Classification of gastric tissue images based on texture characteristics using the Random Forest method

Abstract

Gastric cancer is a group of malignant diseases caused by many factors, including genetics, lifestyle, and environment. This study aims to create additional tools for distinguishing gastric cancer and normal in microphysical biopsy images from the Kaggle database; the dataset includes 98 gastric cancer and 95 normal. The method used in this research utilizes the coarse and delicate nature of the extracted image based on Histogram and Gray Level Co-occurrence Matrix (GLCM) texture features. Image classification uses the Random Forest method in WEKA software. The results showed that the highest accuracy was 94% in folds 15, 20, and 25, while the lowest accuracy was 93% in folds 5 and 10. This research can be an additional tool for differentiating microphysical biopsy images.