The non-invasive detection of intracranial aneurysms: are neuroradiologists any better than other observers?

Can non-neuroradiologists detect intracranial aneurysms as well as neuroradiologists, using CT and MR angiography? Sixty patients undergoing intra-arterial digital subtraction angiography (IADSA) to detect aneurysms also underwent computed tomographic angiography (CTA) and time-of-flight magnetic resonance angiography (MRA). Consensus review of IADSA by two neuroradiologists was the reference standard. Two neuroradiologists, a neurosurgeon, a neuroradiographer and a general radiologist blinded to IADSA, plain CT and clinical data, independently reviewed hard-copy base and reconstructed maximum intensity projection images of the CTA and MRA studies. Thirty patients had a total of 63 aneurysms, 71.4% were </=5 mm in size. Compared with IADSA, mean accuracy per patient for neuroradiologists was CTA 0.87 (95% CI 0.75-0.94), and MRA 0.82 (0.70-0.90); for the other observers it was CTA 0.81 (0.75-0.86), and MRA 0.78 (0.71-0.84). Sensitivity per aneurysm for neuroradiologists was CTA 0.63 (0.50-0.75), and MRA 0.43 (0.6-0.74); for the other observers it was CTA 0.52 (0.44-0.59), and MRA 0.38 (0.31-0.45). Differences between readers and modalities were not statistically significant. Agreement with IADSA was "good" for neuroradiologists: kappa 0.73 for CTA, and 0.63 for MRA. For the other observers, agreement with IADSA was "moderate": kappa 0.59 for CTA, and 0.56 for MRA. Neuroradiologists performed consistently better than the other observers, although differences did not reach statistical significance.