Advanced algorithm for polyp detection using depth segmentation in colon endoscopy
| Autor: | R P Aneesh, Pooja Soman, M Revathy Nair, R G Devika, Aparna Ratheesh |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Endoscope
medicine.diagnostic_test Computer science business.industry Colorectal cancer ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Colonoscopy HSL and HSV medicine.disease Thresholding digestive system diseases 030218 nuclear medicine & medical imaging Colon polyps 03 medical and health sciences 0302 clinical medicine medicine 030211 gastroenterology & hepatology Segmentation Computer vision Artificial intelligence business Algorithm Correlogram |
| Zdroj: | 2016 International Conference on Communication Systems and Networks (ComNet). |
| DOI: | 10.1109/csn.2016.7824010 |
| Popis: | Colon cancer is a major cause of cancer in women and colorectal polyps are the important cause to colon cancer. Colonoscopy is one of the best method for detecting the colon cancer Colon endoscopy is a technique in which the image of the intestine can be obtained through the camera attached to endoscope and video sequence is further analysed. Algorithms for the automatic detection of polyps are being developed, with texture analysis. In this paper, a novel algorithm is proposed for the detection of polyps. In this paper two types of segmentation methods are adapted. In the first method, linear thresholding is used to detect the saturated region from the HSV image. In the second method, Markovian Random Field is used to segment the image depth-wise. The proposed algorithm is based on extracting certain texture as well as color information from the frames captured by the camera. The proposed algorithm, is very simple, fast and efficient method which is highly helpful for the radiologists in detecting polyps. SVM classifier is used to predict the disease condition using the texture vector and color correlogram vector. The density of the polyp areas are also been estimated. This system is successfully tested with colon endoscopy video images and achieved accuracy of 96.7%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|