Clinical Evaluation of Scout Accelerated Motion Estimation and Reduction (SAMER) Motion-Corrected 2D T2-Weighted TSE 3T Brain MRI in the Neurologic Intensive Care Unit

Abstract

BACKGROUND AND PURPOSE: Patient motion remains a major source of image degradation in clinical MRI, particularly in acute and inpatient settings. We evaluated the clinical effectiveness of the Scout Accelerated Motion Estimation and Reduction (SAMER) method for retrospective motion correction in axial 2D T2-weighted TSE brain MRI, particularly in emergency and inpatient neurologic settings, where patient motion commonly degrades image quality. MATERIALS AND METHODS: This prospective single-center study included 275 patients undergoing brain MRI on a 3T MRI scanner located in the neurologic intensive care unit between March and September 2024. Each patient received a T2-weighted TSE sequence embedded with the SAMER motion correction framework. Motion artifacts were assessed using 2 approaches in a blinded, randomized fashion: 1) a motion grading scale applied to individual image series before and after SAMER correction by 2 experienced radiologists, and 2) a head-to-head preference test comparing original and motion-corrected images, evaluated by 2 additional neuroradiologists. Statistical analysis included the Wilcoxon signed rank test, intraclass correlation coefficients (ICC), and Cohen kappa agreement scores. RESULTS: Motion scores improved significantly after SAMER correction, with average motion grades reduced from 0.79 to 0.62 (Rater 1) and 0.68 to 0.52 (Rater 2) (P < .001 for both). For cases exhibiting at least minimal motion, the improvement was even more pronounced. ICCs were high for both original (0.92) and postcorrection (0.90) ratings, indicating excellent interrater reliability. In the head-to-head comparison, SAMER-corrected images were preferred in 22%-23% of cases, with the original images preferred in fewer than 2%. Cohen kappa showed moderate interrater agreement (κ = 0.61) for motion artifact reduction and fair agreement (κ = 0.39) for overall image quality. A small subset of cases exhibited unexpected results, such as increased motion scores or inhomogeneous correction, suggesting areas for further refinement. CONCLUSIONS: SAMER significantly reduces motion artifacts in 2D T2-weighted TSE brain MRI performed in the neurologic intensive care unit, improving diagnostic image quality. The method is reliable across multiple raters, integrates seamlessly into clinical workflows, and offers a promising motion correction solution without additional hardware requirements. Further research is warranted to optimize performance under extreme motion conditions and across additional MRI sequences.