Anormallik tespiti için veri madenciliği

Detection of colon abnormalities is one of the most challenging tasks for gastroenterologists. Colonoscopy is the most common method to record videos and frames from the colon to monitor any abnormality. However, the frames or videos obtained during the procedure are exposed to the significant amount of unwanted artifacts such as motion artifact due to the fast movement of the colonoscopy probe or the capsule, specular reflection (SR) due to the light source used at the probe or in the capsule, improper contrast levels due to insufficient or excessive illumination inside the colon, gastric juice and bubbles, or residuals. The images with such artifacts are called non-informative frames. Disease detection process should be conducted using clear or informative frames. In the first study we investigated the effect of SR and use of image interpolation to remove SR in texture-based automatic polyp detection. For this purpose, we obtained texture features from colonoscopic images with no SR and interpolated images on synthetically added SR with various sizes. We tested whether nearest neighbors, bilinear and bicubic interpolation methods caused any differences in terms of texture features and classification performance to discriminate polyps from the colon background. In the second study the main aim was to compare the performance of conventional machine learning and transfer learning methodologies in detecting non-informative frames. In the machine learning part, we used gray level co-occurrence matrix, gray level run length matrix, neighborhood gray tone difference matrix, focus measure operators and three first order statistics, such as kurtosis, standard deviation, and skewness as features, and random forest, support vector machines and decision tree approaches were used in the classification phase. In the transfer learning part we employed deep neural network architectures like AlexNet, SqueezeNet, GoogleNet, ShuffleNet, ResNet-18, ResNet-50, NasNetMobile, and MobileNet. The last study included the detection of colon abnormalities such as Crohn's, ulcerative colitis, cancer and polyp diseases on informative frames. The aim of this study was first to discriminate healthy frames from diseased ones, and to determine the disease types using both conventional machine learning and transfer learning approaches. We used the same texture features, classification approaches and transfer learning methods as they were employed in the second study.