A Systematic Review: Impact of Preprocessing and Segmentation Methods in Classification of Celiac Disease
In this era of technology, endoscopy emerges as the clinically acceptable imaging modality in the detection and diagnosis of celiac disease. As cases of celiac disease are increasing day by day, so it is necessary to make detection automated. The first step of preprocessing and then image segmentation is considered as the most essential step in research before classifying images, in the last two decades. To process the very large size images, image preprocessing is done, and to divide the image into desirable building blocks, image segmentation is used. As the availability of algorithms is vast, so the analysis of these algorithms might be interesting for researchers. This study compares the results of preprocessing and segmentation methods used. The images are processed using MATLAB and various pictorial results are presented based on image resizing, RGB to gray, edge-based, and ostu thresholding, based on their advantages and disadvantages. The unique aspect of the work is how the two modified filters are employed for segmentation to appropriately assess the intestinal wall. This work compares classification accuracy, sensitivity and specificity with other segmented and nonsegmented images. It is proved that using modified filters, celiac disease can be diagnosed without using a complicated deep learning algorithm or any state-of-art model.