Detection of subarachnoid haemorrhage with magnetic resonance imaging.

OBJECTIVES To measure the sensitivity and specificity of five MRI sequences to subarachnoid haemorrhage. METHODS Forty one patients presenting with histories suspicious of subarachnoid haemorrhage (SAH) were investigated with MRI using T1 weighted, T2 weighted, single shot fast spin echo (express), fluid attenuation inversion recovery (FLAIR), and gradient echo T2* sequences, and also by CT. Lumbar puncture was performed in cases where CT was negative for SAH. Cases were divided into acute (scanned within 4 days of the haemorrhage) and subacute (scanned after 4 days) groups. RESULTS The gradient echo T2* was the most sensitive sequence, with sensitivities of 94% in the acute phase and 100% in the subacute phase. Next most sensitive was FLAIR with values of 81% and 87% for the acute and subacute phases respectively. Other sequences were considerably less sensitive. CONCLUSIONS MRI can be used to detect subacute and acute subarachnoid haemorrhage and has significant advantages over CT in the detection of subacute subarachnoid haemorrhage. The most sensitive sequence was the gradient echo T2*.