An active area of biological research is the construction of neural atlases and repositories of 3D neural images. The goal is to achieve insight into the structural and functional characteristics of classes of similar as well as dissimilar neurons with a view to understand how cellular structure regulates function. However, at present there is no well- established framework that can compare, analyze, and decompose the morphological and geometric information from the databases quantitatively. Current morphology comparison techniques for graphs are not suitable for this purpose since they frequently impose restrictions on the connectivity and degree of the graphs. More importantly, they do not take into account the geometric similarities between branches which are crucial in identifying similar neurons. In this paper, we develop Path2Path, which achieves a fusion of path-matching and morphology comparison into a common mathematical framework. Path2Path handles arbitrary connectivity and number of edges and decomposes the neurons into a connectivity component and the path resemblance component that aides in distinguishing neurons between different functional classes. Preliminary tests on classes of three neurons show an approximate average interclass to intraclass distance ratio of 2.74.