gms | German Medical Science

Introduction: Respiratory motion is a common source of artifacts in MRI, especially in abdominal imaging. Existing approaches try to remove the artifacts either by motion correction in the reconstruction process or already prevent them during the acquisition. Among the latter, a very efficient strategy is breath-holding during the acquisition. However, the ability of patients to comply with predefined breath-hold durations varies greatly. Recently, a sampling pattern was proposed [1], which supports the reconstruction of images from data acquired up to a certain point in time and scan termination at breathing onset. A compromise is made between undersampling artifacts and spatial resolution. In this work, different methods to detect motion for automatic scan termination and the application to water-fat imaging are presented.

Methods: A 3D dual-gradient-echo sequence was chosen to measure the same k-space line twice in a single TR. The specific sampling pattern [1] was generated on-line during protocol specification. A segmented acquisition was performed with motion sensing interleaved. For this purpose, either a 1D pencil beam navigator, positioned at the right diaphragm, or repeated measurements of the k-space centre are performed. During breath-holding, the profiles at the centre stay constant, which is equivalent to maximum correlation. At onset of breathing, the correlation decreases and can be used for motion sensing. The scans were automatically terminated at breathing onset, resulting in an incomplete dataset regarding the Shannon sampling theorem. The chosen sampling pattern ensures incoherence at every point in time and, therefore, permits successful compressed sensing reconstruction. After off-line reconstruction with l1-SPIRiT [2], a Roemer reconstruction [3] was carried out to combine the images from the individual coil elements, followed by water-fat separation [4] yielding water and fat images. Abdominal imaging was performed on volunteers on a 1.5T scanner (Philips Healthcare, The Netherlands), using a 16-element torso coil. A typical FOV of 380 x 280 x 240 mm3 was covered with a target spatial resolution of 1.5 x 1.5 x 3.0 mm3, with a TE1/TE2/TR of 1.3/2.3/3.7ms.

Results: Imaging with termination at breathing onset for varying durations was successfully performed on 11 volunteers. Consistent and good to very good image quality for 3D water/fat-resolved abdominal MR was obtained. Comparison between the two possibilities for motion sensing in scans using both methods indicate breathing onset at the same time. For a comparison without misregistration, shorter breath holding was simulated by further undersampling retrospectively, indicating the desired increase in spatial resolution with the breath-hold duration.

Conclusion: This work advances the previously proposed approach towards clinical applicability. Water-fat resolved abdominal imaging with self-termination provides good image quality adapted to the breath-hold capabilities of the individual patient.