Acute subarachnoid hemorrhage: in vitro comparison of magnetic resonance and computed tomography

The computed tomographic (CT) attenuation values and magnetic resonance (MR) signal intensities of simulated acute subarachnoid hemorrhage were compared systematically. In vitro MR and CT measurements (T1, T2, and Hounsfield units) were made of mixtures of normal human cerebrospinal fluid (CSF) and normal heparinized blood, ranging from 0% to 100% by volume. The mixtures were measured in a plexiglass phantom with a Siemens DR3 CT scanner for attenuation measurements (Hounsfield units) and in the Baylor Bruker Instruments Proton Scanner (6-MHz) using inversion-recovery and spin-echo pulse sequence techniques for T1- and T2-calculated relaxation times. A PRAXIS II (10.7 MHz permanent magnet) nonimaging unit was used to measure the relaxation times of the CSF/blood mixtures independently for comparison. The Hounsfield measurements of the densest parts of the layered mixtures showed increasing values with increasing amounts of hemorrhage (0% blood, 0 H; 100% blood, 66 H) in a nonlinear pattern. The T1 times of the mixtures decreased with increasing amounts of blood, ranging from 2200 msec to 500 msec for 100% CSF and 100% blood, respectively. The inverse of the T1 relaxation times was proportional to the percentage of blood. The T2 data for the mixtures were similar in character to the T1 relaxation times, except for shorter T2 times at high concentrations of blood. It was concluded the MRI distinguishes varying blood/CSF mixtures on the basis of relaxation times better than does CT on the basis of Hounsfield units. CT still has an imaging advantage, since high-concentration hemorrhage is clearly different from normal brain, while concentrated acute subarachnoid blood has relaxation times similar to normal brain and is nearly isointense on MRI.