Preclinical evaluation of a low-frequency transcranial MRI-guided focused ultrasound system in a primate model

Abstract

This study investigated thermal ablation and skull-induced heating with a 230 kHz transcranial MRI-guided focused ultrasound (TcMRgFUS) system in nonhuman primates. We evaluated real-time acoustic feedback and aimed to understand whether cavitation contributed to the heating and the lesion formation. In four macaques, we sonicated thalamic targets at acoustic powers of 34-560 W (896-7590 J). Tissue effects evaluated with MRI and histology were compared to MRI-based temperature and thermal dose measurements, acoustic emissions recorded during the experiments, and acoustic and thermal simulations. Peak temperatures ranged from 46 to 57 °C, and lesions were produced in 5/8 sonicated targets. A linear relationship was observed between the applied acoustic energy and both the focal and brain surface heating. Thermal dose thresholds were 15-50 cumulative equivalent minutes at 43 °C, similar to prior studies at higher frequencies. Histology was also consistent with earlier studies of thermal effects in the brain. The system successfully controlled the power level and maintained a low level of cavitation activity. Increased acoustic emissions observed in 3/4 animals occurred when the focal temperature rise exceeded approximately 16 °C. Thresholds for thermally-significant subharmonic and wideband emissions were 129 and 140 W, respectively, corresponding to estimated pressure amplitudes of 2.1 and 2.2 MPa. Simulated focal heating was consistent with the measurements for sonications without thermally-significant acoustic emissions; otherwise it was consistently lower than the measurements. Overall, these results suggest that the lesions were produced by thermal mechanisms. The detected acoustic emissions, however, and their association with heating suggest that cavitation might have contributed to the focal heating. Compared to earlier work with a 670 kHz TcMRgFUS system, the brain surface heating was substantially reduced and the focal heating was higher with this 230 kHz system, suggesting that a reduced frequency can increase the treatment envelope for TcMRgFUS and potentially reduce the risk of skull heating.

Figure 1

Coronal T2-weighted image of a macaque monkey superimposed on a diagram of the experimental setup. The animal (Monkey 2) was placed on its back with its head tilted back so that it was partially submerged in temperature-controlled degassed water. The position of the geometric focus is indicated by the green “x”. The FUS beam was steered to the thalamic target using the phased array transducer. Acoustic emissions were recorded during sonication using two PCDs mounted in the beam path as well as by detectors integrated into the hemisphere transducer (not shown). The inset shows an axial contrast-enhanced T1-weighted image after creating a lesion in the thalamus.

Figure 2

Speed of sound (−) and absorption (--) values used for the compressional wave propagation in the skull (36) as a function of density.

Figure 3

Temperature rise at the focal point (A) and on the brain surface (B) as a function of the applied acoustic energy and energy density, respectively. The relationship between the temperature rise and the exposure levels was similar for Monkeys 1-3. Less heating was observed for Monkey 4, so it was examined separately. A linear relationship was observed for focal heating, and this relationship was consistent for MRTI obtained in both coronal (open symbols) and sagittal planes (filled symbols). Such a relationship was only evident for sagittal MRTI for brain surface heating, and coronal measurements are not shown in (B). For these measurements, the brain surface was segmented, and the hottest 5% voxels were identified (mean ± SD shown).

Figure 4

MRTI (top) and post-FUS imaging (bottom) obtained in two sessions in Monkey 1. Thermal dose contours at 18 (orange) and 240 (red) CEM43°C were calculated from the MRTI. The plot shows the peak temperature rise as a function of time at the focus for these two sonications (solid line: hottest voxel; dotted line: mean of 3×3 voxel region). Immediately after each session, a small lesion was observed in contrast-enhanced T1-weighted MRI (CE-T1WI). The dimensions of these areas were consistent with the 240 CEM43 contours at this time. At week 2, the lesion produced in the first session was largely non-enhancing in in CE-T1WI, and it appeared in T2-weighted imaging (T2WI) with an increased size that was consistent with the dimensions of the 18 CEM43°C contour. Nineteen months later, only small regions hyperintense in T2-weighted imaging remained. Bar: 1 cm

Figure 5

MRTI (A) and post-FUS imaging (B–D) obtained in Monkey 4. A thermal dose contour at 18 CEM43°C was calculated from the MRTI and is shown in orange. Immediately after the session, a tiny spot of BBB disruption was observed in contrast-enhanced T1-weighted MRI. A lesion was not observed in T2-weighted imaging at that time. This animal was euthanized three days after FUS, and the formalin-fixed brain was imaged post mortem at 7T. The lesion was clearly visible at that time in T2-weighted MRI. Superposition of the 18 CEM43°C thermal dose contour on this lesion [inset in (D)] was consistent with the size of the MRI-evident lesion. Bar: 1 cm.

Figure 6

Post-FUS contrast-enhanced MRI showing a small region with damage next to the skull after the sonications in the second session with Monkey 1. Left: Axial; Right: Coronal. Bar: 1 cm

Figure 7

Power control via acoustic emissions recordings. The signal around the subharmonic (115 kHz) was used to modulate the power level in real time. The system aimed to keep the emissions at a safe level and halted the sonication if the total emissions exceeded a setpoint. In the example shown in (A), the emissions did not exceed this setpoint. In the example shown in (B), the sonication was aborted due to excessive cavitation. (C): Emission spectra from the second example obtained using a PCD integrated into the transducer. This detector had a resonant frequency near the subharmonic; higher frequencies were filtered out. The gray bands indicate the frequency ranges analyzed here for subharmonic and wideband emissions. The black spectrum was obtained at the start of the sonication. The red shows the spectrum with the greatest emission. (D) Emission spectra at these two times obtained using a custom PCD. Here the gray band shows the resonant frequency of the detector, which was 610 ± 10 kHz. Note the presence of higher harmonics and ultraharmonics.

Figure 8

Acoustic emissions magnitude as a function of the peak temperature at the focus, the acoustic power, and the acoustic energy. A clear relationship between focal heating and both subharmonic and wideband emissions was observed, and this relationship was evident for all animals. There was also a relationship between emissions and the acoustic power, except for in Monkey 4 where enhanced signals were never observed. The different symbol shapes represent measurements obtained during the different sessions. The sigmoid curves represents the probability for having thermally-significant emissions, which was defined here as having signal higher than three times the average signal from all the sonications when the temperature rise was less than 10°C. This threshold is indicated by a dashed line. This probability was estimated using probit regression; the shaded areas are 95% confidence intervals of this fit. The point at which the probability was 50% is indicated by “+” symbols.

Figure 9

Measured and simulated sagittal temperature maps. Left: Maximum temperature map obtained with MRTI for a sonication in Monkey 4. Erroneous signals were observed near the skull base (**) due to low signal and on the dorsal brain surface from a blood vessel (*). Right: Corresponding simulation. The measured temperature rise at the focus was similar to the simulation prediction. The heating pattern on the brain surface was also similar, but the location of the greatest heating was near the back of the head in the experiment, but near the front in the in the simulation (arrows). Bar: 1 cm.

Figure 10

Simulated vs. measured temperature rise at the focus and on the brain surface. (A): A good correlation (R²: 0.67) between measurements and simulations was observed for the focal heating. (B): The simulations and measurements at the focus agreed on average for low temperatures. The measured temperatures were consistently higher than the simulations predicted when the focal temperature rise was above approximately 14°C. (C–D): On average, the simulated and measured temperature rise on the brain surface were consistent, but there was more variation (R²: 0.49). The various symbols represent measurements obtained at different sessions for the four monkeys. Measurements and simulations were consistent over the different sessions for each monkey.

Figure 11

Microphotographs showing H&E-LFB stained sections of two lesions. (A): Axial section showing the lesion from Monkey 2 at approximately two hours after sonication (left: 2.5×; right: 40×). A well-defined lesion core was observed with a sharp boundary. This core was pale-stained, and the neurons in the periphery were shrunken and edematous. Blood vessels within the lesion were congested. A larger area, indicated by the dotted yellow contour, contained dead or dying neurons and disorganized fibers. The circles shown on the 2.5× view represent the dimensions of the thermal dose contours at 240 (red) and 18 (orange) CEM43°C, which were obtained with MRTI acquired in a different imaging orientation. At this time, the dimensions of the lesion core was similar to the 240 isodose contour, and those of the larger area with dead neurons was consistent with the 18 CEM43°C contour. (B) Sagittal section showing the lesion in Monkey 4 at 48h after sonication (left: 2.5×; right: 20×). The pale-stained lesion was edematous and lacked structure. Congested blood vessels were observed along with granulated fibers. The thermal dose contour obtained in MRI at 18 CEM43°C is superimposed on the 2.5× view. The accumulated thermal dose for this lesion was approximately 50 CEM43°C. The tissue in some areas around large blood vessels (*) appeared relatively intact. A gap in the lesion was evident in other sections of this lesion and in MRI (Figure 5D); this gap may have been due to a large vessel evident in other sections. Bars: 1 mm (2.5×); 100 μm (40× and 20×).