Affiliations: Biomedical Imaging and Bioinformatics Lab, Machine Intelligence Unit, Indian Statistical Institute, 203 B. T. Road, Kolkata, West Bengal, 700 108, India {pmaji,sumancse_r}@isical.ac.in
Correspondence:
[*]
Address for correspondence: Machine Intelligence Unit, Indian Statistical Institute, 203 B. T. Road, Kolkata, 700 108, India.
Abstract: Color disagreement among histological images may affect the performance of computer-aided histological image analysis. So, one of the most important and challenging tasks in histological image analysis is to diminish the color variation among the images, maintaining the histological information contained in them. In this regard, the paper proposes a new circular clustering algorithm, termed as rough-fuzzy circular clustering. It integrates judiciously the merits of rough-fuzzy clustering and cosine distance. The rough-fuzzy circular clustering addresses the uncertainty due to vagueness and incompleteness in stain class definition, as well as overlapping nature of multiple contrasting histochemical stains. The proposed circular clustering algorithm incorporates saturation-weighted hue histogram, which considers both saturation and hue information of the given histological image. The efficacy of the proposed method, along with a comparison with other state-of-the-art methods, is demonstrated on publicly available hematoxylin and eosin stained fifty-eight benchmark histological images.
Keywords: Histological image analysis, color normalization, circular clustering, rough sets
