Fast spoiled gradient-recalled MR imaging of thoracic aortic dissection: Preliminary clinical experience at 1.5 T

The purpose of this study was to evaluate fast spoiled gradient-recalled (FSPGR) magnetic resonance (MR) imaging in the diagnosis of thoracic aortic dissection (TAD). Twenty-eight patients with suspected TAD underwent MR imaging with FSPGR and either cine or cardiac-gated spin-echo MR techniques. The average scanning lime for the FSPGR images was approximately 1 min. Three readers interpreted the FSPGR images for the presence or absence of TAD. An ROC analysis was done. At a specificity of 90%, the sensitivity ranged from 52% to 90% for the three readers. Pulsatility artifacts and mural thrombus were causes of false-positive and false-negative readings. The areas under the ROC curves (A(z)) ranged from 0.85 to 0.97 for the three readers. There was a statistically significant difference in the A(z) values for two of the experienced readers (p = .02). The correct type of dissection was determined in only 65% of the true-positive diagnoses. FSPGR has a very limited role in screening and for rapid evaluation of the unstable patient. The results are reader dependent and susceptible to pulsatility artifacts. Determination of the type of dissection is limited. With a suspected thoracic aortic dissection, therefore, additional imaging sequences should he obtained to maximize accuracy.