OPTIMIZATION OF FACTORS AFFECTING THE STATE OF NORMALITY OF A MEDICALIMAGE

The purpose of this paper is to examine the first stage of the diagnos tic process in medical imaging, namely determination of the state of n ormality, and to attempt to optimize factors contributing to this stag e. An image of a given type is defined as abnormal if it does not belo ng to the appropriate class of normal images. All images must be pre-p rocessed involving image registration and normalization to align and s cale the images with respect to each other. Normal ranges may be deter mined for each voxel (or other appropriate region) from a representati ve normal sample using univariate analysis, obtaining mean and standar d deviation images, or multivariate analysis, which accounts also for normal patterns of variation (represented as principal components). Fo r a new image, the variation from normality (in SDs) for each region m ay be determined. Since the spatial distribution of this parameter is thought to be relevant, connectivity of abnormal voxels was considered as a possible factor. For the purposes of this study, SPECT images in dicating regional cerebral blood flow were used. Images from 50 normal subjects formed the normal sample. A further 40 normal subjects and 2 00 patients referred with suspected dementia were then analysed using the normal ranges. ROC analysis, using number of SDs as a variable thr eshold, was used to optimize the factors. Normalization to global valu es followed by multivariate analysis using four or five principal comp onents provided optimal discrimination. Connectivity of voxels emerged as an important factor, around 10 connected voxels being optimal for this study.