Improving image quality in transcranial magnetic resonance guided focused ultrasound using a conductive screen

Abstract

Transcranial Magnetic Resonance guided Focused Ultrasound (TcMRgFUS) has been proven to be an effective treatment for some neurological disorders such as essential and Parkinson’s tremor. However, magnetic resonance guidance at 3 Tesla (3T) frequencies and using the large hemispherical transducers required for TcMRgFUS results in artifactual low-signal bands that pass through key regions of the brain. The purpose of this work was to investigate the use of a circular conductive Radio Frequency (RF) screen, that is bent to have a 12 cm radius in one direction and positioned near the top or back of the head, to reduce or remove these artifactual low-signal bands in TcMRgFUS.

The impact of using an RF screen to remove these low signal bands was studied in both imaging experiments and electromagnetic simulations. Hydrophone measurements of the acoustic transparency of the bronze 2 mm diameter square mesh screen used in the imaging studies were compared with temperature measurements with and without the screen in heating studies in the TcMRgFUS system.

The imaging and simulation studies both show that for the different screen configurations studied in this work, RF screen removes the low-signal bands and increases both homogeneity and signal-to-noise ratio (SNR) throughout the region of the brain. Hydrophone and heating studies indicate that even a 2 mm wire mesh provides minimal attenuation to the ultrasound beam. Simulation results also suggest that a 1 cm mesh will provide adequate artifact suppression with even less ultrasound attenuation.

An RF screen that disrupts the natural waveguide nature of the transducer in the 3T MR environment can change the electromagnetic field profile to reduce unwanted artifacts and provide an imaging region which has more homogeneity and higher SNR throughout the brain.

1. Introduction

Transcranial Magnetic Resonance guided Focused Ultrasound (TcMRgFUS) has been shown to be an effective method for treating some neurological disorders such as essential and Parkinsonian tremor [1–6]. New research is also demonstrating the potential of TcMRgFUS for opening of the blood brain barrier and neuromodulation [7–10]. MR guidance includes pre-treatment anatomy imaging and post-treatment assessment as well as MR thermometry and anatomical imaging during treatment. Treatment monitoring is enhanced with high signal-to-noise ratio (SNR) for improved spatial and temporal image resolution and improved temperature measurement precision.

The only commercially available TcMRgFUS device that has received regulatory approval for clinical use is the Exablate Neuro System (Insightec, Inc., Haifa, Israel), which consists of a large 30 cm diameter hemispherical transducer with an electrically conducting ground plane and 1024 ultrasound elements that are focused on the center point of the spherical volume (see Fig. 1). The space between the head and transducer is filled with degassed, de-ionized water that functions as an ultrasound coupling medium. The water is cooled and circulated during treatments to prevent high temperatures on the surface of the skin.

Fig. 1.

The Insightec Exablate Neuro Transcranial Transducer. This transducer is a hemispherical transducer that is 30 cm in diameter, with a conducting ground plane and 1024 transducer elements operating at 650 kHz. The ultrasound elements have a natural focus at the center of the open-end plane of the transducer. It is positioned in a 3T MRI scanner for MR guidance.

It has been hypothesized that the large conducting ground plane of the transducer can have standing wave or waveguide-effects on the gradient and Radio Frequency (RF) magnetic fields of the MR scanner [11]. Further, the linear dimensions of the water volume being slightly larger than the 28 cm RF wavelength in water at 3T, are large enough to support standing waves in the Exablate system. When using the body RF coil for transmit and receive during imaging in the TcMRgFUS system, the waveguide and standing wave effects combine to cause artifactual low-signal bands that significantly reduce the SNR in bands across the brain, as seen in Fig. 2. These undesirable artifacts can reduce the visibility of the anatomy during TcMRgFUS treatments and reduce temperature measurement precision in MR thermometry images.

Fig. 2.

Transcranial transducer low-signal banding artifacts. These artifacts are apparent when using the MR body coil for both transmit and receive making treatment planning difficult and preventing MR thermometry at the surface of the skin and throughout the volume of the brain. A) Artifacts shown in typical clinical treatment planning images. B) Artifacts shown in the skull phantom imaging studies of this work, showing the low signal sensitivity near the focus of the transducer.

We hypothesized that these banding artifacts could be shifted by the insertion of an RF conductive screen in the space between the transducer and the patient’s head [12]. To become a viable solution for the TcMRgFUS system, the screen would need to effectively reflect the RF field, be transparent to the 650 kHz ultrasound fields and not restrict water flow within the transducer. Yan et al. have recently addressed this challenge using a crossed wire solution which becomes a small reflecting antenna within the water bath [11], and other investigators have tried using local transmit and receive coils to reduce the banding artifacts [13–16].

This work investigates the use of an RF conductive screen within the transducer to reduce the artifactual low-signal bands in the MRI images of the TcMRgFUS system. The objective was to modify the distribution of the B1-transmit and receive electromagnetic fields produced by the system body coil inside the TcMRgFUS transducer to improve the field homogeneity throughout the region of the brain and eliminate the banding artifacts in the MR images. Imaging studies, bench tests and electromagnetic simulations were performed to quantify the screen’s impact on image quality and its effects on the ultrasound beam and the water flow required for successful TcMRgFUS treatments.

2. Theory

The SNR of any coil or coil array used in MRI is spatially dependent and is a direct function of the $B_1^{+}$ transmit and $B_1^{-}$ receive sensitivity profiles of the RF coil or coil array. The transverse magnetization from which the signal is generated, is a direct function of the $B_1^{+}$ transmit field [17]:

$${\text{m}}^{+}\propto Wsin\left(V_{scale}\left|B_1^{+}\right|\gamma \tau \right)$$ (1)

The received signal is directly proportional to $\left|B_1^{-}\right|$:

$$\begin{gathered} \hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}S{I}_{calc}\left(x,y,z\right)\propto {\text{m}}^{+}\left|B_1^{-}\left(x,y,z\right)\right| \\ \propto W\left(x,y,z\right)sin\left(V_{scale}\left|B_1^{+}\left(x,y,z\right)\right|\gamma \tau \right)\left|B_1^{-}\left(x,y,z\right)\right|, \end{gathered}$$ (2)

where m⁺ is the transverse component of the excited magnetization, W is the proton density, V_scale is the RF coil excitation voltage gain, γ is the proton gyromagnetic ratio, and τ is the MR excitation pulse duration, respectively. The relative Signal Intensity (SI_calc) is a $B_1^{+}$ related scaling of the receive coil sensitivity profile $\left|B_1^{-}\right|$.

Because the transmit and receive portions of the pulse sequence do not occur simultaneously, their effects are considered separately. Changing the gain of the transmit coil directly affects the tip angle of the magnetization and ultimately the amount of signal generating transverse magnetization.

Although changing the gain of the receive coil can change the received signal amplitude, the noise voltage in the receive coil is a function of the entire distribution of the coil sensitivity. Because of these relationships, simply adding gain where the entire coil sensitivity is increased, will not change the SNR. However, local change in the coil sensitivity distribution can affect the SNR.

From Roemer [18], and assuming sample noise is the dominant source of noise:

$$R_{tot}=\sum _{j,k=1}^N{g}_j{g}_k{R}_{jk}$$ (3)

$$R_{jk}=\sigma \int E_j\left(x,y,z\right)•E_k\left(x,y,z\right)dV$$ (4)

where g_j is the complex gain of coil j, R is the equivalent resistance, representing energy loss and MR noise, and E_j and E_k are the electric field induced by a unit current in coils j and k, respectively. These electric fields are linked to the $B_1^{-}$ sensitivity profile through Faraday’s Law. The total noise in the MR image is integrated over the entire sensitive volume of the sample.

The banding artifact is a function of both the $B_1^{+}$ of the transmit coil array and the $B_1^{-}$ of the receive coil. When the OEM body coil is used for both transmit and receive coils, the banding artifact will be related to both $B_1^{+}$ and $B_1^{-}$ which have very similar distributions. Using any passive or active method to change the distribution of the receiver sensitivity can change the local SNR.

3. Methods

3.1. Imaging studies

The RF screen used for the imaging and hydrophone studies was a 24 cm diameter circular disk cut from a large sheet of wire mesh screen. The screen was made from un-insulated bronze wire that was approximately 0.25 mm in diameter. The wires formed small mesh squares that were approximately 2 mm in diameter. The screen was bent slightly to have an approximately 12 cm radius and was held in position with string and small rubber spacers that were used to isolate the screen from the skull as shown in Fig. 3.

Fig. 3.

Radio Frequency screen used to mitigate artifacts in the transducer images. The screen in this image is in Position #2, over the top of the head.

The phantom for the imaging studies consisted of a human skull, with a vendor-supplied Daily Quality Assurance (DQA) phantom (Insightec, Inc., Haifa, Israel) mounted inside. It was imaged in the Exablate Transcranial Transducer (650 kHz) with a 2D Gradient Recalled Echo (GRE) pulse sequence (TE/TR = 4.03/123 ms, FOV = 350 mm, 320 × 320 acquisition matrix, slice thickness = 5 mm, BW = 220 Hz/pixel) using a 3T MRI scanner (Skyra, Siemens Medical Solutions, Erlangen, Germany) with the OEM body coil for transmit and receive.

Imaging was done with and without the RF screen positioned over the top of the head as shown in Fig. 3. Imaging studies were performed with two different screen positions (see Fig. 4). The first study was done using the screen in position #1, where the screen was positioned toward the back of the skull. The second imaging study was done with the screen in position #2, with the screen on top of the skull.

Fig. 4.

RF screen imaging results for the screen in two different positions. For position #1, the screen was positioned over the back of the head (top row). For position #2, the screen was positioned over the top of the head (bottom row). For both screen positions, there was an increase in signal-to-noise ratio at the focal point of the transducer and throughout the volume of the brain. These results demonstrate the variability in screen position that can help to reduce the low-signal banding artifacts.

Heating studies in the TcMRgFUS system were performed with and without the screen in place. For each situation, five different heating runs at 200 acoustic Watts for 30 s were performed. MR thermometry was performed using a 3D GRE segmented EPI sequence (TR/TE = 29/13 ms, FOV = 280 × 280 × 30 mm, Resolution = 1.1 × 2.2 × 3 mm, ETL = 7, BW = 930 Hz/pixel) [19]. To evaluate the interference of the screen with the ultrasound field, the maximum temperature rise and the full width at half maximum (FWHM) of the temperature distributions were measured. ANOVA tests were performed to investigate if there was any significant difference between the temperature rise and FWHM without and with the screen in place.

3.2. Hydrophone studies

To assess acoustic transparency, pressure field measurements were made using a scanning hydrophone. Pressure measurements were obtained using a 256 element semi-rectangular ultrasound transducer (f = 940 kHz, 14.4 × 9.8 cm aperture, 1.8 × 2.5 × 10.9 mm pressure focal size in water, Image Guided Therapy, Pessac, France, and Imasonic, Voray-sur-l’Ognon, France) and an HNR0500 Hydrophone (Onda Corporation, Sunnyvale, CA, USA) positioned at the acoustic focus. A flat screen was positioned in the hydrophone tank, perpendicular to the propagating beam in the near-field of the acoustic path, at a distance of 2.5, 4.5 and 6.5 cm from the focus. A fourth set of measurements were obtained with the screen at a 45° angle to the beam propagation, positioned in the near-field of the acoustic path at approximately 4.5 cm proximal to the focus. The magnitude of the acoustic pressure at the focus for each screen configuration was compared to that of a water only environment.

3.3. Simulation studies

Two different screens were used for the simulation studies, a solid copper screen and a 1 cm mesh screen. The solid screen was constructed from a solid copper sheet, representing the tight 2 mm mesh bronze screen used in the imaging studies above. The solid screens were modeled as circular copper disks with a 20 cm diameter and 0.3 mm thickness. The wire mesh screen also had a 20 cm diameter, but was made of 0.255 mm diameter copper wires forming a square mesh of 1 cm diameter. These screens were bent in a coronal plane with an approximately 12 cm radius of curvature and were positioned 1 cm from the top of the head and 5 cm from the transducer to mimic position #2 in the imaging studies.

Computer simulations were performed to examine the electromagnetic field behavior and artifacts of the TcMRgFUS system with and without the use of an RF screen. Simulations were performed using commercially available 3D electromagnetic field simulation software (CST Microwave Studio Suite® 2018, Simulia, Providence, Rhode Island). The model consisted of a 16-rung birdcage coil that was 70 cm in diameter and 120 cm long. The birdcage was excited in a lowpass configuration using 16 voltage sources that were each centered on one of the rungs of the coil. Each excitation port was excited with a 1 V sinusoidal source and phase adjusted to drive the birdcage with a circularly polarized current distribution at 123 MHz. The ultrasound transducer ground plane was simulated as a 30 cm diameter solid copper hemispherical dome with the focal point of the transducer concentric with the center point of the birdcage coil. The inside of the dome was filled with distilled water having a dielectric constant of 78.4 and an electrical conductivity of 5.55 × 10⁻⁶ S/m. A human head, neck and partial shoulder model (Gustav, from the CST voxel family with a 2.08 × 2.08 × 2.00 mm resolution) was centered in the birdcage coil with the head positioned inside the transducer with a distance of 6 cm between the top of the head and the top of the transducer as seen in Fig. 6B.

Fig. 6.

Coronal (top row) and Sagittal (bottom row) SI_calc maps for the different simulation cases. The no screen case shows the low signal sensitivity band through the brain and near the focal point of the transducer. For each simulated screen, the signal sensitivity and homogeneity through the brain is increased.

Four different screen configuration cases were simulated and compared to better understand the effects of using an RF screen during TcMRgFUS treatments. Case #1 did not include an RF screen (no screen) and was used as a comparison reference. Case #2 included the solid RF screen (solid) and was simulated because of its effective complete RF screening behavior. Case #3 emulated the more common clinical case for essential tremor treatments where the patients head is not centered in the transducer. The solid RF screen was utilized, and the human model and RF screen remained centered in the birdcage coil, but the transducer ground plane was shifted 2 cm to the left side of the patient head (solid (shifted)). Case #4 simulated the 1 cm wire mesh screen (mesh) that was simulated to assess the performance of an RF screen that should be similar to the solid screen in RF shielding performance while providing more transparency to the ultrasound beam and water flow.

The CST time domain solver was used to compute the $B_1^{+}$ and $B_1^{-}$ for each simulated case. The voltage scaling factor required at the excitation ports of the birdcage coil to achieve a 20° flip angle at the ultrasound focus point of the transducer was determined using the same methods and equations presented by Collins [17], as shown in Eq. 5.

$$V_{scale}=\frac{\alpha }{\gamma \tau \left|B{1}^{+}\left(focus\right)\right|}.$$ (5)

We used a flip angle (α) of 20° and a τ of 2 ms to correspond to pulse sequences typically used for MR thermometry. Assuming the proton density throughout the sample was essentially uniform, the scaled $\left|B_1^{+}\right|$ was used to find the relative signal sensitivity (SI_calc) using Eq. 2.

The CST solver was also used to compute the Specific Absorption Rate (SAR) for 1 g and 10 g SAR for the different cases. The cell size for these CST simulations ranged from approximately 0.25 mm to 2.4 cm.

4. Results

4.1. Imaging results

The imaging results shown in Fig. 4 demonstrate that the screen in both positions moves the banding artifact outside the brain, specifically in the region occupied by the DQA phantom. There was improved signal homogeneity throughout the brain region with the screen in either position. With this improved homogeneity, the SNR was improved on the order of 50% in regions of no artifact and by factors of 5× and more in regions of artifact.

The mean and standard deviation of the temperatures and temperature profile widths from the heating studies are shown in Fig. 5. The 2-way ANOVA test showed no significant difference in maximum temperature rise or FWHM between the screen and no-screen conditions, for both screen positions.

Fig. 5.

3D Transducer Heating Measurement results with the RF screen in positions #1 and #2. For both positions, the temperature heating profiles with the screens were very similar to those without the screen showing no noticeable screen interference. Full-width at half-max measurements were generally more narrow with the screen than the no screen configuration. Note: Not all the transducer ultrasound elements were required to shoot through the RF screen.

4.2. Hydrophone results

The transmission measurements through the 2 mm bronze screen are summarized in Table 1. For all cases, the screen caused an approximate 6% decrease of pressure at the transducer focal point for the different flat and angled configurations when compared to the water-only hydrophone measurement.

Table 1.

Hydrophone measurement results showing the percent attenuation difference between studies with and without the bronze screen in the ultrasound beam pathway. The screen was positioned orthogonal to the ultrasound beam propagation at 3 different distances from the transducer focal point as well as on a 45° angle with respect to the direction of beam propagation.

Intensity attenuation compared to no screen
Screen distance from transducer focus
6.5 cm	4.5 cm	2.5 cm	45° angle
−6.14%	−5.75%	−5.96%	−6.22%

4.3. Simulation results

Simulation results for the different RF screen configurations can be seen in Figs. 6–8. The SI_calc maps given in Fig. 6A and E show the low-signal sensitivity banding artifact that occurs in the transducer when no RF screen is used. Fig. 6B and F show the improved homogeneity and signal sensitivity that occurs in the region of the head with the use of a solid RF screen. Fig. 6C and G demonstrate the effects of using the solid RF screen while the transducer ground plane is shifted 2 cm to the patient left. Although the field between the transducer and the screen is greatly increased, the field within the brain is not substantially affected. Finally, Fig. 6D and H show that the 1 cm mesh screen performs nearly as well as the centered solid conductor RF screen.

Fig. 8.

Specific Absorption Rate maps for the different screen configurations. These maps show an increase in SAR with use of the solid screen and a slight decrease in SAR with the use of the mesh screen.

Fig. 7 displays plots of SI_calc along the central axis of the birdcage coil for the four simulated cases, again showing the improved signal sensitivity profile through the volume of the brain and at the focal point of the transducer with the use of an RF screen. For the shifted transducer case, both the profile through central axis of the birdcage coil and the profile through the transducer focus are shown. The voltage scale factors required to achieve the 20° flip angle at the focal point of the transducer for each case are given in Table 2, indicating the relative efficiency of achieving that flip angle. As the transmit efficiency is proportional to the receive signal sensitivity, these scale factors are also indicative of the gains that might be obtained in SNR and temperature measurement precision when imaging at the level of the focal spot of the transducer.

Fig. 7.

SI_calc profiles for the different simulated cases showing substantial increases in signal sensitivity at the focal point of the transducer and throughout the brain with use of an RF screen.

Table 2.

V_scale voltage gains applied to achieve a 20° flip angle at the transducer focus in the CST simulations, see Eq. 5.

Vscale values applied to achieve 20° flip at transducer focus
No screen	Solid	Solid (shift, focus)	Mesh
200.95	129.03	157.47	134.29

Note: V_scale for the 20° flip angle at the focus of the solid(shift) case is substantially more than it would be if the 20° flip were desired at the center of the birdcage coil as for all the other cases.

Fig. 8 shows the relative SAR maps in three orthogonal views for the different simulated cases based on 1 W input power. The solid screen increases SAR around the skull and skin relative to the no screen case. The 1 cm mesh results in a decreased SAR. All maximum SAR results are presented in Table 3.

Table 3.

SAR results using 1 W input power to the birdcage coil in the CST simulations for the different cases of this work. For each case, the maximum single point SAR was essentially at the same position. These results show that the solid (shift) case has a maximum SAR that is approximately 2× more than the no screen SAR.

SAR results for 1 W input power (W/kg)
	No screen	Solid	Solid (shifted)	Mesh
Total SAR	3.32e-5	3.57e-5	4.37e-5	1.57e-5
max 1 g SAR	1.44e-3	1.83e-3	3.41e-3	0.86e-3
max 10 g SAR	0.51e-3	0.65e-3	1.17e-3	0.62e-3

5. Discussion

This work tested the effects of an RF screen, positioned over the top of the head, on the banding artifact that occurs in the 3T MR images of TcMRgFUS systems with a hemispherical ultrasound transducer. It also tested the effect of the screen on the acoustic field at the focal region of the ultrasound beam, and on the RF power deposition in the skull and brain.

The experiment and simulation results demonstrate that an RF screen positioned between the head and the ground plane of the TcMRgFUS transducer can cause a substantial change in the electromagnetic field behavior inside the transducer volume. This change does not completely eliminate the typical banding artifacts, but it shifts them out of the brain. Results from the imaging studies show that use of the screen positioned either superior to the head or tilted toward posterior, resulted in an increased and more uniform SNR at the transducer focus and throughout the brain than without a screen. Because temperature precision is directly proportional to image SNR, a corresponding improvement in temperature precision and uniformity in precision throughout the brain can be expected when using a screen. These improvements in SNR and homogeneity may provide more safe and informative patient treatment monitoring by enabling 3D temperature measurements throughout the brain and at the skin surface.

Numerical simulations using CST were verified by the experimental observations. The relative changes in simulated $B_1^{-}$ are an indicator of the increases in signal sensitivity that would be achieved using a screen with the TcMRgFUS ground plane. In understanding the effects of the screen on SNR, it is not likely that the presence or type of RF screen will have much effect on the noise power received by the system body coil. This is because the integral in Eq. 4 is over the entire sensitive volume of the receive coil and includes a much larger portion of the body than has been included in these simulations. The square root of the power loss calculated in CST represents image noise generated from the region simulated (water volume, neck and shoulders). Using this measure, the noise contributions for these different cases, solid, solid (shifted), and mesh were only slightly different (−5.5%, +10%, −1.7%) from that of the no screen case. If the simulation had covered the entire body and had included coil losses, the differences in total power loss or noise for each case would be even more negligible. Therefore, given the use of the large transmit and receive birdcage coil, the SI_calc maps are good approximations for relative SNR in these different cases. Similar to the imaging studies, the simulated SNR did increase significantly with use of the screen.

The phantom heating experiments and hydrophone measurements both indicate that the screen has very little impact on the transmitted ultrasound intensity. Fig. 5 demonstrates that there was negligible change in the peak temperature when the ultrasound beam was required to pass through the screen. In addition, the measured temperature distribution (FWHM) was slightly reduced with the screen in place as seen in Fig. 5. The hydrophone measurements indicate that the heating through the screen should be reduced slightly (about 6%) compared to heating without the screen. However, it should be noted that the hydrophone study was performed with a different transducer having a smaller aperture and a higher frequency and the entire ultrasound beam was required to pass through the screen where the attenuation was likely greater due to the shorter wavelength. In the TcMRgFUS system, due to the skull and screen orientation in the transducer, the ultrasound beam from some elements did not pass through the screen in the heating studies. This and the lower frequency (longer wavelength) likely resulted in the reduced impact of the screen on ultrasound delivery during TcMRgFUS treatment compared to that measured in the hydrophone studies. The 1 cm wire mesh screen should have even less interference than the bronze screen used in the imaging studies.

The simulations using a solid screen and a shifted transducer ground plane indicate that better signal sensitivity at the focus of the transducer would result if the screen position were fixed with respect to the transducer requiring both to move over the patient head if needed.

Finally, estimates of the heat deposition (SAR) were obtained using numerical simulations of the electromagnetic fields with and without the screen in place, coupled with a reasonable estimate of the electromagnetic properties of the body. These simulations provide evidence that a solid RF screen can increase the SAR deposition compared to no screen. The SAR was further increased when the transducer was shifted 2 cm laterally. However, it was observed that use of a 1 cm mesh screen could actually decrease energy deposition in the skull and brain over that observed with no screen. The actual energy deposition and tissue heating will always depend on the pulse sequence used, so the scanner software usually places limits on pulse sequence parameters to ensure that allowed SAR limits are not exceeded. If SAR hot-spots occur locally, the scanner software may not protect adequately and this should be studied. At the same time, it appears that the 1 cm mesh could actually decrease the SAR over the case of no screen, improving safety of TcMRgFUS procedures.

We note that the transmit $B_1^{+}$ sensitivity can also adversely affect SAR distribution. A low $B_1^{+}$ at the desired image point can result in an increased transmit gain to achieve the desired flip angle at that position and increase SAR everywhere.

5.1. Limitations

The RF screens used in the simulations were chosen to cover the range of interest from a perfect RF reflector (solid screen) to a more practical reflector that didn’t impede ultrasound and water flow (1 cm mesh). The 2 mm mesh used in the experiments falls in this range. The screens were not optimized for size, shape or position. They were shaped so that the screen wrapped around the head with enough space between the head and screen to prevent any sort of RF or ultrasound burn, that might occur. Further work needs to be done to evaluate the relative SNR performance of different screen shapes and sizes as well as the effect of these screens on ultrasound transmission and treatment efficacy. Also, comparison should be made against other potential forms of B1 field modification [11]. Non-heating studies need to be performed using screens of different hole size on human volunteers in a clinical treatment transducer configuration for better measures of banding artifact elimination and SNR performance.

The banding artifact of the simulations is not identical to the artifacts seen in the imaging experiments. This is due to a number of factors, including the extra water bath that surrounds the phantom in the imaging experiments, a small difference in the size of the screens, and screen positions with respect to the transducer. We observed (not shown) that the imaging artifacts vary slightly with the shape and size of the water bath. However, in the two experimental screen positions, the screen changes the RF field behavior, improving homogeneity and signal sensitivity throughout the volume of the brain. In addition, these simulations used a solid copper conducting ground plane for the ultrasound transducer. Segmenting the ground plane into smaller conductive sections and adjusting the conductivity of the simulated ground plane may provide better correlation between simulated and measured results.

6. Conclusion

In conclusion, for the various imaging parameters of this study, the conductive screen: 1) provided substantial improvement in field homogeneity throughout the region of the brain by eliminating or moving the banding artifacts; 2) increased SNR by a factor of 1.5 to 6 over the region of the brain, and 3) was essentially transparent to the ultrasound beam. We hypothesize that these improvements occurred because the RF screen changed the field boundary conditions and modified the waveguide nature of the transducer ground plane. Future work will more fully optimize and characterize the RF screen functionality.