Quantification of Alzheimer-type neurofibrillary lesions by automated image analysis

Neurofibrillary pathology is seen in a wide variety of disorders such as Alzheimer's disease (AD), progressive supranuclear palsy and the Parkinsonism-dementia amyotrophic lateral sclerosis complex of Guam. To assess the pathological importance of these lesions quantitative studies need to be undertaken. To date, most neuropathological studies have been based on qualitative, or at best semi-quantitative, data reporting the presence or absence of specific lesion types. To obtain such data traditionally involves laborious manual measurements, which rely heavily on the skill of the investigator and tend to have low inter- and intra-rater reliabilities. We have developed a novel analysis technique, using colour image analysis, which can accurately quantify the total amount of neurofibrillary damage present. Furthermore we have developed a set of mathematically defined morphological criteria to allow objective discrimination between the three types of neurofibrillary damage seen in the cortex immunostained with Alz-50. Use of this novel technique provides a reliable, rational means for the classification of neurofibrillary lesions. (C) 1996 Academic Press Limited