Colometer: a real-time quality feedback system for screening colonoscopy

Abstract

Aim: To investigate the performance of a new software-based colonoscopy quality assessment system.

Methods: The software-based system employs a novel image processing algorithm which detects the levels of image clarity, withdrawal velocity, and level of the bowel preparation in a real-time fashion from live video signal. Threshold levels of image blurriness and the withdrawal velocity below which the visualization could be considered adequate have initially been determined arbitrarily by review of sample colonoscopy videos by two experienced endoscopists. Subsequently, an overall colonoscopy quality rating was computed based on the percentage of the withdrawal time with adequate visualization (scored 1-5; 1, when the percentage was 1%-20%; 2, when the percentage was 21%-40%, etc.). In order to test the proposed velocity and blurriness thresholds, screening colonoscopy withdrawal videos from a specialized ambulatory colon cancer screening center were collected, automatically processed and rated. Quality ratings on the withdrawal were compared to the insertion in the same patients. Then, 3 experienced endoscopists reviewed the collected videos in a blinded fashion and rated the overall quality of each withdrawal (scored 1-5; 1, poor; 3, average; 5, excellent) based on 3 major aspects: image quality, colon preparation, and withdrawal velocity. The automated quality ratings were compared to the averaged endoscopist quality ratings using Spearman correlation coefficient.

Results: Fourteen screening colonoscopies were assessed. Adenomatous polyps were detected in 4/14 (29%) of the collected colonoscopy video samples. As a proof of concept, the Colometer software rated colonoscope withdrawal as having better visualization than the insertion in the 10 videos which did not have any polyps (average percent time with adequate visualization: 79% ± 5% for withdrawal and 50% ± 14% for insertion, P < 0.01). Withdrawal times during which no polyps were removed ranged from 4-12 min. The median quality rating from the automated system and the reviewers was 3.45 [interquartile range (IQR), 3.1-3.68] and 3.00 (IQR, 2.33-3.67) respectively for all colonoscopy video samples. The automated rating revealed a strong correlation with the reviewer's rating (ρ coefficient= 0.65, P = 0.01). There was good correlation of the automated overall quality rating and the mean endoscopist withdrawal speed rating (Spearman r coefficient= 0.59, P = 0.03). There was no correlation of automated overall quality rating with mean endoscopists image quality rating (Spearman r coefficient= 0.41, P = 0.15).

Conclusion: The results from a novel automated real-time colonoscopy quality feedback system strongly agreed with the endoscopists' quality assessments. Further study is required to validate this approach.

Keywords: Bowel preparation; Colon cancer; Colonoscopy; Quality assurance; Quality improvement; Withdrawal time.

Figure 1

Conceptual block diagram of the proposed colometer system.

Figure 2

Relative withdrawal speed over time. Red line is the speed threshold configured by the endoscopists.

Figure 3

A sample frame with focused (left) and poor image clarity. Text indicators “adequate speed”, “rapid withdrawal”, “blurry”, were embedded into the video in a real-time to provide a visual feedback to the endoscopist.

Figure 4

An automatically generated chart providing % of time of adequate speed in a colonoscopy procedure (left) and chart showing filtered velocity over time (right).

Figure 5

Comparative results (withdrawal vs insertion) for 10 colonoscopy procedures. The output mean values are marked with a cross.

Figure 6

Correlation between the overall quality ratings from the Colometer system. A: Mean endoscopist overall quality rating (ρ = 0.65, P < 0.01); B: Mean endoscopist withdrawal speed rating (ρ = 0.59, P < 0.01); C: Mean endoscopist image quality rating (ρ = 0.41, P = 0.15).