Began in 2003, the project has resulted in advances in automated analysis methods for video from endoscopy, especially colonoscopy.
Colonoscopy has contributed to a marked decline in the number of colorectal cancer related deaths. Colorectal cancer is the second leading cause of cancer-related deaths in the U.S. Colonoscopy enables the endoscopist to perform a thorough examination of the colon areas that may be difficult to see using other less invasive technologies. More importantly, treatment such as polypectomy can be performed within the same procedure. Quality of colonoscopy is operator dependent and varies significantly. Polyp and cancer have a high probability to be missed under poor quality procedures such as those with a high percentage of blurry images, limited coverage of the colon mucosa, and significant stool covered mucosa.
Outcomes:
- First technology for fully automated capturing of endoscopic procedures. The system works 24-7, capturing over 160 GB of de-identified endoscopic videos daily.
- First technology for automated scoring of quality of colonoscopy: The technology is aimed to assist the endoscopist with real-time feedback to attain an optimal colon inspection during the procedure in order to reduce the polyp miss rate during colonoscopy. Our studies reveal the following.
- Our real-time feedback improves quality of colonoscopy in different aspects for third-year GI trainees. See our ACG 2012 publication.
- Our automated scores correlate well with human-graded quality scores of colonoscopy by GI experts. That is, procedures of good quality receive high automated quality scores whereas procedures of poor quality receive low automated quality scores. The key quality aspects are (1) percentage of the total time which is the withdrawal time, (2) percentage of the withdrawal time with non-blurry images, (3) number of circumference views, (4) cleaning effort by the endoscopist, and (5) hundreds of wall images (with no lumen seen) during the withdrawal phase. See our DDW 2014 publication.
- New content analysis algorithms for polyps, retroflexion, and instruments
First technology for 3D reconstruction of a virtual colon structure from 2D colonoscopy images and video segments. Unlike existing work focusing on reconstruction of 3D colon surface for surgery, we focus on building the structure of the colon with details on alignment of colon folds together with fold width and fold protrusion that may block the view of the endoscope camera. The potential clinical application is to determine the areas inside the colon that are likely unseen by the endoscopist and alert the operating endoscopist to reexamine those areas.- United States Patent No. 7,894,648 “Colonoscopy Video Processing for Quality Metrics Determination,” February 22, 2011
- Establishment of EndoMetric Corporation based on the EMIS technology
Acknowledgment: The research is not possible without the financial support from the following funding agencies. We thank the sponsors for the support on different aspects of the research. Any opinions, findings, and conclusions or recommendations expressed in the publications of the authors are those of authors and do not reflect the views of the funding agencies.
- SEI: Collaborative Research: Endoscopic Multimedia Information Systems (EMIS). National Science Foundation. Award No. 0513809, 051377, 0513582.
- STTR Phase I and IB: Video Analysis Techniques for Computer-Aided Quality Control for Colonoscopy. National Science Foundation. Award No. 0740596.
- NSF-STTR Phase II: Real-time Analysis and Feedback during Colonoscopy to improve Quality. National Science Foundation. Award No. IIP-0956847.
- Improving Colonoscopy Quality through Automated Monitoring (NIH 1R01DK083745 and HS17537)
- Iowa State University, Mayo Clinic Rochester, State of Iowa, and EndoMetric Corp
Major Contributors: Dr. Sean R. Stanek, Dr. Yi Wang, Dr. DongHo Hong, and Dr. Yu Cao
Principal Investigator(s): Prof. Wallapak Tavanapong (Iowa State University), Prof. Johnny Wong (Iowa State University), Prof. Piet C. de Groen (Mayo Clinic, Rochester, MN), Prof. JungHwan Oh (University of North Texas, Denton, TX)