ABSTRACT
Using the bull's eye approach, a reference brain from the single photon emission tomography (SPET) images of 10 subjects aged 62-81 years with excellent mental and physical health was constructed. SPET images were acquired twice, 1 week apart, using a single detector rotating gamma camera collecting 64 planar images over a 360 degrees orbit. The centre of each transaxial slice was first defined with an automatic edge detecting algorithm applied to an anterior-posterior and a side profile of the brain. Each slice was divided into 40 sectors. Maximum counts/pixel in each sector was picked. The 40 maximum count values from one transaxial slice were allowed to form a horizontal row in a new parametric image on the x-axis and slice number from the vertex to the basal parts of the brain on the y-axis. This new image was scaled to a 64 x 16 pixel matrix by interpolation, which meant a normalization of all studies to the same size. The parametric image in each subject was scaled with regard to intensity by a factor calculated by a normalization procedure using the least squares analysis. Mean and SD for each pixel were calculated, thereby constructing a "mean parametric image", and a "SD parametric image". These two images are meant to be used as the reference brain for evaluation of patient studies. This method can be used for objective measurements of diffuse brain changes and for pattern recognition in larger groups of patients. Statistical multifactorial analysis of parameters used for acquisition and data processing is possible.
