Identification of Image Features for Skin Burn Depth Classification
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
UNIVERSITI MALAYSIA SARAWAK
Abstract
Description
A correct first evaluation of burn injury is critical as the first treatment given to the patient based on the evaluation can lead to a better healing process. However, sometimes mistakes on the first evaluation of the burn injury can happen especially when a specialist is not available on site. Besides that, even an experienced doctor may sometimes be confused with the depth of burn through visual examination. In this thesis, a process of evaluating the burn depths to assist inexperienced doctors and to complement the diagnosis of the doctors and specialists is presented. This research specifically investigates the features that can characterise the burn depths in order to more accurately classify skin burn images. Thus, the main focus in this thesis is to propose a feature extraction process to extract features that can specifically characterise each burn depth. In this study, an image mining approach, specifically image classification process, is used to evaluate the skin burn images and classify the burn images into one of the burn depths. The burns depths studied in this thesis are superficial partial thickness burn (SPTB), deep partial thickness burn (DPTB) and full thickness burn (FTB). The proposed feature extraction process extracted six image features that represent the skin burn depths. These features, when used with common image features, were able to improve the performance in the classification of skin burn depths with an accuracy of 87.88% when tested on unseen data.