A feature extracting method using gradient and watershed for tonsillitis diagnosis

Tonsillitis disease is one of major causes intervening for heart attack and pneumonia. Nowadays, it is still difficult to make a proper pathological diagnosis of edge detection into the tonsillitis image because borderline grades are very similar to each other and their diagnosis often leads to a considerable variability. To decrease the burden of those diseases, the effective tonsillitis detection system is required. This paper proposes a brief overview of tonsillitis detection system for objective-analysis of the proper identification of tonsil gland by using a combination of gradient and watershed segmentation method as a fundamental tool. The methods include three main steps starting from calculating gradient image value, then using watershed technique based on mean intensity value to segment the image regions and to detect their boundaries, and finally merging two procedures by making image for calculating the number of pixel of each region. The software is implemented using MATLAB. The method used in this study has perform very well and yield better performance as it achieved the accuracy at approximately 90% comparing with results from the medical doctor. Thus, this study is utilized and will be effective tool for diagnosis tonsillitis disease as the result.