AUTOMATISCHE HEMISPHAREN-ERKENNUNG IN DER ZEREBRALEN RADIONUKLID-ANGIOGRAPHIE

The diagnostic value of brain perfusion studies with radionuclides is based upon the careful determination of intracerebral vascular supply areas in sequential scintigraphic frames. This paper describes an automatic algorithm to determine cerebral hemisphere-borderlines (Regions of Interest). In 134 randomized brain perfusion studies, regions of interest were set by three techniques: automatically by a computer program at the beginning of the venous phase, automatically by using all images derived from the total bolus passage (40 sec.), and manual marking with a light pen at the beginning of the venous phase. A good correlation of these different techniques was obtained (r = 0.87). Furthermore, we found the bilateral perfusion index independent within a wide range, from the size of the marked regions and the quality of the injected indicator bolus. Time saving and lesser carefulness, concomitant with higher precision and reliability, in addition to the standardisation of the marking technique increase the diagnostic value of cerebral radionuclide angiography.