Accuracy of Compressed Sensing Accelerated 3d-Flair in the Diagnostic Work-Up of Multiple Sclerosis at 1.5 T

BACKGROUND: 3D-FLAIR is central to MR-based diagnosis and monitoring of MS, but long acquisition times remain a limitation. OBJECTIVES: This single-center prospective non-inferiority study evaluated whether compressed sensing (CS) - accelerated 3D-FLAIR at 1.5T provides lesion detection comparable to optimized Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA) - accelerated 3D-FLAIR. Optimized CS- and GRAPPA-accelerated 3D-FLAIR sequences were incorporated into MRI examinations in patients with known or suspected MS. The primary endpoint was manual lesion count by two blinded readers using a 10% non-inferiority margin. Secondary endpoints included image quality, SNR, CNR, and automated lesion count and volumetry. RESULTS: Sixty-eight patients were included (mean age, 40.1+/-13.3 years; 61.8% women). CS-accelerated 3D-FLAIR reduced acquisition time by 31.8% compared with GRAPPA (4:00 vs 5:52 minutes). Non-inferiority for manual lesion count was demonstrated for both readers (reader A, p < .0001; reader B, p = .005). Interrater agreement was excellent for CS (ICC = 0.921; 95% CI, 0.837-0.957) and very good for GRAPPA (ICC = 0.851; 95% CI, 0.769-0.906). Subjective image quality was high for both techniques. GRAPPA showed higher objective signal-to-noise-ratio (SNR) and contrast-to-noise-ratio (CNR), whereas automated lesion counts and volumes were higher for CS. CONCLUSION: CS-accelerated 3D-FLAIR provided non-inferior lesion detection with substantial scan-time reduction.