At present, no efficient therapy for pancreatic cancer is available. Unfortunately, 80–85% of patients present with advanced unresectable disease.^₁ Magnetic Resonance Imaging-guided High Intensity Focused Ultrasound (MR-HIFU) has recently been suggested as an alternative treatment modality that is non-invasive, and may be suited for treatment in cases where surgery is not an option.² Therapy guidance using MR temperature mapping (MRT) is essential for controlling thermal MR-HIFU therapy. Limited but promising data has been reported regarding the feasibility of breathhold and multi-baseline MR thermometry in the pancreas body.³ An additional region of interest is the head of the pancreas, where about 65% of all detected tumors are located.⁴ However using proton resonance frequency shift (PRFS) based thermometry ⁵ in this area is highly challenging, because of motion (respiratory, peristaltic) and the presence of air in the digestive tract near the pancreas. In this region, unpredictable motion leads to changes in the magnetic volume susceptibility distribution which result in artifacts in the temperature maps.

In this study, we evaluated the performance of an approach aimed at reducing susceptibility heterogeneity by filling the stomach and duodenum with a fluid in order to improve the stability PRFS based temperature mapping.

Image Acquisition: Experiments were performed on five healthy volunteers on a clinical MR-HIFU system (Philips Sonalleve, Philips Healthcare, Vantaa, Finland) integrated with a clinical 1.5-T MRI scanner (Achieva, Philips Healthcare, Best, The Netherlands). The subjects were positioned prone. The imaging protocol described below was executed twice: first, without any special preparation and then 2 to 5 minutes after oral intake of 500mL of fluid with favorable relaxation times (pineapple juice) to fill the tract. For planning, an anatomical 3D fat suppressed T1-weighted (gradient echo) scan was acquired in a breathhold. For temperature mapping, a dynamic segmented echo planar imaging series consisting of 3 coronal slices covering the pancreas head and part of its body was acquired for a duration of 3min20s. Scan parameters included: TR = 100 ms, TE = 19 ms, FA = 21, reconstructed voxel size = 2.5x2.5x8 mm3, dynamic scan time = 400 ms, 500 dynamics. To create a stable breathing pattern, volunteers were asked to breath according to an acoustic signal indicating the desired inhalation and exhalation phases.

Image analysis: To evaluate the effect of gating, phase images were selected based on the respiratory belt signal, only expiration images were used to compute relative temperature maps.⁵ The first gated phase image was used as a reference for phase subtraction.

Statistical analysis: As a performance indicator, the temporal temperature standard deviation (SD) was calculated over the time course of the retrospectively-gated dynamic series on a voxel-by-voxel basis within a region of interest obtained by semi-automatic segmentation of the pancreas in the anatomical T1 scan.

Figure 1, shows an example of the juice filling effect in one of the volunteers. Without juice the standard deviation varied from 1 to 10 degrees Celcius. After juice intake, the range went down to values between 0.75 to 5 degrees Celcius and the spatial distribution became more homogeneous.

Figure 2, shows that gating clearly reduced the temporal SD in every volunteers. By comparing the Gating/No Juice (green) and Gating/Juice (blue) box plot the figure shows that the use of juice further lowered all median values, by 0.5 to 2.9°C, and reduced the interquartile range (IQR) on average by a factor 2 (range 1.5 to 3).

In all volunteers, oral ingestion of pineapple juice for patient preparation reduced both the median and the IQR of the temporal temperature standard deviation inside the pancreas.

Filling the stomach and duodenum in this way could be an attractive and pragmatic solution for more precise temperature monitoring during MR-HIFU therapy in this challenging region of interest, and in the head of the pancreas, in particular.