Validation of automated brain contour determination in normal and abnormal cerebral single-photon emission tomography

A contour detection algorithm for cerebral studies, using the method of Tomitani, has been implemented on a single-photon emission tomographic (SPET) system. It is based on the detetion by threshold of the brain edge in the sinogram and does not depend on the reconstruction algorithm. Thirteen normal subjects underwent an examination on both computed tomography (CT) and SPET using a head holder to ensure the reproducibility of the positioning. The CT scan contour of the brain was drawn manually according to the brain parenchyma limits. The SPET brain contour was obtained by use of the Tomitani algorithm after the threshold had been determined on an active cylindrical phantom. Using a threshold of 37% of the maximum uptake, the length of the contour as well as the area obtained with SPET and CT were not found to be statistically different. The method of Tomitani, which is simpler and faster then previous methods, provides contours which superimpose very well with CT scan images. Application to patients with unilateral pathological defects is possible by requiring that the contour is symmetrical.