Newcastle University

Publication

Morphological analyses of histological images using fractal (2004)

Author(s): Marghani K, Dlay S, Sharif B, Sims A

• : Morphological analyses of histological images using fractal

Abstract: In order to assist the decision of the pathologist in cancer diagnosis, a new algorithm using morphological analyses based on fractal geometry is investigated. Samples from two different colorectal tissue types were used, consisting of 24 microscopic images represent moderately differentiated adenocarcinoma, and 22 images of normal tissues. Morphological examinations based on boxcounting method were applied and fractal dimension (FD) were estimated. For validation, statistical analyses for assessing the identification of abnormalities (adenocarcinoma vs. normal mucosa) were performed. Obtained results show a very strong significance of 5.57013*10/sup -10/ using analysis of variance for comparing the means of two different populations of the independent FD observations. The potential for applying morphological analyses of histological microscopic images based on fractal features is established. In brief, quantitative measurements, based morphological features, provide useful information can help decision-making. Further research of combining advanced feature based morphology for pathological diagnosis is required. (21 References).

Notes: Dlay SS Newcastle upon Tyne, UK. Communication Systems, Networks and Digital Signal Processing. CSNDSP 2004. Fourth International Symposium. Newcastle upon Tyne, UK. 20-22 July 2004.

• Short Title: Morphological analyses of histological images using fractal

• Date: 20-22 July 2004
• Conference Name: Communication Systems, Networks and Digital Signal Processing. CSNDSP 2004. Fourth International Symposium. Newcastle upon Tyne Univ. 2004
• Pages: 520-523
• Publication type: Conference Proceedings (inc. abstract)
• Bibliographic status: Published