Temperature dependence of canine brain tissue diffusion coefficient measured<i>in vivo</i>with magnetic resonance echo-planar imaging

Abstract

The intensity of conventional spin-echo diffusion-weighted magnetic resonance (MR) images is approximately linearly dependent on temperature over a restricted range using conventional diffusion-weighted spin-echo magnetic resonance imaging (MRI). However, conventional diffusion-weighted MRI is too motion sensitive for in vivo thermometry. The present work evaluated rapid diffusion-weighted echo-planar imaging (EPI), which is less sensitive to motion, for application to non-invasive thermometry in acrylamide gel materials and in vivo in canine brain tissue for applications in therapeutic hyperthermia. The rapidly switched, strong gradients needed for EPI were achieved using a 'local' z-axis gradient coil. Gel materials were heated with a small (10 cm diameter) spiral surface microwave (MW) applicator at 433 MHz, while in vivo heating was accomplished with whole body RF hyperthermia using an annular phased array (130 MHz). The MW or RF fields associated with heating and imaging (64 MHz) were decoupled using bandpass filters providing isolation in excess of 100 dB. This isolation was sufficient to allow simultaneous imaging and MW or RF heating without deterioration of the image signal-to-noise ratio. Using this system in a gel, temperature sensitivity of the diffusion coefficient was observed to be (3.04 +/- 0.03)%/degrees C which allowed temperature changes of 0.55 degrees C to be resolved for a 1.8 cm3 region in < 10 s of data acquisition. In vivo, cardiac gating of the pulse sequence was necessary to minimize motion artifacts in the brain. The temperature sensitivity of brain tissue was (1.9 +/- 0.1)%/degrees C allowing temperature changes of 0.9 degrees C to be resolved in a 0.9 cm3 volume in < 10 s of data acquisition. We conclude that with further optimization of the data acquisition conditions it will be possible to determine 0.5 degrees C temperature changes in 1 cm3 volumes in < 10 s using this technique.