An improved border detection in dermoscopy images for density based clustering

BMC Bioinformatics

Proceedings

Mutlu Mete¹  Sinan Kockara²  Sait Suer² 

[1] Computer Science and Information System, Texas A&M University, 75429, Commerce, TX, USA;Computer Science, University of Central Arkansas, 201 Donaghey Ave., 72035, Conway, AR, USA;

关键词: Convex Hull;    Lesion Area;    Query Point;    Simple Polygon;    Core Point;   

DOI  :  10.1186/1471-2105-12-S10-S12

来源: Springer

【 摘 要 】

BackgroundDermoscopy is one of the major imaging modalities used in the diagnosis of melanoma and other pigmented skin lesions. In current practice, dermatologists determine lesion area by manually drawing lesion borders. Therefore, automated assessment tools for dermoscopy images have become an important research field mainly because of inter- and intra-observer variations in human interpretation. One of the most important steps in dermoscopy image analysis is automated detection of lesion borders. To our knowledge, in our 2010 study we achieved one of the highest accuracy rates in the automated lesion border detection field by using modified density based clustering algorithm. In the previous study, we proposed a novel method which removes redundant computations in well-known spatial density based clustering algorithm, DBSCAN; thus, in turn it speeds up clustering process considerably.FindingsOur previous study was heavily dependent on the pre-processing step which creates a binary image from original image. In this study, we embed a new distance measure to the existing algorithm. This provides twofold benefits. First, since new approach removes pre-processing step, it directly works on color images instead of binary ones. Thus, very important color information is not lost. Second, accuracy of delineated lesion borders is improved on 75% of 100 dermoscopy image dataset.ConclusionPrevious and improved methods are tested within the same dermoscopy dataset along with the same set of dermatologist drawn ground truth images. Results revealed that the improved method directly works on color images without any pre-processing and generates more accurate results than existing method.

【 授权许可】

Unknown   
© Suer et al; licensee BioMed Central Ltd. 2011. This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.