Application of Higuchi’s fractal dimension in analysis of images of Anal Intraepithelial Neoplasia

To reduce computational complexity, 2-D images of tumors are firstly preprocessed following our method to form two 1-D sequences, called horizontal and vertical landscapes, and then their fractal dimension is estimated using Higuchi’s fractal dimension. We validate application of this method for grading Anal Intraepithelial Neoplasia (AIN) tumors using 120 images of histological slices graded by an experienced pathologist. The method is easy and quick and we demonstrate that it may be successfully applied for differentiation between AIN stages. It can easily be adapted to other histological specimen. Of course, like any other semiautomatic method it may support the pathologist’s decision.

► New fractal method for grading Anal Intraepithelial Neoplasia tumors is validated. ► The method is based on constructing from the analyzed 2D image two 1D quasi-signals. ► These signals, so called landscapes, are analyzed using Higuchi’s fractal dimension. ► Receiver Operating Characteristic curve analysis is used for grades’ discrimination. ► The method is easy and quick and can be used to support the pathologist’s decision.