News |  Sitemap |  Contact
PDF Export  | 

Computer-assisted segmentation and measurement in digital imaging of skin lesions

Abstract:

Introduction

Evaluation of skin lesions is performed visually by assessing different parameters as size, shape, depth color etc. The size of the lesion is an important parameter, indicating changes due to treatment. Therefore an indication of the skin lesion is required, which is typically done manually in digital images of the skin. To overcome the problem of inter- and intraobserver changes in this task an automatic correction was introduced.

Material and Methods

Digital images were acquired with a standard digital camera. The developed method to assist in the task of registration was a live-wire model, were the user indicates a polygon, which roughly corresponds with the border of the skin lesion. The segmentation of the lesion was done by the use of the cost-minimizing algorithm. After segmentation area, diameter were calculated The developed method was implemented on a standard PC using the ImageJ library. Evaluation was done in comparison with manual measurements.

Results

The developed algorithm successfully segmented the skin lesions. In a preliminary test the differences between manual and automated measurement showed a difference in area of less than 2%. The computation and display of the results is performed in less than a second.

Discussion

The developed method proved to be a good assistance in the segmentation and measurement of thee size of skin lesions. Variation due to a manual segmentation are avoided in this method. The user time for interaction is minimized, because the user only has to roughly indicate the skin lesion. Future work will include a clinical study of the tool and further evaluation of the skin lesions.

Keywords:
Computer-assisted detection

Author
Julian Stücker

Organisation   

Homepage   

Short Biography   
Diploma in medical informatics at the University of Applied Sciences Giessen-Friedberg
Competence in software development, image processing

© Luxembourg Institute of Science and Technology | Legal Notice