Percutaneous hyperthermal ablations are intended to destroy cancerous tissues by locally increasing the temperature over 60°C. The MRI guidance of these minimally-invasive procedures offers the possibility of measuring the temperature in real-time. Nevertheless, no information related to structural properties of the tissue is available during the procedure. Measuring elasticity during the procedure has been proposed and liver stiffness has been shown to increase after a thermal ablation ^1,2. Magnetic Resonance Elastography (MRE) allows measuring the mechanical properties of tissue in vivo from MR phase images by estimating the wave propagation speed generated by a mechanical exciter ³. However, current MRE systems do not meet the demanding conditions of interventional MRI in terms of bulk and update rate. An all-in-one interventional MRE system has been proposed, and feasibility of monitoring elasticity changes in real-time using this method has been demonstrated on phantoms ⁴. The objective of this work is to demonstrate the feasibility of monitoring thermal ablations in vivo in real-time using interventional MRE combined to MR thermometry.

The interventional MRE system is composed of: 1/ a vibrating needle MRE driver that generates waves directly within the region of interest, 2/ a fast and interactive spoiled gradient-echo MR pulse sequence that encodes the motion on MR phase images, and 3/ an online LFE-based inverse problem solver that reconstructs elasticity maps in real-time. The MRE pulse sequence has been adapted to allow for respiratory triggering during the long in vivo thermal ablation procedures, making continuous monitoring of the procedure possible without any need for breath-holding. Each respiratory trigger is followed by a mechanical excitation trigger, dummy cycles that allow the steady-state to be reached and finally, two phase images with opposite gradient polarities. On the one hand, these two images are subtracted to enhance the wave image and remove the background phase. On the other hand, these images are also used for temperature measurement by adding them together ⁵, allowing simultaneous MRE and MR thermometry. The shift between the respiratory trigger and the mechanical excitation trigger is incremented so that three phase-offsets, covering one mechanical period, are acquired every three respiratory cycles (Fig.1). Thanks to a sliding window scheme, one elastogram is reconstructed with every respiratory cycle, i.e. with every new phase difference image.The number of MRE phase offsets is reduced to three in order to optimize elasticity measurements while minimizing temporal smoothing effects⁴.

First, in order to assess the stability of the system, 127 elastograms are acquired during 10 minutes with the described protocol in a swine liver in vivo.

Second, a thermal laser ablation (DIOMED 25) is performed in another swine liver in vivo and is monitored in real-time over 20 minutes by simultaneous MRE and MR thermometry. Four optical fibers are inserted in the liver by an experimented interventional radiologist (Fig.2). A total power of 12 W is delivered through the fibers during the first 10 minutes of the procedure. The needle MRE driver is positioned so that mechanical waves are propagating through the planned ablation zone, i.e. at the optical fiber tips.

The animals are continuously ventilated with a respiratory rate of 12 breaths per minute. A respiratory belt is placed on their abdomen for triggering during the expiration phase. Relevant MRE parameters include: excitation frequency 60 Hz, encoding frequency 90 Hz, acquisition matrix 102×128, GRAPPA ×2, MEG amplitude 20 mT/m, TE/TR 9.34/16.67ms, flip angle 13°, through slice encoding, one slice orthogonal to the needle MRE driver. Total acquisition time per respiratory cycle is 2.76s corresponding to the dummy cycles and the acquisition of the two images.

A standard deviation inferior to 5 % of the average shear modulus is obtained during the first experiment (Fig.3). Wave images acquired during the ablation are shown in Fig.4 The simultaneous evolutions of shear modulus and temperature in the lesion are plotted in Fig.5. When the laser is turned on, the wavelength continuously increases in the fiber tip zone corresponding to the increase of the shear modulus in the elastograms. As expected, the target ablation temperature of 60°C is reached after 10 minutes. When the laser is turned off, the temperature decreases while the shear modulus stabilizes in the lesion.For the first time, in vivo thermal ablation is monitored by simultaneous MR elastography and MR thermometry with a refresh rate of one elastogram with every breathing cycle, i.e. every 5 seconds. This experiment highlights the complementarity of simultaneous temperature and elasticity monitoring during thermal ablations.