Disk-caused nerve compression in patients with acute low-back pain: diagnosis with MR, CT myelography, and plain CT.

Ninety-five patients with acute low-back and radicular pain underwent magnetic resonance (MR) imaging and either plain computed tomography (CT) (n = 32) or CT myelography (n = 63) for diagnosis of herniated nucleus pulposus-caused nerve compression (HNPNC). Patients were followed up for at least 6-12 months. Fifty-six patients underwent surgery, and 39 received conservative treatment. Receiver operating characteristic (ROC) analysis was performed on correlation of results of blinded image reading with "true" diagnoses determined by an expert panel [corrected]. Results in subgroup analysis for ROC curve areas were MR, 0.84, versus plain CT, 0.86; MR, 0.81, versus CT myelography, 0.83; and MR, 0.82, versus findings with both CT techniques, 0.85. Results indicate no statistically significant difference in diagnostic accuracy of HNPNC among the three modalities. Thus, factors of cost, radiation dose, and invasiveness influence selection of modality. On the basis of accuracy findings, the authors suggest that MR should replace CT myelography because of the invasiveness of myelography but that MR should not replace plain CT because plain CT is equally accurate and much less costly.