Image processing for skin lesion identication

Dissertant

al-Rabii, Zaynah Abd al-Latif Ahmad

University

University of Technology

Faculty

-

Department

Applied Sciences Department

University Country

Iraq

Degree

Master

Degree Date

2006

English Abstract

Nowadays the frequency of skin cancer is rapidly increasing.

A useful treatment is possible but only if an early and precise diagnosis is done.

The automatic image analysis helps dermatologist to give precise and reproducible measures, and to set up a database of the lesions in order to follow their evolution in time.

One of the most common features used in the diagnosis of skin cancer is based on the boundary of the lesion to illustrate the irregularity border of cancer in comparison with another skin lesion.

Principal step is to extract the lesion from the surrounding skin in the image and then draw the histogram to show random distribution of pixels for cancer because of division in cells.

Two segmentation methods were used to extract the lesion from its background.

They are segmentation depending on maximum and minimum threshold values and global segmentation.

These two methods give good information on skin lesions identification.

Many colour systems are applied to select the desired colour system.

Then the usual different approaches are applied to the same selected samples of images of variegated cells for comparison.

Statistical value gives good information about different skin lesions.

A new suggested method was implemented for identification of skin lesions.

The method involves calculating the gradient angle values, histogram of gradient angle, and deviation of gradient.

The result clearly shows that this method could be a useful tool for identification.

Main Subjects

Medicine
Information Technology and Computer Science

Topics

• Cancer
• Digital electronics
• Skin
• Diagnosis

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-305720