A discrete level set approach for texture analysis of microscopic liver images

In this paper the analysis of microscopic liver tissue images is addressed to identify abnormal zones due to the presence of tissue with necrosis, or to malignant lymphoma; the study is performed by texture analysis. A discrete level set approach is considered, applying the well know segmentation algorithm to a new data constituted by a linear combination of the matrices of Uniformity, Contrast and Entropy. The proposed method makes use of the classification capability of the discrete level set analysis applied to a suitable transformation of the original data. The algorithm is applied to a significant set of liver tissue, showing encouraging results.