Figure 1
Pipeline depicting the extended motion compensation (MoCo) approach. During a separate planning scan (1), the region of interest on each image navigator is selected interactively by the user (2). These reference images and bounding-boxes are stored on the reconstruction system (3). Timing of the navigated, interleaved MR sequence with the image navigator (red), tracking and data processing (black) and the target sequence, e.g. a 31P MRS sequence (blue). The navigator may consist of an arbitrary number of slices (three shown here). In this example, the trigger delay before the navigators is set for the 31P MRS acquisition to occur during end-systole, which is typically 250–300 ms after the R-wave. With the start of the navigated MRS sequence, the references are loaded to initialize an independent OpenCV tracker for each image navigator (4). Following a trigger signal, the image navigators are acquired (5), reconstructed and the respective trackers are updated (6). In-plane translations (7) are transformed to the patient coordinate system (8) and combined to a single 3D translation vector using SVD (9). The vector is transferred to the target pulse sequence, e.g. MRS, to update the volume of interest (10) before signal acquisition (11). The cycle is repeated for each MRS or MRI transient (12). Transmit and receive frequencies are switched between each image navigator and target acquisition (1H and e.g. 31P, respectively) using multi-nuclear interleaving.
