Instrument visualization using conventional and compressed sensing SEMAC for interventional MRI at 3T.

Background: Interventional magnetic resonance imaging (MRI) at 3T benefits from higher spatial and temporal resolution, but artifacts of metallic instruments are often larger and may obscure target structures.

Purpose: To test that compressed sensing (CS) slice-encoding metal artifact correction (SEMAC) is feasible for 3T interventional MRI and affords more accurate instrument visualization than turbo spin echo (TSE) and gradient echo (GRE) techniques, and facilitates faster data acquisition than conventional SEMAC.

Study Type: Prospective.

Phantom And Subjects: Cadaveric animal and 20 human subjects.

Field Strength/sequence: TSE (acquisition time 31 sec), GRE (28-33 sec), SEMAC (128 sec), and CS-SEMAC (57 sec) pulse sequences were evaluated at 3T.

Assessment: Artifact width and length, signal-to-noise (SNR), and contrast-to-noise (CNR) ratios of 14-22G MR-conditional needles were measured in a phantom. Subsequently, high-bandwidth TSE and CS-SEMAC sequences were assessed in vivo with 20 patient procedures for the size of the metal artifact, image sharpness, image noise, motion artifacts, image contrast, and target, instrument, and structural visibility.

Statistical Tests: Repeated-measures-analysis-of-variances and Mann-Whitney U-tests were applied. P ≤ 0.05 was considered statistically significant.

Results: CS-SEMAC and SEMAC created the smallest needle artifact widths (3.2-3.3 ± 0.4 mm, P = 1.0), whereas GRE showed the largest needle artifact widths (8.5-8.6 ± 0.4 mm) (P < 0.001). The artifact width difference between high-bandwidth TSE and CS-SEMAC was 0.8 ± 0.6 mm (P < 0.01). SEMAC and CS-SEMAC created the lowest average needle tip errors (0.3-0.4 ± 0.1 mm, P = 1.0). The average tip error difference between high-bandwidth TSE and SEMAC/CS-SEMAC was 2.0 ± 1.7 mm (P < 0.01). SNR and CNR were similar on TSE, SEMAC, and CS-SEMAC, and lowest on GRE. CS-SEMAC yielded smaller artifacts, less noise, less motion, and better instrument visibility (P < 0.001); high-bandwidth TSE showed better sharpness (P < 0.001) and targets visibility (P = 0.007); whereas image contrast (P = 0.273) and structural visibility (P = 0.1) were similar.

Data Conclusion: CS-SEMAC is feasible for interventional MRI at 3T, visualizes instruments with higher accuracy than high-bandwidth TSE and GRE, and can be acquired 55% faster than conventional SEMAC.

Level Of Evidence: 2 Technical Efficacy: Stage 6 J. Magn. Reson. Imaging 2018;47:1306-1315.