Patients with implanted devices, including deep brain stimulators and pacemakers, often have restricted access to MRI [1,2] due to RF tip heating concerns. RF tip current must also be limited when visualizing electrophysiology (EP) catheters or dipole micro-transceivers in interventional MRI [3]. Thermo-acoustic ultrasound, in which thermoelastic expansion caused by heating creates measureable pressure waves, is a promising method for detection of local heating. Here, we compare sinusoidal and square wave transmit modulation schemes to pulse trains for thermo-acoustic ultrasound. We then demonstrate the first thermo-acoustic ultrasound detection of tip heating in an MRI scanner.

For initial tests comparing alternative transmit methods, we generated thermo-acoustic signals by driving current directly into a coaxial cable immersed in saline solution with its center conductor exposed at the tip (Figure 1e). All methods used the same average transmit power level and a total transmit duration of 4ms.

Continuous modulation of RF at ultrasound transducer frequencies generates continuous wave (CW) thermo-acoustic ultrasound [4,5]. Linear frequency modulated CW (FMCW) results in frequency-encoding of depth. To create a thermo-acoustic signal, a 64MHz carrier was amplitude-modulated by a 2ms RF chirp with a 375kHz bandwidth centered at 1MHz using the FMCW technique (Figure 1a).

The RF power of a sinusoidal amplitude-modulated signal has an envelope similar to a square wave. Because thermo-acoustic pressure depends on transmit power, we tested the ability of square wave modulation to create thermo-acoustic signals. A 64MHz carrier was square-wave modulated at a 50% duty cycle, but with pulse repetition frequency modulated using FMCW with a 375kHz bandwidth and 1MHz center frequency (Figure 1b). A similar method, with the square wave on-time fixed at 500ns but with FMCW-modulated pulse repetition frequency, was tested as well (Figure 1c).

Classic ultrasound pulse trains were also tested. A pulse length of 500ns was chosen so that the power profile of the transmitted signal had significant spectral content at 1MHz - the ultrasound transducer frequency. Repetitions of 64MHz carrier pulses were used (Figure 1d).

To detect the thermo-acoustic signal from tip heating induced in MRI, we interfaced our thermo-acoustic detection setup with the body coil transmitter of a GE Signa Excite 1.5T scanner (Figure 3). For 250 averages, a Medusa­ console­ [6] configured in slave mode transmitted a 3ms FMCW signal centered at 1MHz with a 375kHz bandwidth when triggered by the body coil transmit gate. The signal amplitude modulated a 63.855MHz carrier frequency, which was applied to the body coil amplifier through an RF SPDT switch. The transmit signal waveform (Figure 3c) was verified using a pickup loop inside the magnet. A long wire was placed in the scanner bore, and one end of the wire with an exposed tip was immersed in saline solution. An Olympus V303 immersion transducer was placed in the solution and attached to an RF-shielded box containing a preamplifier and LED driver powered by a non-magnetic battery. The LED optically coupled the amplified transducer signal to a photo-receiver outside the MRI scan room to minimize RF interference from the body coil. This signal was then filtered and sent to the Medusa RF receiver.

Figure 2 shows the depth profiles of the thermo-acoustic signals from the different transmit and receive schemes when directly driving the stripped coaxial cable as a signal source. We also found that signal sensitivity dropped by ~4x when the driven coax tests were repeated inside the magnet bore. Figure 4 shows the thermo-acoustic signal resulting from coupling between the body coil’s electric field and a long wire in the bore. System nonlinearities and RF interference created an observable feedthrough signal at 0cm. In both figures, the signal source was present at a depth of ~6cm.

For the chosen transmit parameters and using a 1MHz transducer, the SNR of the different transmit and receive methods are comparable. For the modulated CW techniques, resolution of the reconstructed signal was ~5mm for the 375kHz bandwidth transmit signal. For the pulse train transmit, the resolution was limited to ~3mm by the 500kHz bandwidth of the Medusa receiver. The required scan time was significantly longer for the pulse train because unlike the modulated CW methods, transmission and reception could not occur simultaneously.

Our first demonstration of thermo-acoustic detection of RF tip heating from an MRI body coil achieved > 100 SNR. Conceptually, a thermo-acoustic modulation system could augment any MRI scanner through a SPDT RF routing switch to the body coil transmitter. This demonstration bodes well for the prospects of using thermo-acoustic ultrasound to assess RF heating risks of implanted and interventional devices.