Automatic Colon Malignancy Recognition using Sobel & Morphological Dilation

The role of digital image processing in medical science is very advantageous. Colon malignancy is one of the perilous infections which are very hazardous for human health. It starts on the large intestine and later infects other nearest organs of the body, which is lethal if left untreated. Colorectal diagnosis is very expensive if it is not treated timely, so the early phase identification of malignancy is necessary for better health. To diminishing this problem we develop an automated system for recognizing colorectal malignancy in an initial stage. The prime aspire of this framework is to inspect the colorectal CT image to identify whether the colon has malignancy or not. Usually, most of the existing techniques may distort the actual detail that creates false prediction and may reduce accuracy and precision which is very dangerous for patients but a proposed novel approach is capable of accurately detect colorectal cancer at very less processing instant. It consists of different phases namely Pre-processing, Thresholding, Sobel filter, and morphological dilation operation. Sobel algorithm executes a 2-D spatial gradient measurement on the picture and emphasizes the vicinity of high spatial frequency that corresponds to edges. It is easy to apply and gives more accurate edges information about the scene. After that, we apply a morphological operation for extracting picture elements and also advantageous for telling about object shape. The system obtained 98.48% accuracy by testing 198 colon CT samples.