Automatic Classification of Gastrointestinal Diseases Based on Machine Learning Techniques
Bahaa Saifalnasr Rabi, Omneya Attallah, Mohamed S. Zaghlool, Maha Sharkas · 2019
Gastrointestinal (GI) diseases are common diseases which affect the GI tract. Treatment of GI diseases is quite expensive, complicated and challenging. Machine learning techniques could considerably reduce the cost of examination processes and improve the speed and quality of diagnosis. Therefore, this paper presents an approach to classify GI diseases using machine learning techniques. The proposed method consists of 4 steps; image pre-processing, feature extraction, feature reduction, and classification. First the images are resized. Afterwards, the discrete wavelets transform (DWT) and histograms of oriented gradient (HOG) are used as feature extraction methods. Next, the feature data were reduced by utilizing principle component analysis (PCA) and single value decomposition (SVD). Finally, the reduced features are used to train different classifiers. Results show that classification with complex decision tree achieved an accuracy of 99.8% using SVD method. The results were compared with recent related work. The comparison showed that the proposed approach is capable of diagnosis and classification of GI diseases compared to other work. Thus, it can be used to help medical experiments in investigation procedures.