A Novel Expandable Catheter Wireless Amplified NMR Detector for MR Sensitivity Accessing the Kidney in Rodent Model

Abstract

This article demonstrates the enlarged effective range for MRI sensitivity enhancement with a deformable catheter MRI coils integrated with a wirelessly powered amplifier. The expandable balloon Wireless Amplified Nuclear magnetic resonance Detector (WAND) is constructed on a copper-clad polyimide film to resonate at the first and second harmonics of the proton Larmor frequency at 7 Tesla. The WAND is then mounted on a balloon catheter system for easy delivery inside confined orifice. Upon reaching the region of interest, it is unfolded out of the sheath tube to increase its effective size. Magnetic Resonance (MR) imaging experiments with and without the WAND are performed both in a water phantom and in a live rat to evaluate the WAND’s sensitivity advantage. Expanded from a 3 mm diameter in its folded state, this deformable WAND can change its width by > 100% in its inflated state to at least 6 mm, leading to a sensitive detection region extending to up to 20 mm in the transverse direction. When the deformable WAND is placed in an artery in the region of the kidney of a live rat, it could achieve at least a 10-fold SNR gain over images acquired by a standard external detector of 22 mm diameter, even though the region of interest is separated from the WAND’s surface by a distance larger than the WAND’s own width. The proposed expandable catheter WAND could significantly enlarge the effective range for MR sensitivity enhancement in-vivo, enabling versatile applications in interventional MRI.

I. INTRODUCTION

MANY biomedical applications such as human and small animal subjects in Magnetic Resonance Imaging (MRI) require high spatial resolution, together with high signal-to-noise ratio (SNR) and reasonably short acquisition time for proper diagnosis and treatment of medical conditions [1]. This would be even more crucial and challenging when imaging deep lying tissues of interest. Although, there has been significant progress in improving the MRI sensitivity detection by increasing the field strengths [2], working with RF coil arrays [3], and parallel imaging acquisition [4]-[8], these techniques still have their own limitations to obtain sufficient signal sensitivity, especially in target regions that are deep inside the body. Recently, a complementary method is demonstrated to improve the detection sensitivity of the deep lying tissues using a Wireless Amplified Nuclear Magnetic Resonance (NMR) Detector (WAND) [9]-[12]. In this approach, weak MR signals, which are emitted from deep lying Regions-of-Interest (ROIs), could be sensitively detected by the use of a localized WAND with an integrated amplifier. In contrast to traditional micro-coils [13], [14] using wired connections, the WAND can sensitively detect deep-lying target regions by amplifying weak MR signals before transmitting them to the external receiving coil, thus reducing sensitivity loss during passive inductive coupling [15], [16]. The WAND has a built-in parametric amplifier that can harvest a wirelessly provided pumping power to amplify weak MR signals in-situ. A parametric amplifier [17]-[22] is a radio-frequency (RF) amplifier that can amplify the power level of weak input signals under the influence of energy utilized from the external pump source, leading to periodic variation of its circuit parameters, such as its capacitance [9], [23]. As a result, due to the time varying capacitance of the varactor diode [24], linear amplification of input signals could occur.

WANDs [9]-[11] are initially developed as implantable detectors to observe individual nephrons in vivo, but with sub- optimal homogeneity. Subsequently, they are redesigned with cylindrical symmetry [12], [25], [26] to get a panoramic view of their surrounding areas. Yet, all these detector designs are restricted by their limited diameters to effectively observe wider surrounding regions. As a result, their dimensions are determined by the narrowest orifice along the insertion pathway, making them unable to observe relatively larger openings in internal body cavities. For example, blood vessels reaching the heart are narrower than cardiac chambers where there is a need for an expandable detector. As another example, it is challenging to pass a fixed-volume detector through the stenosis or plaques in blood vessels as the path is narrowed by fat deposition. Similarly, other desirable applications could be inserting the detector into male urethra to reach the bladder to detect nearby organs like the prostate, or into the female vagina to observe ovaries, etc. Thus, our approach is focused on bridging this gap by successfully designing and fabricating a deformable WAND. The present study proposes an endo-rectal balloon catheter system with an expandable WAND mounted on it. Fabricated on a single layer of deformable substrate, the WAND that is initially rolled-up within a 3 mm diameter sheath can be inflated by ≥6 mm diameter expandable balloon in the target cavity to observe enlarged regions of interest in its surroundings. The electrical and mechanical characteristics are presented and tested in phantom and in vivo experiments.

II. Materials and Methods

A. Operation Principle

The expandable double frequency WAND works on the principle of non-degenerate parametric amplification. In contrast to the degenerate parametric amplifier, where both the signal and idler frequencies are identical, in a non-degenerate parametric amplifier there are interactions between the three distinct waves the pumping wave at frequency ω₃, the signal wave at frequency ω₁ and the so called ‘idler’ wave at frequency ω₂ = ω₃ – ω₁ that is generated during the interaction. The term idler frequency is used in parametric amplifiers (paramps) to specify the difference between the pumping frequency (ω₃) and the signal frequency (ω₁). During signal reception, the varactors in the WAND enables a weak MR signal at the Larmor frequency ω₃ with phase Φ₁ to mix [27] with a strong pumping signal at ω₃ with phase Φ₃ that is applied orthogonally to create an amplified output at the idler frequency ω₂ = ω₃ – ω₁ with phase Φ₃−Φ₁. This amplified output at ω₂ can mix back with the pumping signal at ω₃ to create a second amplified output at ω₁ with phase Φ₃−(Φ₃−Φ₁), which can be detected by a standard external receiving coil (Fig. 1) [28], [29]. Basically, the enhancement procedure suggests that the pumping signal provides power needed for in-phase signal amplification at the input frequency.

Fig. 1:

The schematic diagram of a deformable WAND with a higher frequency resonance mode that is sensitive to the pumping signal at ω₃, a lower frequency resonance mode that is sensitive to MR signals at ω_1· and the ‘idler’ signal at ω₂ that is created during frequency mixing.

The deformable WAND is normally a non-linear double frequency resonator, which can receive both the MR signal at the Larmor frequency ω₁ and the pumping signal at approximately twice the Larmor frequency i.e., ω₃ ~ 2 ω₁. It is required that the difference | ω₃−2 ω₁| should be greater than the imaging bandwidth such that the amplified MR signal at ω₁ and the generated idler signal at ω₂ do not interfere with each other. Before the actual imaging experiment, the required pumping power is empirically adjusted until a single intense peak is observed in the spectrum window. A reduced power level of 0.5 dB below this threshold is used for subsequent imaging experiments. In principle, both the MR signals and the idler signals can be detected simultaneously if the image bandwidth is large enough. To be compatible with standard image processing procedures, however, the idler signals are often filtered out when the image bandwidth is reduced below |ω₃−2 ω₁|. By setting the pumping frequency ω₃ slightly above 2 ω₁, the idler frequency ω₂ is also slightly above the signal frequency ω₁, but nonetheless both signal and idler frequencies are well below the pumping frequency.

Fig. 2 shows the Fast Spin Echo (FSE) pulse sequence diagram that includes the RF excitation pulse during the transmit (Tx) mode, as well as the pumping pulse and the reception channel during the receiving (Rx) mode. In this representative diagram, a single 90° RF-excitation pulse is followed by a series of 180° refocusing pulses to generate a train of echoes during which a pumping pulse is kept on. During each RF-transmission pulse, the deformable WAND is detuned by the strong excitation pulse because varactors are strongly modulated during RF excitation. During signal reception, weak MR signals are amplified by the WAND through the parametric frequency mixing process [30]-[32]. These amplified signals are wirelessly coupled to the external receiving coil before being processed by the scanner console.

Fig. 2:

A representative diagram of Fast Spin Echo (FSE) pulse sequence showing the RF excitation pulse, followed by pumping pulse and simultaneous signal reception. The WAND is detuned by the RF excitation pulse but is wirelessly activated during the subsequent signal acquisition period.

B. Resonance Modes

The design of an expandable catheter WAND is based on a multi frequency resonator. Unlike previous designs of WANDs [9]-[11], [26] made of rigid circuits utilizing orthogonal inductors to create multiple resonance modes in 3D structures, the deformable WAND proposed here is mounted on a planar conductor pattern that can be easily scrolled into a roll (Fig. 3). This pattern has a lower frequency dipole resonance mode, which is centered around the Larmor frequency ω₁ of a 7 T MR scanner. As with conventional surface coils, this lower frequency dipole resonance mode exists when the majority of the current flows around the circumferential conductor legs. In this case, the currents from the two sub-sections of the WAND circulate in-phase to make this mode sensitive to the precessing nuclear magnetization (Fig. 3a). To create additional resonance modes, vertical conductor legs are incorporated in the center symmetric plane to minimize their influence on the dipole mode. When a continuous vertical leg is incorporated to neutralize extra charges across the varactor diodes accumulated during the RF excitation pulse, a low frequency butterfly resonance mode is created. This mode has counter- rotating currents flowing through the center vertical leg that separate the circuit’s two subsections (Fig. 3b). This mode is not utilized for parametric frequency mixing, so it will not be considered in the following discussion. To create an additional resonance mode at about twice the Larmor frequency, two additional vertical legs split by chip capacitors are incorporated to the left and right sides of the center continuous leg (Fig. 3c). Because capacitors have lower impedance while inductors have higher impedance at higher frequencies, the majority of current passes through these two chip capacitors in the high frequency butterfly resonance mode. Just as in Fig. 3b, the current flows out-of-phase in the circuit’s two sub-sections (Fig. 3c). When the circuit is curved around a cylindrical balloon, the mode illustrated in Fig. 3c will be sensitive to a pumping field applied horizontally. Because the cathodes of the two diodes are connected to the same terminal, they will undergo the same level of modulation by the pumping field. To enable efficient energy exchange between the dipole mode and the higher-frequency butterfly mode, varactor diodes C₁ and C₂ are mounted on the left-most and right-most legs where sufficient current flows at both frequencies. As a result, multi-stage parametric frequency mixing can occur. It is noteworthy to mention that all varactor diodes used in our WANDs are zero-biased; their capacitance is wirelessly modulated by the pumping signal in a sinusoidal way. The requirement for no bias voltage [33] has enabled the implementation of a wireless amplifier in a compact circuit.

Fig. 3:

Current flow in different resonance modes. (a) In the horizontal dipole resonance frequency mode (low frequency mode) that is sensitive to precessing nuclear magnetization, currents from the two sub-sections circulate in phase. (b) In the low frequency butterfly resonance mode that is used to stabilize diodes in the circuit, the majority of the current flows through the center continuous leg (c) In the high frequency butterfly resonance mode that is sensitive to a pumping field, the out-of-phase currents from the two subsections circulate through the two chip capacitors in the center.

To characterize the deformable WAND, its transmission coefficient (S₂₁) is measured by placing it above a dual pick-up loop connected to the two ports of the vector network analyzer. According to the S₂₁ vs frequency curve in Fig. 4, the WAND in its passive state has a low resonance frequency of 299.6 MHz (Q = 57) when in water. This lower resonance frequency is about half of its higher resonance frequency at 610 MHz (Q = 67). When the pumping field B_pump is applied, the curve has a significant rise in peak height with a shift in the peak center. This increase in gain and corresponding decrease in bandwidth of the deformable WAND is due to efficient multistage frequency mixing that provides regenerative energy to compensate for circuit losses. Thus, the effective quality factor of the WAND, defined as the ratio between the resonance frequency and its bandwidth at −3 dB <mi>, would increase with the application of pumping power. When the applied pumping field is sufficiently large to induce enough modulation in the zero- biased varactor diode, the parametric WAND will begin to self-oscillate at a frequency that is half of the pumping frequency. To enable a gain of 17.5 dB with a large enough bandwidth of ~690 kHz, it is necessary to reduce the pumping power by 0.5 dB below its oscillation threshold. As shown in Fig. 4, even though the passive resonator is detuned from the Larmor frequency, its peak response can still be brought back to the Larmor frequency by setting the pumping frequency to approximately twice the value. The WAND’s unique capability for wireless adjustment would make it particularly suitable for use inside the body where wired connections are either inconvenient or impractical.

Fig. 4:

S₂₁ transmission curve of the WAND. The solid curve is measured using the passive WAND in the absence of a pumping field, while the dashed curve is measured using the active WAND with the pumping power maintained at 0.5 dB below the detector’s oscillation threshold.

C. Balloon catheter deformable WAND Fabrication

The deformable copper loop for the WAND is fabricated from a polyimide foil (FR8501, DuPont), which is etched using FeCl₃ solution, the final version of which is shown in Fig. 5b. This so called ‘endo-rectal’ WAND is mounted on an expandable balloon catheter system as shown in Fig. 5a. The rectangular circuit pattern has a total length of 12.1 mm, a width of 7.2 mm and a strip width of 0.5 mm (Fig. 5c). The WAND consists of a continuous vertical leg in the center. To the left and right sides of this continuous vertical leg are two vertical legs both of which are split by a 1 mm gap. The capacitance values of varactor diodes C₁ and C₂ are 22 pF at zero bias and the chip capacitors C₃ and C₄ have values of 10 pF. This polyimide-etched WAND is left with a spare region of 12 mm at the top and its bottom (Fig. 5b) to mount it around a 7 mm × 20 mm balloon catheter system (Fig. 5a). Before mounting, the circuit is coated with a urethane seal coat spray (18411, CRC Industries, Inc.) to improve biocompatibility and electrical isolation. With clear appearance and high dielectric strength, this product dries fast without disturbing the circuit’s flexibility. After complete drying of the coating layer, the WAND is mounted around the balloon surface of a percutaneous transluminal angioplasty (PTA) balloon dilatation catheter (DR8062, Dorado) using an adhesive (Loctite 422 and SF 7452, Henkel). In addition, the tip of the balloon catheter is mounted with a 3 mm long plastic screw with a 3 mm diameter round cap (Fig. 5a). When the balloon is deflated, the WAND is squeezed inside a red sheath along with the balloon. Protected by the round plastic screw at its tip (Fig. 5d), the sheath can easily slide inside a confined orifice. Upon reaching an open cavity near the region of interest, the deformable WAND can be unfolded by the balloon when they are pushed out of the sheath. The balloon catheter has a diameter of 7 mm, allowing the WAND to reproducibly expand from 3 mm diameter in its rolled-up configuration to at least 6 mm width in its expanded state, in order to obtain a larger field-of-view (FOV).

Fig. 5:

An expandable catheter Wireless Amplified NMR Detector (WAND), (a) A practical realization of a deformable catheter WAND mounted at the distal tip of a 7 mm × 20 mm Dorado PTA balloon dilatation catheter. Inset: Expanded view of the deformable WAND. The WAND is partially inflated, and it includes a small atraumatic screw cap tip to facilitate the catheter’s insertion through a narrow orifice. The guide wire is 80 cm long. The proximal portion of the catheter consists of a female luer-lock hub connected to an inflation lumen, and a female luer-lock hub connected to a guidewire lumen. A re-wrapping tool is also available on the catheter shaft (transparent white) but this function is performed by a 3 mm diameter tube (red) that is inserted later in the catheter guide wire. (b) Actual circuit used as a deformable WAND in (a). (c) Schematic circuit diagram sketch of the WAND shown in (b). C₁ and C₂ are 22 pF capacitances associated with varactor diodes. C₃ and C₄ are 10 pF ceramic capacitors. Dimension of the WAND is 12.1 mm × 7.2 mm and the distance between the two center vertical legs with chip capacitor is 1 mm. (d) WAND in its folded state. The red tube shown in (a) is advanced to fold the balloon catheter WAND inside and before it is inserted into the rat’s rectum for MR imaging.

To demonstrate its robust operation, we intentionally compress and inflate the WAND for multiple cycles to measure its resonance frequencies on bench. Table I lists frequency peaks in the S₂₁ response curve (ω₁ and ω₃, respectively) measured using the vector network analyzer (VNA) during the WAND expansion in air and in water. It summarizes the slight frequency shift due to a slight structural variation each time the WAND is inflated. However, because the Larmor frequency always falls within the bandwidth of the WAND in its inflated state, it is very easy to wirelessly readjust the WAND’s optimal operation frequency by setting the pumping frequency slightly above twice the Larmor frequency for efficient amplification. After imaging, the WAND could again be retracted into the sheath and folded into the initially rolled-up configuration owing to the tapered geometry created at the bottom half of the polyimide film.

Table I:

Resonance Frequencies of the Inflated WAND

Serial Number (S.N.)	Inflation in Air (MHz)	Inflation in Water (MHz)
1.	305.8,623.6	297.9,610.7
2.	308.1,625.3	299.6,612.2
3.	306.7,620.8	299.3,610.9
4.	306.3,622.2	297.8,611.3
5.	307.5,623.3	298.5,610.6

D. MR Imaging

Our deformable WAND does not require an impedance matching network or a control chain for varactor diodes, because the WAND can be wirelessly adjusted and detuned. To demonstrate the performance of the deformable WAND, we first place it in the center of a rectangular box fully filled with water to acquire amplified images. To wirelessly feed pumping power to the WAND, a pair of horizontal rectangular loops mounted on a yellow plastic cylinder is inserted and positioned perpendicular to the static magnetic field B₀ (Fig. 6a). Both pumping loops are driven by the same coaxial cable. To receive amplified MR signals emitted from the WAND, an external surface coil is placed beneath this detection object, which is about 2.5 cm from the surface of the expandable catheter WAND. In this way, MR signals are amplified by the WAND in situ before they reach the external surface coil. The whole apparatus assembly is then loaded inside a 7 T MR scanner equipped with a 75 mm horizontal bore, inside which a standard volume coil (Bruker Biospin Bilerica, MA) is used for slice excitation. Decoupling of the WAND from the volume coil is done wirelessly by the RF excitation pulse that induces a modulated voltage across the varactor diode. Because both diodes are connected in a head-to-head orientation, during each half of the modulation cycle, at least one diode will be switched on to detune the resonator as soon as the induced voltage across the diode exceeds the forward threshold voltage (~0.2 V for the diodes used in this work). Wireless detuning of the transmitted pulse minimizes the amount of radio frequency (RF)-induced heating, as confirmed by a temperature probe immersed immediately beneath the WAND that shows no measurable temperature rise. During signal acquisition, less than 10 mW of RF pumping power is required on an external RF signal generator to wirelessly activate the deformable WAND. The requirement for a very low level of pumping power again corroborates the undetectable RF-induced heating.

Fig. 6:

Experimental setup for animal MR imaging. In the schematic diagram shown in (a), external pumping signal is provided to the WAND via a pair of inductor loops mounted on the yellow cylinder shown in (b). A rat is placed in the supine position with a flexible WAND non-surgically inserted inside the colon through its rectum. The RF excitation pulse is provided by a Bruker volume coil (not shown) while signal reception is performed with a 22 mm diameter surface coil placed beneath the rat which then is connected to the MR receiving channel through a Bruker preamplifier. The rat is continuously given anesthetic inhalation and monitored with a breathing sensor during scanning.

All animal experiments are operated in accordance with guidelines set by the Institutional Animal Care and Use Committee (IACUC). During this process, a 250-gm rat is anesthetized with 2% to 3% isoflurane in 96% oxygen prior to insertion of the endo-rectal WAND. Then, the rat is placed in its supine position in the sample cradle and its abdomen is taped to the bench surface to reduce any physiological motion artifact during MR imaging. When the animal lies inside the horizontal bore MR scanner, large portions of their digestive tract, such as the esophagus and rectum, are parallel to B₀. Because the dipole mode of the WAND is perpendicular to its long axis, when we insert the WAND mounted on a catheter into the rectum, it will be naturally aligned at the correct orientation for MR signal reception (Fig. 6b). To insert the WAND near the kidney region, the insertion depth of the catheter is manually adjusted by observing the marking point on the guide wire until the folded WAND is located near the kidney region as confirmed by localizer images. Subsequently, the endo-rectal WAND is pushed out of the sheath and inflated by a syringe connected to the terminal end of the catheter. After amplified images are acquired, the WAND is retracted into the sheath before being removed in its contracted state. For sensitivity comparison, images are taken with the external surface coil alone, using the same acquisition parameters as are used with the endo-rectal WAND.

E. Image Processing with MATLAB

In order to evaluate the signal sensitivity enhancement of the deformable WAND in the water phantom, intensity normalized images acquired by the active resonator are divided by the reference images acquired by the external receiving coil to obtain the relative sensitivity maps. For kidney imaging, normalized intensity profiles are obtained for each pixel to demonstrate the signal intensities acquired with and without the use of the endo-rectal WAND.

F. Simulation of Specific Absorption Rate (SAR)

3D simulation is performed using CST Studio (Dassault Systèmes, France) to evaluate the potential heating effect. For computational simplicity, the curved detector circuitry (Fig. 7a) is modeled to be soaked inside a uniform medium (ε = 1, σ = 0.59 S/m), with all capacitive components soldered in the center positions of vertical legs. To calculate the Specific Absorption Rate (SAR) induced by the slice excitation pulse, the left most and the right most legs are both modeled to have an excitation port whose excitation power is adjusted to obtain a peak voltage of 0.2 V, i.e. the diodes’ threshold voltage. According to the transverse and longitudinal SAR profiles (Figs. 7b, 7c), there is non-negligible SAR only within narrow regions around the two varactor diodes (C₁ and C₂) where voltage drop is appreciable. The peak SAR value of ~0.22 W/kg is much smaller than the lower level of whole-body SAR recommended by IEC 60601–2-33 (2 W/kg). On the other hand, to calculate the SAR induced by the pumping signal, the excitation port is mounted on the center continuous leg (Fig. 7d) whose excitation power was adjusted to obtain a peak voltage of 0.044 V that is required to induce a 1.75% modulation in capacitance. This level of capacitance modulation can completely compensate for energy loss of the resonator with a <mi> of 57 at its lower resonance mode, based on the simple relation <mi> between the modulation index M and the detector’s quality factor <mi> [9]. As shown by the axial (Fig. 7e) and longitudinal (Fig. 7f) SAR profiles, there is non- negligible SAR value only near the two chip capacitors in the center (C₃ and C₄). Within this narrow region, the highest SAR is 2 W/kg, which is still within the safely level. According to the linear relation ΔT/Δt = SAR/C_p that correlated the temperature rise AT with SAR i.e. an SAR of 2 W/kg corresponds to a temperature rise of 0.5 mK per second, assuming the tissue specific heat C_p to be ~4 kJ kg⁻¹ K⁻¹. This is equivalent to a small temperature rise of 0.5 K after 17 mins of continuous pumping. This is a time frame sufficient for most MRI experiments. The small temperature rise could be further reduced by turning off the pumping field during the recycle delay.

Fig. 7:

To simulate the dipole resonance mode at the Larmor frequency, the circuit diagram in (a) is modeled to have a symmetric pair of voltage sources distributed on the left most and the right most legs, with the calculated SAR patterns displayed in the transverse (b) and vertical (c) plane. The white dashed line in (b) defines the position of the longitudinal slice shown in (c), and white solid lines define the outer contours of the WAND. To simulate the butterfly resonance mode at the pumping frequency, the circuit diagram in (d) is modeled to have a single voltage source in the center leg, with the corresponding SAR patterns displayed in the transverse (e) and vertical (f) plane.

G. Dynamic range

To evaluate the dynamic range of the WAND, the apparatus assembly depicted in Fig. 6 is used, where the input signal comes from a 1 cm circular loop that is cable-connected to an external signal generator to emulate MR signals with controllable input level. By placing this single-turn loop 5 mm away from the WAND and aligning it perpendicular to the external surface coil, direct coupling from the input loop to the external surface coil is minimized. As shown in Fig. 8, an input level higher than 108 dBm would lead to an observable output level of 109 dBm that stands above the spectrum baseline. The slightly reduced output level compared to the input is a result of efficient amplification that mostly compensates for signal intensity loss due to passive inductive coupling. The linear relation between the input and output level is maintained within a 78 dBm range, until the input power level reached a level of −27 dBm to induce a 1 dBm compression in the output signal. Because the WAND is optimized for signal amplification of local regions, this 78 dBm dynamic range would be enough for most imaging experiments.

Fig. 8.

The linear region of the WAND that can amplify signals relayed from the input loop to the external receiving coil.

III. RESULTS

A. Phantom Imaging

Phantom images are acquired using a multi-slice fast-spin echo (FSE) sequence based on the following sequence parameters: TR/TE = 1000/7.9 ms; FOV = 60 × 30 mm²; matrix size = 256 × 128; FA = 90°; imaging-bandwidth = 50 kHz; slice thickness = 0.8 mm; NA = 1 and TA = 24 s. In Fig. 9, row 1 depicts the longitudinal (left) and transverse (right) images taken by the WAND in its active state. Relative sensitivity maps (row 2) are computed by dividing the images in row 1 with those images obtained by the external surface coil only. As shown in the transverse profiles, this expandable catheter WAND has a distance dependent detection sensitivity, just as any other surface coils. However, this endorectal WAND can achieve more than 10-fold sensitivity gain for regions separated from the surface of the WAND by a distance larger than its own width. This level of sensitivity gain is preserved within a 15 × 13 mm² region in the longitudinal profiles. The WAND’s favorable feature for sensitivity enhancement is attributed to increased signal transmission efficiency made possible by in situ amplification. Furthermore, as seen from the transverse profile, the WAND in its initial 3 mm diameter roll-up configuration is enlarged to at least 6 mm in width around it, leading to a sensitively detectable region extended to up to 20 mm. The fine structures shown in the transverse image are just the outer contour of the inflated balloon and the curved detector circuitry mounted on a scrollable film. Overall, the detector has a pretty homogeneous profile in its vicinity, with the hyperintense spots near its edges corresponding to regions immediately adjacent to its vertical conductor legs.

Fig. 9:

Longitudinal (Column 1) and transverse (Column 2) images and their corresponding relative sensitivity maps evaluated after the WAND is inserted inside a water phantom and inflated. Parameters used are TR/TE = 1000/7.9 ms, 90° FA, 60 × 30 mm² FOV, 256 × 128 matrix, 50 kHz bandwidth, NA = 1, and TA = 24 s. Row 1 are the images acquired when the deformable WAND is active, and Row 2 are the corresponding relative sensitivity profiles obtained with the active detector divided by the images acquired with an external surface coil. The fine structures in the transverse image are actually the outer contour of the inflated balloon and the curved detector.

B. In vivo Imaging

During small animal imaging, the rat is first scanned with the endo-rectal WAND coupled with the external surface coil, using the apparatus assembly shown in Fig. 6. To demonstrate the detection sensitivity of the deformable WAND, a series of MR images are then acquired in the coronal plane, using fat suppressed fast-spin-echo (FSE) T₁:-weighted images with the following parameters: TR/TE = 1175.05/35.5 ms, 90° FA, 25.6 × 25.6 mm² FOV, 256 × 256 matrix, 50 kHz bandwidth, slice thickness = 0.8 mm. NA = 4 and TA = 47 s. For comparison, images are also acquired with only the external surface coil using the same acquisition parameters. As shown by the images listed in row 1 (Fig. 10) the external surface coil has limited sensitivity for deep-lying regions with a penetration depth (~2.5 cm) that is larger than the diameter of the external coil (~2.2 cm). This limitation is particularly obvious for fast- spin-echo acquisition when the number of spin echoes reaches 16 in a single acquisition shot. Compared to images in row 1, the images obtained by the active WAND in row 3 have significantly improved detection sensitivity around the kidney region. These images (Figs. 10 D1, E1, F1) show clearly distinguishable vascular boundaries along the renal artery and its junction with the aorta. However, nothing similar is distinguishable from the corresponding images acquired with the external receive coil alone (Figs. 10 A1, B1 C1). In addition, a one dimensional SNR profile plotted along the same horizontal line (yellow dashed line) acquired from the endo-colon WAND (Figs. 10 D2, E2, F2) has ≥ 10-fold gain in normalized intensity compared to the corresponding profiles obtained from the external surface coil (Figs. 10 A2, B2, C2), enabling accurate differentiation of benign and malignant lesions with adequate signal contrast.

Fig. 10:

T₁-weighted longitudinal MR imaging of rat kidneys. A1, B1, C1: Images obtained using external detector only. D1, E1, F1: The same images obtained with the deformable endo-rectal WAND inserted near the kidney region via the rectum before being inflated. A2, B2, C2: The normalized intensity has been plotted for each spot corresponding to each image in row 1 along the horizontal yellow dashed line and highlighted within the vertical red dashed lines. D2, E2, F2: The normalized intensity is plotted with the same region as in row 1 along the yellow dashed line and highlighted within the same region.

IV. DISCUSSION

We have successfully constructed and studied the sensitivity enhancement properties of an expandable balloon catheter WAND. This catheter WAND can enhance detection sensitively by at least 10-fold for regions separated from its surface by a distance larger than its width in the expanded state. The deformable WAND could be non-surgically inserted into the rodent rectum. After navigation into open cavities, the detector can be inflated to sensitively observe deep-lying structures such as the kidney and its surrounding arteries. Moreover, our deformable WAND is integrated with a parametric amplifier that can effectively utilize wirelessly provided pumping power to amplify MR signals in situ. Less than 10 mW power is required on the external signal generator to efficiently operate the WAND at a penetration depth of ~2.5 cm away from the external surface coil. This is the largest depth obtainable on a 250-gm rat to demonstrate the sensitivity advantage of the WAND, when the detection depth is similar as the diameter of the external surface coil. Based on our simulations and measurements, the required power for pumping signal is much smaller than the required power for RF excitation, demonstrating the better safety feature of the WAND over other endoluminal coils with wired connections. For any deeper regions, the WAND can always enhance detection sensitivity. Of course, the required power level will scale up as the detection depth, in a similar way as the power level required for RF excitation.

Unlike previous versions of WANDs [9], [10], [25], [26], where the fixed geometries restrained their capability to sensitively detect relatively larger openings in internal body cavities, this polyimide etched deformable WAND is mounted on a PTA dilatation balloon catheter with a dimension of 7 × 20 mm² that can be conveniently inserted in the target regions via a 3 mm diameter sheath tube when the coil is in its rolled- up configuration. Upon deployment, the detector can be unfolded to at least 6 mm width to sensitively observe relatively larger regions. It creates a symmetric detection profile in the longitudinal direction covering an area of at least 15 × 13 mm² with an extended field-of-view in the axial direction, up to 20 mm. This expandable detector can be used either as an endo-esophageal or an endo-colon WAND to sensitively detect adjacent vascular walls and abdominal organs, such as the kidney. The in vivo application of the WAND as an endo-colon detector is strongly supported by high-resolution kidney images performed on a rat using a 7 T horizontal bore scanner. Compared to the external surface coil, the endo-colon WAND can enhance sensitivity of the arterial region by a factor of ~10 for regions with at least 5 mm separation from its surface. This signifies that the expandable WAND could easily differentiate signals emitted from various parts of the tissues within the organs of interest, leading to better diagnosis and treatment of lesions. It allows us to sensitively observe multiple organs from inside the digestive tract, and to observe important vessels without the need for surgical incisions. In addition to kidney imaging, improving prostate image quality will also be a suitable application for this novel technique.

Even though parametric amplifiers are somewhat noiser [34], [35] than transistor-based amplifiers, the WAND is still much more sensitive than external detectors. Its wireless operation feature will make interventional MRI [36]-[38] safer, thus promoting this important but underutilized application. Its compact design is amenable to further miniaturization. When both the WAND and a cable-coupled internal coil are soaked in water, the WAND is at least 10% more sensitive than the cable coupled internal coil because it can effectively compensate for signal loss. In contrast, there is much SNR loss in the cable traps and along the RF cable soaked inside water. Of course, it is also possible to boost the sensitivity of conventional internal coils with an on-board amplifier. But compared to parametric amplifiers that are compactly integrated into the detection coil, transistor-based amplifiers require a DC power source and biasing circuitry to operate. In addition to its operation convenience, the WAND can also overcome a major challenge when a long RF cable is used inside the body: the inherent risk for excessive local heating along cables during the RF excitation pulse. Although the enhancement and the field homogeneity obtained by the WAND is still limited over a small ROI, we can provide improved sensitivity for larger ROIs by using arrays of WANDs. Active work is being performed along this direction.

V. CONCLUSION

This paper presented a deformable Wireless Amplified NMR Detector (NMR) mounted on the cylindrical surface of a balloon catheter. This novel expandable endo-rectal detector could be used to observe a large range of deep lying regions after it is navigated through the narrow orifices. The WAND’s capability for sensitivity enhancement will be effective in all penetration depths and at all available magnetic field strengths, so the WAND might provide an additional opportunity to achieve comparable performance of the more expensive higher field magnet (such as 7 T) at lower field strength (1.5 T). Although the clinical utility of this configuration needs further demonstration, the outcomes of this study already indicate that the endo-colon deformable WAND might be beneficial not only in precision diagnostics but also in interventional MRI.

Biographies

Roshan Timilsina received the B.S. and M.S. degrees in physics from Tribhuvan University, Kathmandu, Nepal, in 2009 and 2012, respectively. In 2014, he joined the Department of Physics, University of Akron, OH, USA to pursue his international MS degree in physics.

He is currently working toward his Ph.D. degree in biomedical sciences: medical physics at Oakland University. As part of the degree requirement, he is conducting his research in the Department of Radiology, Michigan State University in conjunction with Department of Physics, Oakland University designing and fabricating wireless amplified radio frequency (RF) coils to develop advanced detection sensitivity and accuracy for next generation MRI diagnosis.

Chunqi Qian received his BS degree in chemistry from Nanjing University and his Ph.D. in physical chemistry from the University of California, Berkeley, in 2007. Following postdoctoral trainings at the National High Magnetic Field Laboratory and the National Institutes of Health, he became a principle investigator in the department of Radiology in Michigan State University. He is focusing on the development of advanced imaging technology for biomedical research.