In this project, machine learning algorithms will be developed for wireless capsule endoscopy (WCE). WCE is a swallowable technology designed primarily to provide diagnostic imaging of the whole digestive tract. Data collected from this swallowable capsule is in images/videos format. WCE has great advantages over traditional endoscopy because it is portable and easy to use, especially in remote monitoring health-services. However, during the WCE process, the large amount of captured video data demands a significant deal of computation to analyse and retrieve informative video frames. In order to facilitate efficient WCE data collection and browsing task, we aim to develop WCE video summarization framework that can extract representative keyframes of the WCE video contents by removing redundant and non-informative frames. Representative keyframes are the images which clearly shows tissues of gastrointestinal tract.
We have already developed some porotypes and this is a well-defined problem in our group. Our aim is to eliminate redundant frames and classify informative and non-informative frames. Once the redundancy from WCE videos is removed and non-informative frames are detected, we will have a summarized WCE video. This summarized video/data can lead to the quick analysis and diagnosis of the gastrointestinal disorders as listed:
• Stomach pain • Ulcers, gastritis, or difficulty swallowing • Digestive tract bleeding • Changes in bowel habits (chronic constipation or diarrhea) • Polyps or growths in the colon
The ideal candidate should have good knowledge on machine learning. Good programming skills are desirable.