4D image data are used in radiation therapy planning to estimate motion of the tumor or regions at risk. While today 4D CT is commonly used for this purpose, MRI offers better soft tissue contrast and quantitative and functional imaging. Therefore MRI is increasingly used in therapy planning and there is also a growing interest in 4D MRI techniques. 4D MRI may be of additional relevance for repeated inter-fraction re-planning. A prospective 4D MR imaging approach based on a highly-efficient 4D MRI method was presented recently [1,2] where the standard respiratory belt sensor is used as motion sensor to infer the tissue motion. However, correlation of sensor and actual tissue motion is a major concern in these techniques. Therefore, an implementation based on the MRI pencil beam navigator approach is proposed here. A refined MR Navigator method was developed as accurate, reliable and robust motion sensor in 4D MRI for therapy-planning independent of anatomy or scan geometry.

As in [1,2] the volume of interest is covered by a stack of 2D image slices with an image of each slice and respiratory phase being acquired within a few hundred milliseconds (e.g. using single-shot TSE or TFE) to obtain a respiratory-cycle-resolved 4D data set based on a short initial respiratory signal calibration phase. As in [1] we utilized a highly efficient scheme with dynamic choice of phase and slice and dynamic reduction of sequence inherent idle time. For highest robustness manual interaction was integrated into the scanning workflow: If navigator-measured tissue position drifts out of the acceptance window of the inital calibration phase [1], a decision dialog will be presented whether to instruct the subject to breathe normally again and proceed with scanning, to finalize the scan prematurely in case all statistically significant motion states have already been captured or to restart the acquisition. Even if incomplete, the obtained 4D data can be imported into existing therapy planning systems. The 4D MR navigator methos was refined to trigger robustly and accurately independent of scan and navigator geometry, the most prominent issue being MR Navigator slice saturation artifacts, i.e. artifacts resulting immediately after a slice was acquired in case the slice intersects the MR navigator:The algorithm estimates the severity of the expected artifact based on the intersection angle and excludes a definable number of navigator acquisitions for triggering to let the MR navigator regain its steady state and to get rid of eventual saturation artifacts.

27 volunteer subjects were included in a feasibilty study. A navigator acquisition rate of 20Hz was used. Different geometries and target organs were tested to verify robustness of the method. MR navigator triggered scans were acquired (FOV 375 (FH) x 260 (AP) mm2, 32 slices, slice thickness 5 mm, resolution 1.5mm x 1.5 mm, single-shot TSE, TR 6400 ms, TE 77 ms, echo spacing 5.2 ms, flip angle 90°, refocussing angle 120°, 10 phases, shot duration 330 ms) and compared to respiratory-belt triggered scans with same parameters for scan efficiency, image quality and overall robustness. All scans were performed on a 3T Ingenia MR system (Philips Healthcare, Best, The Netherlands) using the digital anterior and posterior receive coils.

The navigator method allows robust and accurate triggering in various tested scan and navigator geometries using the saturation prediction algorithm (Figure 1). 4D MR images with high geometric consistency could be generated in a volunteer study with 27 subjects (Figure 2). The mean scan time for a 4D scan with 10 respiratory phases and 32 slices was (334+-36)s. It was found that navigator-based triggering results in more geometrically consistent 4D MRI images than possible with the respiratory belt. The quality difference can be especially high in cases where the belt signal correlates poorly with the internal motion which was found reproducibly in certain volunteers with frequent switching between breathing modes or generally less differentiatable breathing pattern (Figure 3). Another advantage of the proposed technique is the faster workflow and the ability to place the MR navigator pencil beam close to the target anatomy. The latter results in 4D scans with highest possible geometric fidelity which is crucial for therapy planning. An efficient and precise prospective 4D MRI method was presented using a refined MR-based respiratory navigator as motion sensor to accurately track the tissue motion at the target anatomy. The main usage scenarios of this 4D MRI technique are EBRT planning. The technique may be also of interest for combined MR-EBRT or MR-HIFU systems.