Measurement of patient temperature during MRI scanning is becoming increasingly important as patient heating may occur even when FDA/IEC limits are followed^1,2. Unlike the case of highly localized heating (relevant in for example MRI-monitored HIFU), temperature measurements to monitor surface SAR do not need to have a very high spatial resolution, but should ideally be obtained from all over the surface of the body. Typical MRI-compatible temperature sensors are fluorescent-based, and can only cover a very small number of specific locations. MRI-based methods such as the proton reference frequency (PRF) method are not well-suited to measuring surface temperatures. Here we introduce a new method of large volume distributed surface temperature measurement based on Raman spectroscopy. Both the peak position and the peak width of the Raman spectra are affected by changes in temperature, due to the anharmonic nature of vibrational modes.

A Silixa iDAS system was connected to ~50 metres of optical fibre. The fibre first passes through an ice-bath, then is formed into a rectangular spiral to cover an area of ~60 x 90 cm, as shown in Figure 1, before passing back through the ice-bath and back to the data acquisition system. The ice-bath is used to calibrate the system and length of optical fibre. The time resolution of the system corresponds to a spatial point spread function with FWHM of ~12.5 cm.

Temperature measurements in phantoms were performed by placing a small vial of saline onto one length of the optical fibre, as shown in Figure 1. A small surface coil was deliberately driven at high power levels in order to induce heating. This experiment was performed on a 7 T scanner for the ease of interfacing custom-built transmit/receive RF coils. In the second experiment, a volunteer lay on top of the fibre-bed in a 3 Tesla scanner order to illustrate the whole body coverage of the temperature measurements.

Figure 2(top) shows a van De Giesen Plot (length along cable on the horizontal axis, time on the vertical and color representing temperature) for the small hot water phantom placed on the bed, followed by the RF excitation. The localized increase in temperature can clearly be seen. The two dark blue areas correspond to the ice bath calibrations. Figure 2(bottom) shows a plot of temperature vs time, showing the rapid temporal response to the increase in temperature. There is a small additional heating induced by the RF coil, shown as a slight increase in the slope of the graph. After RF heating the water vial is removed from the coil and the temperature decreases sharply.

Figure 3 shows results of a volunteer lying on top of the mat in the whole body 3T scanner. There is a clear increase in temperature all over the fibre optic. The plot shows that, for the clinical protocol scans used (which lies within FDA guidelines), there is minimal surface heating of the subject.

This works shows a new technique for non-invasive surface measurements within an MRI system. The principle is based on Raman backscattering of light, and since a simple fibre-optic is used to sense the temperature there is no interference in the thermal measurements from either RF or gradients. The great advantage over current techniques is that the measurements are distributed over a very large volume. The intrinsic resolution is currently on the order of 12.5 cm, but this can easily be increased by using a criss-cross pattern of overlapping fibres and reconstruction methods based on Anger logic or more sophisticated algorithms. In addition to monitoring patient heating, the technique can also potentially be used to monitor the temperature within the RF coil, or the magnetic field gradient coils.