles to demonstrate a working resonant marker prototype operating at 3T.

We discovered several trends during this study that provide insight into different design and microfabrication methods. The distributed capacitance of the long twin-lead transmission line allows the interventional radiologist to visualize the length and orientation of the catheter, which is similar to the tracking characteristics of some catheters tracking markers under X-ray. Similar wireless devices have been tracked using an interleaved sequence, where a low flip angle RF pulse, with phase-dithering applied, is followed by gradient-echo readouts along the x, y, or z gradient axes, which is multiplexed with a Hadamard scheme [11]. The unique pattern at the tip of the catheter clearly delineates the tip, and pattern recognition could be used to determine the orientation and direction of the catheter. We experimentally observed different magnetic field zones, in the helical region, and the leaking flux region of the transmission line capacitor, generating unique field patterns. For the aerosol jet deposition printing, the total amount of time required was approximately 6–7 hours. A spray coating method to apply the top insulation layer was explored in order to save processing time; however, the coating showed uneven distribution of ink leading to ink pooling and fractures. We did not observe any image artifact from the polymer catheters themselves, and we anticipate minimal susceptibility-induced artifact, since PEEK has less than a 3 ppm difference in magnetic susceptibility from that of water and tissue [26].

This study considered printing markers on catheters that are manipulated by interventional radiologists; however, these techniques may be helpful for other MR applications. One group recently published a closed-loop method for controlling a rotational joint using robotics in combination with inductively coupled coils as fiducial markers [27]. Clinical applications may include printing these markers on catheters for MR-guided electrophysiology procedures in the large cardiac chambers [28]. A wireless resonant marker could also be incorporated onto an endovascular imaging coil, as has been demonstrated previously [29].

While the strengths of this work include the proof of concept of printing a complete LC circuit using aerosol deposition and a double helix inductor, this study does have limitations. The circuit design is limited by the significant length of the capacitor. Inductively coupled coils amplify the specific absorption rate (SAR) by coupling to both the electric and magnetic fields. It is possible that there are unwanted parasitic capacitance effects through the water phantom, due to the proximity of the circuit components near each other, and there are likely some common RF mode contributions due to the electric field coupling. A majority of signal appears to come from the capacitor, due to the displacement current. A dielectric ink with a higher permittivity could be used to shorten the capacitor. While one group explored the use of lead zirconate titanate as a dielectric with higher permittivity [25], our approach avoids the safety issues with using lead by keeping the cure temperature at a temperature just low enough (140°C) to avoid melting the outermost layer of catheter, since the thermoplastic elastomer has a melting temperature of 134–174°C. Different capacitor designs could also be used to reduce the overall electrical length, while keeping the capacitance the same. The skin depth of silver at 128 MHz should be 5.603 μm, and the conductive traces are estimated to have a total thickness of about 4 μm. Assuming similar values for the silver nanoparticle ink traces, the conductor thickness is thinner than the skin depth, and so with a thicker conductive path, the losses could be less, and the B₁+ amplification higher. Since the circuit was sealed with the polyimide ink, we did not measure the total equivalent resistance of the coil. Additionally, at each cross-over point of the double helix inductor, a capacitor is effectively formed, resulting in some self capacitance.

While we observed no obvious temperature increase in the temperature probe experiment and the MR thermometry experiment, the possibility remains that the local flip angle amplification could result in heating and increased local SAR. The resonant marker was tested at the center of the magnet bore, but the highest electric fields, and therefore heating, are expected at the edge of the bore. Thorough safety analyses using the standard measurement for RF-induced heating, including both experiments and simulations, should be performed before any human studies [22].

There are several other possible solutions to mitigate RF-induced heating. One common hardware solution is to decouple the circuit during RF transmission by adding anti-parallel or crossed decoupling diodes to the circuit to limit the amplification [30]. These commercial diodes are currently available as thin wafers, which we will consider in the future. One group implemented optically detunable LC circuits [31], [32], using a laser diode to illuminate a photodiode through an optical fiber, in order to decouple the resonant marker. Another group recently demonstrated feasibility of an all-printed diode at ultra-high frequency, which could be printed with a similarly low profile as the resonant marker [33]. A software-based solution to potential heating could be to use a parallel transmission system and optimization approach to control RF induced heating [34]. An important safety consideration is transmit calibration, which may be heterogeneous throughout the volume.

The MRI experiments and post-processing also contain some limitations. The regions of interest drawn include a heterogeneous amount of signal due to the magnetic field pattern of the double helix, which may affect the relative signal amplification. In vivo, there are a variety of possible motion artifacts, due to respiration, cardiac, or pulsatile blood flow effects. We did not compensate for any of these effects, and in future experiments, motion compensation techniques, such as respiratory gating, may improve the signal amplification factor.

V. Conclusion

Imaging, B₁+ mapping, and temperature experiments demonstrated that the wireless resonant circuit printed on an endovascular guide catheter produces an induced current, transferring the RF into local signal amplification. We demonstrated proof-of-principle of microfabricating a fully aerosol jet-deposited complete LC circuit with a double helix inductor. The in vivo experiment at 3T further demonstrated that this method could potentially be used for a low profile guide catheter that can be tracked under MRI for endovascular neurointerventional applications.