Analysis of whole slide images of equine tendinopathy

Abstract

We present a method for the automatic analysis of whole slide histological images of equine tendinopathy. This computer-aided analysis is a pre-screening tool that helps veterinarians doctors to evaluate the efficacy of new treatments. A set of textural, arrangement, and alignment features are extracted to reproduce visual histological criteria, each of them representing different feature views of the initial data. To efficiently combine these different views of the data for clustering, tensor-based multi-view spectral clustering is considered and provides an unsupervised classification of the tissue zones. © 2012 Springer-Verlag.