Introduction & Purpose

Pancreatic cancer has a very poor prognosis. Currently, its treatment is complex and invasive as it often requires an open surgery. Laser ablation (LA) is a minimally invasive procedure already used in clinical practice for the treatment of tumors in deep-lying organs like the prostate, and thus could be an appropriate treatment procedure for pancreatic tumors. In order to ensure the safety and efficacy of the procedure, it is highly desirable to monitor the temperature increase in real-time, which can be performed with MR-based thermometry. Tests have already been carried out ex vivo with T1-weighted sequences¹, but this method requires calibration. Proton resonance frequency (PRF)-based phase shift MRI is more widely used to guide local thermal ablation treatment². The purpose of this preliminary work was to investigate the feasibility to get accurate PRF thermometry images in the in vivo swine pancreas during LA.

Methods

The experiments were approved by the local ethics committee. Three pigs (weight 47-58 kg) were intubated and placed in a 1.5T clinical MRI scanner (Magnetom Aera, Siemens, Germany) in lateral decubitus position. During all the procedure, anesthesia and uniform ventilation were maintained with 2% isoflurane and intermittent administration of curare. The abdomen was surrounded by a four-element flexible phased-array coil. Two custom-made carbon fiber MRI compatible needles were inserted inside the pancreas under MRI guidance and were ~5mm apart. An optical fiber guiding Nd:YAG laser light (2 to 3.5 W power) was inserted in the first needle, and a fiber Bragg grating sensor providing real-time temperature increase was inserted in second needle³. The guidance sequence was a T1-weighted sequence acquired during apnea, providing anatomical images at the same time (VIBE sequence, TE/TR=1.98/4.69ms, pixel resolution 1.17*1.17*2mm³, TA=50s). Two LA were performed for each pig in the head of the pancreas: after the first ablation, both needles were moved a few centimeters away and a second ablation was performed without overlapping with the former one. PRF-thermometry images were acquired with a dedicated respiratory-gated GRE-EPI sequence providing magnitude, phase subtraction and motion images. The ablation began 30s after the beginning of the acquisition and stopped ~2 minutes before the end of the acquisition, and lasted 10 (for the first two pigs) or 15 min (for the last pig). The following parameters were used: TE/TR=13/43ms, FoV 320*320cm², in-plane resolution 128*128 pixels, three 3.5mm-thick axial slices, fat suppression, total number of scans 240/300 (80 or 100 for each slice). Pigs were euthanized at the end of the procedure and the pancreas was removed for histological assessment. Temperature maps were reconstructed from the phase subtraction images with a homemade Matlab routine. Phase unwrapping was performed in case of aliasing and B₀ drift was corrected by considering the phase at each time step in a ROI chosen in the pig dorsal muscle where no heat effect occurred.

Results & Discussion

An example of temperature map superimposed on the magnitude image is shown in Fig1. This figure shows a small area where consistent temperature increases were found as compared with the sensor (Fig. 2A and 2B). In this area, as the distance from the tip of the laser increases, lower temperatures and greater thermal inertia were found as expected (Fig. 2C and 2D). Macroscopical and histological examination confirmed the small size of the lesion for all the animals (~1cm diameter , see Fig. 3). More surprisingly, a large adjacent area was found with negative ΔT. This effect was seen for all the ablations and was attributed to the presence of gas bubbles (water steam due to very high temperatures), leading to cavitation-induced phase shift in our images. Such apparent decrease in the temperature map has already been reported during radiofrequency ablation, and it has been demonstrated that this effect did not occur at low power⁴. A first order correction has been proposed by the same authors to minimize the measurement errors⁴. This result emphasizes the importance of the chosen therapeutic power for the reliable monitoring of LA.

Conclusions & Future Work

PRF thermometry images and consistent temperatures curves were obtained during LA in the in vivo swine pancreas, despite the presence of a susceptibility artifact that could hamper the monitoring of the procedure. We intend to perform new experiments to investigate the effect of the laser power on our thermometry images.

Acknowledgements

We thank Thibaut Eisele and Khalid Ambarki from Siemens for their kind support about the MRI protocol and the IHU team of technicians for the management of animal anesthesia.