Automating the Detection of Atypical Pigment Network Using Texture Segmentation
Department
Electrical and Computer Engineering
Major
Computer Engineering
Research Advisor
Shrestha, Bijaya
Advisor's Department
Electrical and Computer Engineering
Funding Source
Missouri S&T Opportunities for Undergraduate Research Experiences (OURE) Program
Abstract
A software program called CVIPtools is used to calculate texture features in skin lesion images with the goal of finding atypical pigment networks in these images. The atypical pigment network is a critical feature in attempting to diagnose melanoma versus benign nevus from an image. Using CVIPtools is a long and arduous process. New software was developed to automate this laborious process. The software uses a map to segment an image for atypical pigment networks. The segmented image would mark the areas with and without atypical pigment networks. The results of the automated process are identical to CVIPtool’s results when using grayscale images, and only 1-5% different when using a color image as the input.
Biography
Cory Gassner completed his primary and secondary education in Jefferson City, MO. He is currently working on his Bachelor’s degree in Computer. He worked as an Undergraduate Research Assistant with the DERMVIS group of skin cancer research from August 2008 to April 2009.
Research Category
Engineering
Presentation Type
Oral Presentation
Document Type
Presentation
Location
Gasconade Room
Presentation Date
08 Apr 2009, 10:00 am - 10:30 am
Automating the Detection of Atypical Pigment Network Using Texture Segmentation
Gasconade Room
A software program called CVIPtools is used to calculate texture features in skin lesion images with the goal of finding atypical pigment networks in these images. The atypical pigment network is a critical feature in attempting to diagnose melanoma versus benign nevus from an image. Using CVIPtools is a long and arduous process. New software was developed to automate this laborious process. The software uses a map to segment an image for atypical pigment networks. The segmented image would mark the areas with and without atypical pigment networks. The results of the automated process are identical to CVIPtool’s results when using grayscale images, and only 1-5% different when using a color image as the input.
