Computational imaging of cerebral perfusion by real time processing of DSA images. Clinical applications

Real time and high resolution functional imaging of cerebral perfusion was developed, which displays the color coded image of the cerebral perfusion index (PI) from the serial images of digital subtraction angiography (DSA). The small (3 x 3 pixels) regions of interest (ROI) were set on the images, and the time density curves of the contrast media for each ROI were obtained. The perfusion indices, corresponding to the mean transit time (MTT), were calculated for all the ROIs on the image in real time (5 sec) by the area over height method and then, converted to the PI image. A large ROI (64 x 64 pixels) was also used to estimate the parenchymal perfusion. In the normal subjects, consistent PI values at the various parts of the cerebral arteries were obtained, which reasonably agreed with the published values when converted to the blood flow. Then the study on the pathological subjects were made. Firstly, the alteration in the PI images agreed well with the CBF SPECT study in the case of moyamoya disease with the synangiosis operation. Secondly, even a 5% change of the parenchymal perfusion could be detected with the PI imaging, which was obscure when inspecting the series of DSA images, in the case of intra-arterial papaverine infusion for delayed vasospasm. Thirdly, the vortex inside a large aneurysm could be estimated with the same resolution as DSA. Our method offers real time, high resolution, projection angle independent and semiquantitative imaging of the cerebral perfusion from the conventional DSA images without introducing any new expensive devices. This method could be used to evaluate the therapeutic change and especially to monitor the rapidly changing cerebral perfusion in interventional angiography.