US Patent 8483454 Methods for tissue classification in cervical imagery

A rule-based unsupervised process for classifying cervical tissue by serially applying classifiers selected from the group that consists of determining size of texture region, opacity parameter, size of acetowhite regions, number of coarse and fine punctations, size of coarse and fine mosaics, size of atypical blood vessels and demographic data, so that the cervical tissue can be classified into no evidence of disease, low-grade dysplasia, high-grade dysplasia or cancer.