"The incidence of malignant melanoma, the deadliest skin cancer, has been rapidly rising. One of the key features that are used to discriminate malignant melanoma from nevi in dermoscopy skin lesion images is the presence of blotches, also called structureless areas. In this research, variations of blotch features developed in a previous study are investigated for discriminating malignant melanoma from nevi skin lesions. Absolute and relative color methods are explored to extract the blotchy areas from individual skin lesion images. Features are computed from the blotchy areas within each skin lesion image, including eccentricity measures, size of the blotches relative to the size of the skin lesion, irregularity index of the largest blotch, and dispersement index of the blotches. Using a neural network classifier, skin lesion discrimination results are compared for the different approaches for computing features from blotch regions in dermoscopy skin lesion images. The best diagnostic accuracy of 81.2% is achieved based on computing blotch features from blotchy areas determined using fuzzy logic techniques for three plane relative color and size"--Abstract, page iii.
Stanley, R. Joe
Moss, Randy Hays, 1953-
Stoecker, William V.
Electrical and Computer Engineering
M.S. in Computer Engineering
University of Missouri--Rolla
ix, 44 pages
© 2006 Azmath Ullah Khan, All rights reserved.
Thesis - Restricted Access
Melanoma -- Diagnosis
Moments of inertia -- Measurement
Skin -- Cancer -- Diagnosis
Print OCLC #
Link to Catalog Record
Electronic access to the full-text of this document is restricted to Missouri S&T users. Otherwise, request this publication directly from Missouri S&T Library or contact your local library.http://merlin.lib.umsystem.edu/record=b5792741~S5
Khan, Azmath, "The extension of blotch features for melanoma discrimination in dermoscopy images" (2006). Masters Theses. 5932.
Share My Thesis If you are the author of this work and would like to grant permission to make it openly accessible to all, please click the button above.