Investigating the Morphological Categories in the NeuroMorpho Database by Using Superparamagnetic Clustering

The continuing neuroscience advances, catalysed by multidisciplinary collaborations between the biological, computational, physical and chemical areas, have implied in increasingly more complex approaches to understand and model the mammals nervous systems. One particularly important related issue regards the investigation of the relationship between morphology and function of neuronal cells, which requires the application of effective means for their classification, for instance by using multivariated, pattern recognition and clustering methods. The current work aims at such a study while considering a large number of neuronal cells obtained from the NeuroMorpho database, which is currently the most comprehensive such a repository. Our approach applies an unsupervised clustering technique, known as Superparamagnetic Clustering, over a set of morphological measurements regarding four major neuronal categories. In particular, we target two important problems: (i) we investigate the coherence between the obtained clusters and the original categories; and (ii) we verify for eventual subclusters inside each of these categories. We report a good agreement between the obtained clusters and the original categories, as well as the identification of a relatively complex structure of subclusters in the case of the pyramidal neuronal cells.