Generalized Clustering Methods for Multivariate Data

Efficient analysis (including segmentation and classification) of multivariate data is an inherently complex task in which features occur as salient members of clusters in a multi-dimensional data space. The clusters assume a variety of distributions and frequently overlap, leading to difficulty in segmentation and classification. In many cases, similar but distinct features in the dataset are grouped into the same cluster in global classification. Additionally, partial voluming effects from the acquisition process complicate the reconstruction of accurate spatial features. In this paper, we present a novel framework for the analysis of multivariate datasets. We employ simple linear material models under partial voluming assumption, and recursively classify data samples, thus obtaining both global and regional segmentation in feature space. The simplicity of our approach makes it computationally more efficient compared to other methods for classification of multivariate data. We present results that employ data from light microscopy scanners and the Visible Human repository. CR Categories: K.6.1 [Management of Computing and Information Systems]: Project and People Management—Life Cycle; K.7.m [The Computing Profession]: Miscellaneous—Ethics