Magnetic Resonance Metal Artifact Evaluation with Routine Clinical Cardiac Sequences
D Joyner, A Rivard, D Craft, T Liu, S Callaway, J James*, University of Mississippi Medical Center, Jackson, MSSU-E-I-69 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: To examine schemes to grade the severity of metal susceptibility artifacts on image quality using cardiac MRI pulse sequences.
Methods: A post-thoracotomy patient was simulated with a stainless steel sternal wire (Syneture,MA; size=6, diameter=48mm), placed securely on an ACR MRI phantom. Phantom was scanned on a 1.5-T Siemens using cardiac MRI sequences:1)TrueFISP, 2)Gradient-Recalled-Echo (GRE), 3)Turbo-Spin-Echo (TSE), 4)Turbo-Inversion-Recovery-Magnitude (TIRM), 5)Dark-blood-IR-FS (DBFS) with and without the wire (FOV=30x30cm, slice-thickness/slice-gap=7.0/1.5mm, matrix size=192x192, slices=17). Image quality degradation was assessed in terms of signal loss and spatial deformation; signal loss by a) measuring the largest diameter of signal drop and b) number of slices with a signal drop and spatial deformation in VelocityAI (Atlanta,GA) by computing the rigid transformation indices between the phantom's internal grid with and without the metal.
Results: Image quality was evaluated in terms of signal loss, spatial deformation and ring artifacts. Signal loss: TruFISP and GRE showed the largest signal drop diameter (13 and 16cm respectively). GRE sequence showed a signal drop in ~12 slices where as signal drop occurred in only ~4-5 slices with other sequences. Spatial deformation: GRE sequence showed the maximum with a ~9mm grid deflection followed by TSE and DBFS (~8mm). An average deflection of 5.4mm was observed on most of the sequences except T rueFISP and TIRM (0mm). Rigid body transformation showed a maximum x,y,z-translation of -4.7, 0.3 and 1.69 mm and x,y,z-angular rotation of 0.2, -1.5 and 0.5º for GRE sequence followed by TSE and DBFS confirming the spatial distortion results. Concentric ring artifacts with signal loss were also observed on TrueFISP and DBFS images.
Conclusions: Quantification of cardiac MR sequences to metal tolerance and the impact on image quality has shown that GRE and TrueFISP are the most metal-susceptible and TIRM is the most metal-tolerant sequence in terms of both signal loss and spatial deformation. This study helped in creating a separate cardiac metal protocol comprising of mainly metal-tolerant sequences thus reducing scan time and patient discomfort.