
ED/Neurorad @mghimaging | MRI translation @MGHmartinos
- Tweets60
- Following342
- Followers207
- Likes16.3K

ALT One of the biggest challenges in pediatric MRI is managing motion artifacts. Sedation or anesthesia is frequently used, to minimize the motion artifacts. However, these approaches carry increased medical risks and additional costs. Motion correction techniques offer a promising solution, however, no single method has yet gained widespread clinical acceptance due to the stringent requirements for robustness, reliability, and seamless integration into clinical workflows. Data-driven retrospective motion correction offers a promising solution to these challenges. It derives the motion trajectory information from the acquired k-space data of the imaging sequence and produces the motion-corrected images without the need for additional tracking hardware or disruptive sequence modifications. Moreover, a significant advantage over prospective techniques is that radiologists can directly compare the motion-modeled reconstruction with the original, non-corrected images.

ALT Data-driven retrospective motion correction can be formulated as a “do-no-harm” strategy. These methods extract the motion information and the motion-free image directly from the k-space data of the imaging scan without requiring additional tracking hardware or disruptive sequence modifications. This ensures that the baseline image quality and contrast will be maintained and that the motion-mitigation strategy only adds value to the exam. Radiologists have the opportunity to directly compare the motion-modeled reconstruction against the original non-corrected images for improved diagnostic utility. Unfortunately, data-driven motion correction is computationally demanding because it often requires the inversion of a large and sophisticated nonlinear physics model that describes the effect of motion on the acquired k-space data. The scout accelerated motion estimation and reduction (SAMER) algorithm will reduce the computational burden of motion estimation.

ALT The scout accelerated motion estimation and reduction (SAMER) algorithm will reduce the computational burden of motion estimation by leveraging an ultra-fast scout scan along with the repeated acquisition of a very limited number of additional k-space encoding lines. This approach will facilitate robust and highly efficient retrospective motion correction with clinically acceptable computation times.

ALT To take advantage of the efficiency of EPI while mitigating signal-to-noise (SNR) and susceptibility-induced losses to image quality, the authors have developed a multi-shot EPI (msEPI)-based prototype that leverages a deep learning-based image reconstruction to provide high-SNR T1w, T2w, T2*, and T2-FLAIR, complemented by ssEPI diffusion imaging within a 2-minute exam. The prototype’s msEPI acquisition acquires readout lines in an interleaved fashion across several shots to achieve shorter echo times and reduce geometric distortion. At the same time, the deep learning (DL)-based image reconstruction produces high-SNR images and limits g-factor noise amplification. To provide greater flexibility and robustness to variations in the clinical workflow, the prototype allows each contrast to be acquired individually, in any orientation, and the DL-reconstruction includes a tunable parameter which can be automatically selected to optimize image quality and data fidelity.






























