Diagnostic accuracy of a clinical carotid plaque MR protocol using a neurovascular coil compared to a surface coil protocol

Background: Carotid plaque imaging with MRI is becoming more commonplace, but practical challenges exist in performing plaque imaging with surface coils. Purpose: To compare the diagnostic performance of a carotid plaque MRI protocol using a standard neurovascular coil (Neurovascular Coil Protocol) to a higher-resolution carotid plaque MRI using carotid surface coils (Surface Coil Protocol) in characterizing carotid plaque. Study Type: Prospective study comparing two MR techniques in plaque characterization. Population: Thirty-eight consecutive carotid artery disease patients. Field Strength/Sequence: Patients underwent 3T MRI using 1) a Neurovascular Coil Protocol including the following sequences: 3D-FSE T₁ pre/postcontrast and precontrast 3D IR-FSPGR, and 2) a Surface Coil Protocol using standard multicontrast MRI sequences. Assessment: Plaque characteristics analyzed by two independent neuroradiologists included intraplaque hemorrhage (IPH), lipid-rich necrotic-core (LRNC), and thin/ruptured fibrous cap (TRFC). Statistical Tests: Diagnostic performance of the Neurovascular Coil Protocol was compared to the Surface Coil Protocol reference standard using receiver-operating curves. Results: For IPH, sensitivity, specificity, and area under the curve (AUC) of the Neurovascular Coil Protocol were 91.1% (95% confidence interval [CI]=78.8-97.5%), 87.0% (95% CI=66.4-97.2%), and 0.92, respectively. For LRNC without IPH sensitivity, specificity, and AUC were 73.3% (95% CI=44.9-92.2%), 85.7% (95% CI=67.3-96.0%), and 0.84, respectively. For TRFC, sensitivity, specificity, and AUC were 35.3% (95% CI=14.2-61.7%), 97.6% (95% CI=87.4-99.9%), and 0.66 respectively. Interobserver agreement for IPH, LRNC, and TRFC using the Neurovascular Coil Protocol were k=0.87 (95% CI=0.75-0.99), k=0.54 (95% CI=0.29-0.80), and k=0.41 (95% CI=0.08-0.74), respectively. Data Conclusion: Our Neurovascular Coil Protocol has high sensitivity, specificity, and accuracy in identifying IPH and LRNC but is limited in assessment of TRFC.