The Identification of Stomach Cancer with Semi-Automatic Region Growing Segmentation Method

Abstract

This study was conducted so as to identify the cancerous area in the stomach in a semi automatic way using image processing techniques. The aim was to identify the area suspected of being cancerous by studying the point which the doctor suspected of being cancerous in the patients. Although we cannot say that the whole area which was diagnosed as cancerous, it would be helpful for doctors to identify the area and the site of the biopsy taken. In this way, the results of the biopsy would make it possible to offer more accurate results to the patients and the doctors in order to come up with more precise results. In our study, Accuracy, Sensitivity, Specificity, Confusion Matrix and ROC analysis values were calculated through examination after endoscopic images were taken from the patients. When the results were evaluated, it was seen that sensitivity values of 98.93% were achieved in terms of choosing the proper point, and when irrelevant areas were marked, a sensitivity value of 0% was obtained or an excellent result (92.35%) was obtained through ROC analysis. It can be concluded from the high predictive values applied that the system of semi automatic image processing is fast, reliable and risk-free and therefore could be of help to the physician.