Background and study aims: Neoplastic lesions can be missed during colonoscopy, especially when cleansing is inadequate. Bowel preparation scales have significant limitations and no objective and standardized method currently exists to establish colon cleanliness during colonoscopy. The aims of our study are to create a software algorithm that is able to analyze bowel cleansing during colonoscopies and to compare it to a validate bowel preparation scale.
Patients and methods: A software application (the Clean Colon Software Program, CCSP) was developed. Fifty colonoscopies were carried out and video-recorded. Each video was divided into 3 segments: cecum-hepatic flexure (1st Segment), hepatic flexure-descending colon (2nd Segment) and rectosigmoid segment (3rd Segment). Each segment was recorded twice, both before and after careful cleansing of the intestinal wall. A score from 0 (dirty) to 3 (clean) was then assigned by CCSP. All the videos were also viewed by four endoscopists and colon cleansing was established using the Boston Bowel Preparation Scale. Interclass correlation coefficient was then calculated between the endoscopists and the software.
Results: The cleansing score of the prelavage colonoscopies was 1.56 ± 0.52 and the postlavage one was 2,08 ± 0,59 (P < 0.001) showing an approximate 33.3 % improvement in cleansing after lavage. Right colon segment prelavage (0.99 ± 0.69) was dirtier than left colon segment prelavage (2.07 ± 0.71). The overall interobserver agreement between the average cleansing score for the 4 endoscopists and the software pre-cleansing was 0.87 (95 % CI, 0.84 - 0.90) and post-cleansing was 0.86 (95 % CI, 0.83 - 0.89).
Conclusions: The software is able to discriminate clean from non-clean colon tracts with high significance and is comparable to endoscopist evaluation.