Thermal therapies, such as high-intensity focused ultrasound (HIFU) and laser-induced thermal therapy (LITT) are increasingly being applied in oncology and neurology. MRI is an attractive modality for guidance since it can track temperature and has a good soft-tissue contrast, useful for visualizing the lesion. It is important that the PRFS thermometry sequence is fast enough to keep up with temperature changes, to reduce motion artifacts and to allow for enough spatial coverage. Our hypothesis is that the acquisition can be accelerated and the spatial coverage can be increased by using Simultaneous multi-slice imaging (SMS) [1] in combination with CAIPIRINHA [2], and that the potential SNR advantage that this method has over SENSE translates in a higher precision of the temperature measurements. We use HIFU and LITT for heating and compare the temperature curves acquired with SMS-CAIPIRINHA accelerated sequences with those acquired with unaccelerated sequences, SENSE accelerated sequences and hybrid SENSE/SMS-CAIPIRINHA accelerated sequences.

To test our hypothesis two different experiments were performed. The first used HIFU heating in ex-vivo bovine tissue with a GRE sequence and the second used LITT heating in an agar phantom with a multi-shot EPI GRE sequence.

HIFU: The HIFU experiment was performed on a 1.5T Achieva scanner using a Sonalleve HIFU system (Philips, Best, NL) for continuous heating at 30W during 1 minute. A fat-suppressed GRE sequence with α/TE/TR = 10°/9.2ms/12ms, FOV 24x24cm², voxel size 1.5x1.5x5mm³ and a 20 channel coil array were used for imaging. To achieve CAIPIRINHA shifting, different MultiBand (MB) RF pulses were cycled each phase encoding line [3]. The MB pulses were calculated in Matlab (The Mathworks, Natick, USA). The stack contained two slices and the dynamic scan time of the unaccelerated sequence was 6.1s. Four different scans were compared: an unaccelerated scan, a SENSE scan with acceleration factor 2, a SMS-CAIPIRINHA scan with acceleration factor 2 and a SENSE/SMS-CAIPIRINHA scan with total acceleration factor 4. For all scans the total scan time was 4 minutes consisting of the non-heating, heating and cool-down regimes.

LITT: The LITT experiment was performed on a 1.5T Ingenia scanner (Philips, Best, NL) using a Nd:YAG laser (TMS, Umkirch, DE) for continuous heating, operating at 36W during 1 minute. A multi-shot EPI GRE sequence with α/TE/TR = 12°/15ms/35ms, FOV 21x21cm², voxel size 1.5x1.5x5mm³ and a 15 channel coil array was used for imaging. CAIPIRINHA shifting was achieved by using gradient blips [4] instead of cycled RF pulses. With this imaging sequence the same comparison as in the HIFU experiment was made. The dynamic scan time was 2.8s and the accelerations were used to increase the spatial coverage.

Analysis: The images were further processed in Matlab where temperature curves were obtained from averaging over a ROI of 2x2 mm². The precision of the temperature measurements was quantified by calculating the standard deviations of the curves in the non-heating regime.

The HIFU transducer created a small lesion in the tissue (fig. 1) of which the temperature was monitored. The temperature curves of the accelerated scans of the HIFU experiment (fig. 2b-d) were in good agreement with the temperature curves of the unaccelerated scan (fig. 2a). The standard deviations were σ_­unacc ­= 0.68 °C, σ_SENSE = 1.1°C, σ_SMS = 0.66 °C and σ_SENSE/SMS = 1.1 °C. Similarly in the LITT experiment the accelerated EPI scans (fig. 3b-d) were in good agreement with the unaccelerated EPI scan (fig. 3a). The standard deviations were σ­_unacc ­= 0.3°C, σ_­­­SENSE = 0.44°C, σ_SMS = 0.32°C and σ_SENSE/SMS = 0.63°C. This shows that SMS-CAIPIRINHA can accelerate both GRE and EPI GRE scans without loss of precision, although most PRFS thermometry sequences use EPI readouts. The standard deviation is inversely proportional to the SNR. It was therefore to be expected that σ_SMS < σ_SENSE. The improvement in precision was quantified by the dimensionless quantity σ_­­­SENSE / σ­­­_SMS (fig. 4). In regions with negligible geometry factor this quotient was close to the theoretical value of √2. This was expected since the SENSE scan acquired half of k-space while SMS-CAIPIRINHA acquired the full k-space albeit for two slices simultaneously. PRFS thermometry can be accelerated with SMS-CAIPIRINHA, which results in a higher precision compared to SENSE acceleration. The acceleration can be used for increasing the temporal resolution as demonstrated in the HIFU experiment, or for increasing the spatial coverage as demonstrated in the LITT experiment.