Simulation study of acute subarachnoid hemorrhage using water density images of dual energy CT

The purpose of this study was to evaluate the detection of a minor leak of a subarachnoid hemorrhage (SAH) using material-density images (MDI) in a computer simulation. Iron (Fe) in the structure of hemoglobin was used as a focus for the imaging process, and the MDI were reconstructed using both iron and water as basis materials. The effect of the selection of the two X-ray energy spectra used for data acquisition on the MDI was also investigated. Numerical simulation was performed using both energy spectra and substance attenuation data. An acute SAH was simulated by mixing cerebrospinal fluid (CSF) and blood (hematocrit [Hct], 45%) ranging from 0% to 100%. The single energy computed tomography (SECT) images and the MDI using iron and water as basis materials (Iron density image: IDI and water density image: WDI) were obtained. The computed tomography (CT) values of the SECT images and the values associated with the IDI and WDI were measured for each different blood concentration. Furthermore, the SAH / CSF ratio and contrast to noise ratio (CNR) of the SECT image and the WDI were calculated. In order to investigated the influence on the combination of two X-ray energy spectra used for data acquisition, three combinations of the energy spectra were set and used: 80-120 kVp, 80-150 kVp, 80-150 kVp (0.5 mm tin filter addition: + Sn). The CT number of the SECT increased linearly with increasing blood concentration in the CSF. The iron density of the IDI demonstrated a decreasing tendency. However, the water density of the WDI linearly increased. It was shown that it was difficult to detect a minor leak from a SAH using IDI, but it was detectable using information obtained from the WDI as an indication of the increase in erythrocytes. In the scaled WDI, the slope of the SAH / CSF ratio was larger than SECT. However, the CNR of the SECT image showed a larger value than the scaled WDI. It was shown that the scaled WDI may allow a greater chance of detection of the minor leak of a SAH than the SECT. However, it was also noted that the CNR decreased due to image noise. It was also observed that the greater the separation imposed between the two X-ray energy spectra used for data acquisition, the smaller the standard deviation in the scaled WDI.