Optothermal profile of an ablation catheter with integrated microcoil for MR-thermometry during Nd:YAG laser interstitial thermal therapies of the liver—An in-vitro experimental and theoretical study

PURPOSE
Flexible microcoils integrated with ablation catheters can improve the temperature accuracy during local MR-thermometry in Nd:YAG laser interstitial thermal therapies. Here, the authors are concerned with obtaining a preliminary confirmation of the clinical utility of the modified catheter. They investigate whether the thin-film substrate and copper tracks of the printed coil inductor affect the symmetry of the thermal profile, and hence of the lesion produced.


METHODS
Transmission spectroscopy in the near infrared was performed to test for the attenuation at 1064 nm through the 25 μm thick Kapton substrate of the microcoil. The radial transmission profile of an infrared high-power, light emitting diode with >80% normalized power at 1064 nm was measured through a cross section of the modified applicator to assess the impact of the copper inductor on the optical profile. The measurements were performed in air, as well as with the applicator surrounded by two types of scattering media; crystals of NaCl and a layer of liver-mimicking gel phantom. A numerical model based on Huygens-Fresnel principle and finite element simulations, using a commercially available package (COMSOL Multiphysics), were employed to compare with the optical measurements. The impact of the modified optical profile on the thermal symmetry was assessed by examining the high resolution microcoil derived thermal maps from a Nd:YAG laser ablation performed on a liver-mimicking gel phantom.


RESULTS
Less than 30% attenuation through the Kapton film was verified. Shadowing behind the copper tracks was observed in air and the measured radial irradiation correlated well with the diffraction pattern calculated numerically using the Huygens-Fresnel principle. Both optical experiments and simulations, demonstrate that shadowing is mitigated by the scattering properties of a turbid medium. The microcoil derived thermal maps at the end of a Nd:YAG laser ablation performed on a gel phantom in a 3 T scanner confirm that the modified irradiation pattern does not disrupt the thermal symmetry, even though, unlike tissue, the gel is minimally scattering.


CONCLUSIONS
The results from this initial assessment indicate that microcoils can be safely integrated with ablation catheters and ensure that the complete necrosis of the liver tumor can still be achieved.


INTRODUCTION
Minimally invasive MR-guided thermal therapies are increasingly used to destroy lesions in the liver and considerable efforts are being made to improve their safety and efficacy. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] Given the difficulty of accurately modeling the thermal distribution due to the absorption of photons during laser interstitial thermal therapies (LITTs) with typically inhomogeneous tumor regions surrounded by a complex network of vasculature, MR-guidance of these operations is of utmost importance.
It is clear that earlier thermometry approaches 16 based on thermocouples, Bragg sensors, or fluorooptic sensors in addition to being invasive, cannot provide the required spatiotemporal monitoring. Compared to other noninvasive thermometry methods such as ultrasound thermometry, MRthermometry can provide high-resolution updates of the spatial thermal profile in close to real-time, while unlike CTthermometry, 17 it is radiation free.
By accurately monitoring the thermal and structural changes that the tissue undergoes during ablation, it is possible to enable dynamic adaptation of the thermal dose and ensure the complete necrosis of the lesion while sparing the surrounding vital structures. One of the main objectives is to increase the resolution and the accuracy of the temperature measurement during MR-thermometry. The thermal information can be then used as a feedback for safely adjusting the thermal dose, the position of the ablation catheter, or for introducing additional catheters.
Microcoils colocated with the ablation catheter have the potential to improve the control of MR-guided thermal therapies. It has been convincingly demonstrated that they are inherently less affected by body noise compared to external coils, [18][19][20] implying that significantly higher signal-to-noiseratio (SNR) and hence increased temperature accuracy are possible within a small field-of-view (FOV), typically sufficient for the lesion dimensions treated. However, there are only limited studies of their effectiveness for local MRthermometry. [21][22][23] In a recent study, 24 we demonstrated that significant improvements in the overall control of LITTs can result from the integration of flexible microcoils 20 with commercially available laser catheters. The results from an Nd:YAG gel-phantom ablation [where the microcoil proton-resonance frequency (PRF) MR-thermometry performance was compared with that of an array coil] indicated that the SNR and temperature accuracy are improved by ten times within a radius of 20 mm.
Here, we consider how the integration of a microcoil with the ablation catheter can affect the irradiation pattern of a diffusive fibre tip, and we examine the thermal implications by assessing the microcoil derived thermal maps of axial thermometry slices, produced during the Nd:YAG laser ablation in Ref. 24. More specifically, we interpret how the measured and simulated optical profile of the modified catheter correlates with the microcoil derived thermal maps. The effects of the microcoil substrate and printed copper inductor are independently assessed.

2.A. Laser applicator and laser system
A Somatex water-cooled laser applicator (Power-Laser Applicator Set, SOMATEX, Teltow, Germany) and a Nd:YAG laser (MY 30, Marting Medizin-Technik, Tuttlingen, Germany) with a wavelength of 1064 nm and 25 W output power were used for the ablation of a liver-mimicking gel phantom. The laser light was delivered through a 12 m length of fibre with a 400 µm diameter core and a 20 mm diffuser active tip (Surgical Laser Technology). The internal circulation cooling system was maintained by a continuous flow of 0.9% NaCl at a rate of 60 ml/min.

2.B. Integration of the microcoil with the ablation catheter
The microcoil receiver is shown in Fig. 1(a). It is based on a copper-clad Kapton thin film (35 µm thick Cu on 25 µm thick Kapton) and consists of a 60 mm long, 4.5 mm wide, two-turn spiral inductor with a line width of 150 µm. A pair of integrated capacitors C T and C M enable tuning and matching, while a subminiature coaxial cable and a non-magnetic pin diode enable active detuning. The details of the modified ablation catheter are shown in Fig. 1(b). A PTFE tube was slid over the applicator and the microcoil, tuned and matched to the loaded condition (at 127.6 MHz), was wrapped around it and held in place with a heat-shrink sleeve, leading to a final diameter of 5 mm. Although this diameter is still sufficient for minimally invasive operations, we are investigating custom made solutions that will further reduce it.

2.C. Gel phantom and experimental setup
The gel phantom with liver-mimicking MR-properties [T 1 = 800 ms and T 2 = 50 ms (Ref. 24)] was prepared by mixing 31.5 g agar with 58 ml n-propanol and 642 ml salt solution, a solution containing 3.37 g/L NiCl 2 .6H 2 O and 2.4 g/L NaCl. To enable absorption at 1064 nm, the gel was doped with 8% India ink.
The complete setup is shown in Fig. 2(a). The ablation phantom has cylindrical symmetry, with the probe cast horizontally in the middle, as shown in Fig. 2(b) which is a photograph of the phantom mould before casting the inkdoped gel. The cylindrical mould, consisting of 11, 9.5 mm thick Perspex slices has been cast inside clear agar gel which emulates body loading [see Fig. 2(a)]. 24 The details of a cross section through the slices that were monitored during the ablation experiment are illustrated in Fig. 2(c).
The details of the two thermometry slices that were monitored, using PRF MR-thermometry, simultaneously during ablation are illustrated on the schematic of Fig. 2 was inserted in each, providing temperature references at two discrete locations [at d 1 = 5 mm (L 1 ) to the probe in slice 1 and at d 2 = 10 mm (L 2 ) in slice 2] during ablation for comparison with the MR-inferred temperatures.
The thermal maps of each slice (containing either L 1 at d 1 = 5 mm or L 2 at d 2 = 10 mm) were derived offline by postprocessing the acquired phase images in ™. The pixel-by-pixel values of the phase differences (∆ϕ) were converted into temperature differences according to Eq. (1). 5 Here, T ref is the temperature recorded with the Luxtron sensors at the beginning of the experiment (baseline), a is the PRF shift coefficient (ppm/ • C), γ/360 • is the gyromagnetic ratio (45. 2 MHz/T), B 0 is the static magnetic field (3 T), and TE is the echo time (3.872 ms). The thermal coefficient a was determined from the linear fit of the phase differences [in the region of interest (ROI) corresponding to L 1 ], plotted against the temperature differences recorded with L 1 . Two non-heated ROIs (10×30 pixels) were used to correct for non-temperature related phase changes (1)

2.E. Near infrared transmission spectroscopy of the Kapton film substrate
The transmission spectrum of the 25 µm thick Kapton substrate was measured in the near infrared (NIR) range to determine the absorption at 1064 nm. The setup of Fig. 3 was employed for this measurement. A broadband light source was used and the excitation wavelength was varied over the range of 600 to 1200 nm in steps of 10 nm, using a computer-controlled monochromator. The transmitted intensity through the film at each wavelength was recorded using a Ge photodetector and a lock-in amplifier as shown in Fig. 3. To determine the absorption, the transmission spectrum through Kapton was normalized to that of air.

2.F. Measured irradiation through the modified catheter
The copper track of the microcoil inductor is 35 µm thick and hence is a perfect conductor at 1064 nm, leading to a region of low optical intensity. The extent of this shadowing effect was assessed by characterizing the radial transmission profile of an infrared high-power, light emitting diode (LED) (M1050F1, Thorlabs) with >80% normalized power at 1064 nm through a cross section of the modified applicator. The LED was coupled to a multimode fiber (MMF), fixed in the F. 3. Experimental set up for near IR transmission spectroscopy of the Kapton film and for the characterization of the absorption coefficient µ α of the ink-doped ablation gel. center of the probe as shown in Fig. 4 and it was driven at a constant current of 50 mA. The diffuse characteristics of the ablation fiber were not emulated in this case, so that the fiber produced a forward propagating collimated and spatially coherent light instead of a diffuse light.
A rotating mechanical shaft, controlled from a PC by means of Labview™ code, allowed for the transmitted light intensity through the probe to be recorded over a full revolution in steps of 2 • using an imaging system and a Ge detector perpendicular to the axis of rotation. To reduce noise due to the background light, the detector output was fed into a lock-in amplifier, which was locked to the frequency of a modulating square wave (3 V square wave, with 1 kHz frequency) that was applied to the LED. The measured light intensity was normalized to the intensity without the microcoil. An optical system, consisting of a pair of magnifying lenses, at d 1 = 6 mm to the fiber and at d 2 = 9 mm to the detector, ensured high spatial resolution.
To determine the extent of shadowing, three measurements of the transmitted light intensity through the modified catheter were performed: (a) in air, (b) with a thin (2 mm) surrounding layer of scattering medium provided by NaCl crystals, and (c) with a thin (2 mm) surrounding layer of the ink-doped gel in order to compare with the thermal profiles obtained using the microcoil MR-data from the ablation. The absorption coefficient µ a of the gel was determined as 0.12 mm −1 using the setup of Fig. 3 and employing the Beer-Lambert law Eq. (2), where z is the thickness in mm of the cuvette containing the sample, I o the intensity of light through the empty cuvette, and I the intensity of light through the cuvette filled with the ink-doped gel (2)

2.G. Huygens-Fresnel diffraction model
The distribution of light around the opaque copper can be modeled using the Huygens-Fresnel principle. Every point on an optical wavefront can be considered as a source of secondary spherical waves, which interfere with each other to produce a new wavefront at a given distance away. Due to the cylindrical symmetry of the geometry, we considered only the profile through the middle of the inductor, and assumed an infinitely thin inductor. The model was implemented in ™ for comparison with the measured results of the radial irradiation pattern obtained in air. The diffuse nature of the ablation fiber was incorporated, by considering a spatially incoherent light source.
Every point x i of the line across the inductor was modeled as a discrete point source of intensity E o (x i ,0) (V/m) with a random phase ϕ i assigned to it. As shown in Fig. 5(a), the intensity E z (ξ k ,0) at a given point d distance away (in the z-axis) was calculated by summing the contributions from all point sources x i which were assumed to have intensity equal to 1 (so that attenuation due to the Kapton film was ignored) with the exception of the conductor tracks, where zero intensity was assumed.
The field intensity at a distance d was calculated according to Eq. (3), where λ = 1064 nm, k = 2π/λ, and f (θ) = [1 + cosθ]/2 accounts for backscattering propagation. 25 When diffuse light was considered, the terms exp (− jϕ i ) in Eq. (3) accounted for a spatially incoherent source and the phase ϕ i of each point source x i was assigned a random value. Otherwise, the phases ϕ i , were set to zero

2.H. Impact of a turbid medium: A multi-parametric study using a full diffraction model
The effect of scattering-a property of a turbid medium such as tissue-on the attenuation of the field intensity behind the conductors was studied through a more realistic planar model of the microcoil developed using a commercially available finite element analysis software package (COMSOL Multiphysics). This allowed the rigorous treatment of diffraction by solving the electromagnetic wave equations for a given 2D geometry of finite thickness, compared to the simpler 1D Huygens-Fresnel model.
The details of the model are illustrated in the schematic of Fig. 5(b). The model shown in Fig. 5(b) is the planar model for one of the inductor tracks as indicated in Fig. 5(a) (dotted line). A plane wave of wavelength of 1064 nm with transverse magnetic (TM) polarization, traveling through a region with refractive index n 1 , was assumed to be incident on a 35 µm thick boundary comprising a 75 µm wide metal obstacle, modeling the half width of one of the four copper tracks and a 75 µm wide dielectric of n 2 = 1. The dielectric is intended to model a region of Kapton film but it is assumed to be air here in order to study the independent effect of the inductor.
A third region of n 3 , below the metal obstacle and dielectric opening, extending up to a depth of 200 µm modeled either a turbid medium or air (n 3 = 1). The incident wave travels through the dielectric opening and it is fully absorbed by a perfectly matched layer (PML) at z = 200 µm. Perfect magnetic conductor boundaries were used on the left and right boundaries to reflect the symmetry and a mesh size smaller than λ/4 was used.
Triangular mesh elements were generated by the built-in meshing function and the maximum mesh size was bounded to λ/4. The upper limit ensures the spatial resolution is fine enough to accurately capture the wave propagation. Mesh convergence studies were carried out to confirm this. Exploiting symmetry allowed significant reduction in the simulation time.
The following four cases were considered: (a) spatially coherent incident wave and uniform external medium (air), (b) spatially coherent wave and turbid medium, (c) diffuse planar wave and uniform medium (air), and (d) diffuse planar wave and turbid medium. The scattering property of the turbid medium was modeled as random spatial variation δn ′ of the refractive index n ′ about a mean value ⟨n ′ ⟩ = 1.4. 26-29 A normal distribution was assumed as codified in Eq. (4) where the variance δn ′2 reflects the degree of scattering.
Absorption was included in the model by introducing a lossy component n ′′ = µ α λ/4π such that n ∼ N(⟨n ′ ⟩, δn ′2 ) + jn ′′ . Finally, for a more realistic model, the diffuse nature of the ablation fiber was incorporated by assuming that the region of n 1 through which the incident wave travels before reaching the microcoil interface is also a scattering one. This additional scattering is modeled in the same way as for the tissue region n 3 n ′ ∼ N(⟨n ′ ⟩, δn ′2 ). (4)

3.A. Symmetry of the thermal profile during ablation
The microcoil derived thermal maps at t = 10 min and at t = 20 min are shown for each of the two thermometry slices in Fig. 6. In all four cases, the thermal maps are perfectly symmetric as expected from the cylindrical symmetry of the phantom and confirm that the integration of the microcoil did not distort the heating profile. Figure 7 shows a plot of the MR-inferred temperatures versus the Luxtron data. The readings of the Luxtron sensors are in good agreement with the microcoil MR-inferred temperatures and the temperature standard deviation less than 0.75 • C up to a radius of 20 mm 24 while the phase noise does not exceed 0.5 • in a 15 × 15 mm ROI centered at the catheter. Worth noting that no artifacts due to the Luxtron sensors were observed, neither did the laser fiber affect the Luxtron readings. In our case, the closest distance of the Luxtron sensor to the laser fiber is 5 mm which is larger than the distance at which others have observed effects. 16

3.B. Transmission spectroscopy of the Kapton film substrate and thermal implications
The transmission spectrum of Kapton between 600 and 1200 nm wavelength is shown in Fig. 8. At 1064 nm, over 70% is transmitted, confirming the expected low absorption of Kapton at the wavelength of interest. 30  To appreciate the possible thermal implications of absorption in the Kapton film, we also examined the thermal maps shortly after the laser was turned on. Figure 9 shows the microcoil derived thermal maps of the thermometry slice containing L 2 in a 23 × 23 mm ROI at t = 2 min and at t = 3 min as well as the corresponding magnitude baseline image that indicates the location of the microcoil on the probe.
At t = 2 min, there is a clear elevation of temperature in the region of the Kapton film compared to the upper part of the probe. This temperature difference leads to an initial asymmetric thermal profile in the immediate vicinity of the catheter. The temperature difference is, however, only equal to 3 • C.
The temperature map one minute later (t = 3 min) indicates that the asymmetry is no longer present and one possible explanation is the effect of the water cooling system which does not allow a large thermal gradient to develop between the two parts of the probe.  Figure 10(b) presents the profiles with the surrounding layers of ink-doped gel and NaCl crystals. All measured profiles have been normalized to the transmitted light through the original catheter.

3.C. Shadowing due to the copper inductor and thermal implications
Since d >> the 35 µm thick copper conductor, our assumption to neglect the finite thickness and assume an infinitely thin inductor is valid. This is also verified by the excellent correlation of the model results in air with F. 8. Near IR transmission spectrum of a 25 µm thick Kapton film.
F. 9. Thermal maps in a 23 × 23 mm ROI of the thermometry slice containing L 2 centered at the probe: (a) at t = 2 min, (b) at t = 3 min, and (c) corresponding magnitude baseline image.
the experimental results (with the exception of the Kapton absorption which is not taken into account). The optical notches around 30 • and 150 • correspond to the copper tracks which as expected block light. Our optical system provided enough resolution to separate the effect of all four tracks of the two-turn spiral coil. Light is blocked over an angle of 30 • at the two regions of the copper tracks and this corresponds to an arc length of 1 mm, assuming the coil is wrapped on a catheter with 4 mm diameter.
The measurement with the thin surrounding layer of inkdoped gel resulted in the same irradiation profile as that obtained in air. Referring to the thermal maps of Fig. 6 (at t = 10 min and t = 20 min) as well as to the thermal maps at the start of the ablation, shown in Figs. 9(a) and 9(b), there is no thermal effect due to the optical notching. This implies that heat conduction must rapidly prevent the development of a thermal gradient.
In the case of the purely scattering surrounding medium, the irradiation profile is very different. Shadowing is effectively countered due to scattering, and this suggests that the ink-doped gel must be minimally scattering. The mechanism by which scattering counters shadowing is further investigated in Sec. 3.D.

3.D. Simulation results: A multi-parametric study using a full diffraction model
The calculated electric field below the metal-dielectric boundary is shown in Fig. 11 for three different conditions of scattering for the region of n 3 , namely δn 3 ′ = 0, 0.05, and 0.1. The figures illustrate the results for the left hand side of the model in Fig. 5(b). The right hand side has been taken into account by imposing appropriate boundary conditions as explained in Sec. 2. Figures 11(b)-11(d) correspond to a spatially coherent plane wave and Figs. 11(e)-11(g) to a diffuse wave, modeled by a standard deviation (std) δn 1 ′ = 0.1. Our model verifies that scattering redistributes photons from the plane wave that traveled in medium 3 through medium 2, resulting in a nonzero field in the region immediately below the metal obstacle. The E-field increases more than twofold in the case of increased scattering while below medium 2, scattering does not affect the value of the E-field. The use of diffuse illumination does not have a significant effect when scattering is present. In air, however, it leads to a slightly increased field away from the metal boundary compared to the coherent source case. It should be noted that this effect could not be captured through the Huygens-Fresnel model, where there appeared to be no dependence in shadowing on the illumination. A plausible explanation could be the different implementation of a diffuse source in the two cases.
To allow more direct comparison with the optical measurements and thermal maps, we also introduced absorption in region 3. We tested two scenarios: (a) µ α = 0.12 mm −1 by setting and n 3 ′′ = 10 −6 (a realistic case) and (b) µ α >> 0.12mm −1 by setting n 3 ′′ = 0.0013. The mechanism of scattering in countering shadowing is not affected in the lowloss case (µ α = 0.12 mm −1 ). With significant absorption on the other hand, scattering fails to counter the shadowing below the metal. Absorption seems to be the dominant effect in determining the value of the E-field in that case. These results correlate extremely well with the optical measurements. Although the thermal maps from the gel phantom ablation did not indicate any thermal effect due to the optical notching we observed in the transmission measurements, the simulations further ensure that the shadowing will be minimized due to the scattering properties of tissue.

DISCUSSION
We have studied the independent and combined impact of the Kapton substrate and copper tracks on the irradiation pattern of the modified catheter. We have verified through a liver-mimicking phantom Nd:YAG laser ablation, optical measurements, and electromagnetic simulations that the modification of the catheter does not affect the thermal profile. Absorption due to the thin film substrate is low and leads to insignificant heating and the shadowing caused by the conductor tracks is countered by tissue scattering, which acts to redistribute the light.
Despite the simplistic 1D Huygens-Fresnel model, which assumed an infinitely thin coil structure, the shadowing effect behind the copper tracks was captured fairly accurately and the results of the model correlated really well with the experimental results [see Fig. 10(a)] which accounted for the physical structure of the microcoil. The microcoil thickness was considered in the more complete 2D electromagnetic model of diffraction we used to simulate the effects of scattering and absorption of a turbid medium on a plane electromagnetic wave propagating through the microcoil into the turbid medium.
We modeled tissue scattering as a spatial variation of the refractive index as previously proposed in Refs. 26-29, while liver tissue absorption was modeled as a lossy component of the refractive index, based on the measured value of the absorption coefficient of the liver-mimicking gel phantom. Although these parameters are semi-empirical, they have provided reasonable qualitative answers which were further validated by experimental results.
The excellent sensitivity of the microcoil (see Fig. 7) allowed for an accurate assessment of the thermal profile of the modified catheter without the need for a thermal model. The symmetry of the thermal maps, where the effects of substrate absorption and track obstruction are combined, indicates that any small remaining inhomogeneity of irradiation is most likely corrected by the heat conduction under the operation of the water cooling system of the applicator.
Although more thorough testing will be required using porcine liver, our MR-data based on the gel ablation indicate that Kapton absorption will not compromise tissue necrosis due to exposure to different temperatures neither will it pose a safety risk due to overheating. In the unlikely case of a failure of the water-cooling system during an ablation, undesirable effects such as tissue charring, significantly reduced lesion size, and potentially serious patient safety risk are expected weather a microcoil has been integrated or not with the catheter. 32,33

CONCLUSION
Microcoils, which can easily be integrated with ablation catheters, can improve the control of laser interstitial thermal therapies, and lead to excellent signal-to-noise ratio and temperature accuracy during local MR-thermometry within a radius matching typical lesion dimensions. Here, we have considered the additional effects of microcoil integration on the optothermal performance of the modified catheter. The results of this study provide a preliminary confirmation of the clinical utility of the modified catheter in liver LITTs.