PROSPECTIVE COMPARISON OF OPTIMIZED COMPRESSED SENSING AND GRAPPA-ACCELERATED 3D-TOF-MRA FOR EVALUATION OF UNTREATED INTRACRANIAL ANEURYSMS AT 1.5 T: A NON-INFERIORITY STUDY

RATIONALE AND OBJECTIVES 3D-TOF-MRA is a standard non-invasive method for detecting and monitoring intracranial aneurysms, but long scan times remain a limitation. Compressed sensing (CS) acceleration for intracranial aneurysm evaluation has mainly been studied at 3T. This prospective single-center non-inferiority study compared CS-accelerated 3D-TOF at 1.5T with GRAPPA-accelerated 3D-TOF-MRA for evaluating untreated intracranial aneurysms. MATERIALS AND METHODS Sample size was determined via a preliminary retrospective study. 25 patients with 32 untreated aneurysms underwent both CS- (factor 7, 3:04 min) and GRAPPA-accelerated (factor 2, 5:00 min) 3D-TOF, yielding a 38.7% scan-time reduction. Two blinded neuroradiologists assessed overall image quality (primary endpoint) and artifacts, subjective CNR, aneurysm delineation and diameters (secondary endpoints). Non-inferiority was tested via one-sided t-test (prespecified 10% margin); interrater agreement used ICC. RESULTS CS demonstrated non-inferiority to GRAPPA for overall image quality in both readers (p <= .023), though interrater agreement was low (ICC 0.241-0.262), reflecting subjective variability. Secondary exploratory analyses showed heterogeneous preferences. ROI-based CNR was lower for CS (median 169.6; IQR, 91.0-226. vs. 278.2; 181.0-366.2). Aneurysm diameter measurements showed excellent interrater agreement for both techniques (CS ICC, 0.899-0.980; GRAPPA ICC, 0.877-0.973; all p < .001). CONCLUSION CS-accelerated 3D-TOF at 1.5T met non-inferiority criteria vs. GRAPPA-TOF for overall image quality, reducing scan time by 38.7% while preserving aneurysm measurement reliability. However, reduced CNR and reader-preference variability important trade-offs.