Paper
Computer-Aided Detection of Gastrointestinal Disorders with Endoscopy Capsule Images
- Authors:
- Faria L.; Diogo D.; Moreira R.; Gon?alves G.
- Abstract
- The endoscopy capsule (EC) is a reliable, secure, and non-invasive tool for diagnosing disorders in the gastrointestinal tract (GIT). To aid in early detection diagnostic efficiency for various disorders in difficult-to-reach areas of the GIT, the authors developed the PhotonicPill EC. This EC uses photonics technology to facilitate a more accurate and timely diagnosis of various pathologies. The main objective of this research project was to develop a clinical decision support system that uses EC-acquired data for the early identification of bleeding and polyps, and for the detection of vascularization inside the GIT. The bleeding detection method was based on the evaluation of minimum and maximum values for each RGB (red-green-blue) channel, combined with morphological and local operations. Regarding the polyp detection method, segmentation was accomplished with a marker-controlled watershed transform, using measures related to the typical shape of a polyp for decision purposes. The vascularization detection algorithm used the calculation of maximum values for each channel to select a maximum threshold value for segmentation of the vascularization area. For validation of the developed methods, the EC images with known diagnoses were used as ground truth. The results showed very satisfactory percentages of classification, which fulfilled the objective for bleeding, polyps, and vascularization detection.
- Keywords
- Endoscopy Capsule; Bleeding; Early Detection; Polyps; Image Processing; Photonic Pill
- StartPage
- 30
- EndPage
- 37
- Doi
- 10.5963/PHF0404001