Automatic detection of lesion border and edge-related structures in dermoscopy images

"Early detection of malignant melanoma is critical because early stage diagnosis results in a higher survival rate. As pre-processing steps of digital dermoscopy images in an automatic diagnosis system, line structure identification and lesion border identification algorithms are important for accuracy of diagnosis. This work presents the methodology and procedures to process dermoscopy images using computer vision and data mining methods. A watershed-based adaptive skin lesion border finder was developed and implemented for dermoscopy images...An improved SharpRazor algorithm was developed to remove hairs from dermoscopy images, based on a previously existing DullRazorʼ algorithm for hair removal...Work is also reported here on classifiers for detecting atypical pigment network in skin lesions based on texture"--Abstract, page iv.

Name thesis:degree_name

Ph. D. in Electrical Engineering

Grantor

University of Missouri--Rolla

Year dc:date.available

2016

Author dc:creator

• Chen, Xiaohe

• Dermoscopy
• Electrical and Computer Engineering

Repository record dc:identifier

https://scholarsmine.mst.edu/doctoral_dissertations/1752

OAI identifier oai:identifier

oai:scholarsmine.mst.edu:doctoral_dissertations-2754