Alzheimer‘s Disease is the most common form of dementia. The morphological correlate for the associated neuro damage is a synaptic loss which appears as a reduction of the dendritic spines along the dendrites of neurons. Therefore research on neural damage associated with Alzheimer‘s Disease requires the counting of large numbers of spines, which currently is performed by hand.
In the present study a software-based system consisting of an effective preprocessing routine and a sophisticated evaluation algorithm for the identification and classification of dendritic spines in fluorescence image stacks of two-photon laser scan microscopy images is presented. The preprocessing routine is implemented in the Java based image processing and evaluation program ImageJ. The dendritic spine evaluation algorithm makes use of the Definiens Cognition Network Technology. The algorithm is divided into four major steps to reproducibly identify the dendritic spines and classify them into the three different classes called ‘mushroom’, ‘stubby’, and ‘thin’. The robustness of the system was evaluated by analyzing different image sceneries.
Due to its high degree of automation the image analysis system presented in this paper offers the possibility to evaluate high amounts of dendrites captured in 3D image stacks regarding their dendritic spine equipment. Thus the system allows a far more detailed and faster examination of the relevant histopathological alterations and thus supports the speed-up of research for a better understanding of Alzheimer’s Disease.