SURF-SVM Based Identification and Classification of Gastrointestinal\n Diseases in Wireless Capsule Endoscopy

Endoscopy provides a major contribution to the diagnosis of the\nGastrointestinal Tract (GIT) diseases. With Colon Endoscopy having its certain\nlimitations, Wireless Capsule Endoscopy is gradually taking over it in the\nterms of ease and efficiency. WCE is performed with a miniature optical\nendoscope which is swallowed by the patient and transmits colour images\nwirelessly during its journey through the GIT, inside the body of the patient.\nThese images are used to implement an effective and computationally efficient\napproach which aims to detect the abnormal and normal tissues in the GIT\nautomatically, and thus helps in reducing the manual work of the reviewers. The\nalgorithm further aims to classify the diseased tissues into various GIT\ndiseases that are commonly known to be affecting the tract. In this manuscript,\nthe descriptor used for the detection of the interest points is Speeded Up\nRobust Features (SURF), which uses the colour information contained in the\nimages which is converted to CIELAB space colours for better identification.\nThe features extracted at the interest points are then used to train and test a\nSupport Vector Machine (SVM), so that it automatically classifies the images\ninto normal or abnormal and further detects the specific abnormalities. SVM,\nalong with a few parameters, gives a very high accuracy of 94.58% while\nclassifying normal and abnormal images and an accuracy of 82.91% while\nclassifying into multi-class. The present work is an improvement on the\npreviously reported analyses which were only limited to the bi-class\nclassification using this approach.\n