Digital images of multipolar neurons from the human dentate nucleus: topologic and morphometric analysis accompanied with the classification by cluster analysis
The present study shows morphometric analyses of the 2D digital image of a neuron from the human dentate nucleus. Based on the minimum number of parameters that describe three image properties, this study improves the existing anatomical classification of large multipolar neurons. Considering that the complete classification of brain neurons requires more than the morphological description (e.g., the functions), this study can be accepted as preliminary. This research shows morphometric analysis based on computational parameters of neuronal images, previously classified by the anatomical or histological criteria. Furthermore, the correct way of homogenizing the sample and presenting the final results is shown.
The analyses were performed on a sample of 272 microscopic images of dentate nucleus neurons. Each image is characterized by eleven morphometric parameters: six Euclidean and five monofractal. Five of these parameters quantified neuron size, three are shape measures, and three measure the complexity of analyzed neurons. The results suggest a severe uncertainty of the current anatomical/topological classification. A large sample of two groups (116 border and 153 central neurons) differed only in three monofractal parameters, which quantify two properties of neurons. Also, cluster analysis classified neurons into three different groups, which differed significantly concerning all parameters currently used in the research. Finally, we conclude that large multipolar neurons from the human dentate nucleus can be divided into three groups according to their morphology. Further research will provide the answer to how these groups are related to the functional properties of the dentate nucleus.