In recent years, the prevalence of gastrointestinal diseases such as gastric cancer, gastric ulcers and tissue destruction has increased due to lifestyle changes, stress, fast food and air pollution. Endoscopic imaging is an effective method for diagnosing these diseases, but image interpretation requires high expertise and a lot of time, which may lead to human error. To reduce these errors, the use of image processing methods and computer-aided diagnosis (CAD) techniques is proposed. In this study, a new method for diagnosing and classifying gastrointestinal diseases using endoscopic image processing is presented. The proposed method includes the steps of image preprocessing, feature extraction using morphological component analysis, feature enhancement using discrete wavelet transform and dimensionality reduction using principal component analysis. Then, the reduced features are used to classify the images into five different classes including gastritis, ulcer, esophagitis, bleeding and healthy, using a random forest (RF) classifier. The evaluation results on the KVASIR database show that the proposed model achieved 98% accuracy, 99% precision, 98% recall rate, and 99% specificity. These results indicate that the proposed method is accurate and reliable in diagnosing gastrointestinal diseases.
Emadi et al. (2025) studied this question.