The Degree of Skin Burns Images Recognition using Convolutional Neural Network

Abstract

In recent years, Convolutional Neural Network (CNN) model is the stat of art model successful for image analysis. In this research, we suggest integrating CNN model for burn images recognition, one kind of medical images. The aim of this paper is to build to automated computer aided for identifying the degrees of burn images. The burn images dataset has been collected from Burn faculty of Cho Ray hospital, Vietnam and published by the collaboration research project of Cho Ray doctors, and the Information Technology lab of Ho Chi Minh University of Pedagogy. The pre-processing involves Local Binary Pattern (LBP) operators based on the burning expert’s suggestion. CNN model was adapted to be automated method of burn images classification into 4 degrees following the classification of burning patients in Cho Ray hospital. Let it called Burn Convolutional Neural Network (B-CNN). The experimental results showed the feasibility of the proposed model B-CNN. This burns analysis will be helpful for remote hospital in Vietnam where the hospital service must be improved. The remote doctors could use this computer aided module to classify the degrees of burns, and give the suitable medical decision. Keywords: Burn Convolutional Neural Network (B-CNN), Burn Image Classification, Local Binary Pattern (LBP), Object Tracking, Skin Burn Images, Skin Burn Images