Guidance alerting endoscopists to adjust their peripheral gaze significantly improved adenoma detection in colonoscopies — particularly for flat lesions — regardless of the endoscopist’s experience level and without increasing procedure time, according to a recent study in Clinical Gastroenterology and Hepatology.

The study was led by Fumiaki Ishibashi, PhD, with the Department of Gastroenterology at the International University of Health and Welfare Ichikawa Hospital in Chiba, Japan. In the study, a real-time eye-tracking system monitored endoscopists’ gaze and prompted them to look at the peripheral area. Four centers in Japan were included, and patients were eligible if they were aged 40-90 years and undergoing colonoscopy for a positive fecal immunochemical test, other screening or diagnostic indications, or post-polypectomy surveillance.

Participants were randomized either to the intervention or control groups using a web-based system. Patients were blinded to group allocation, but endoscopists were not. Both expert endoscopists (who had performed at least 3000 procedures) and nonexpert endoscopists performed the colonoscopies.

Low-Pitched Tone Prompts Gaze Adjustment

In the intervention group, endoscopists heard a high-pitched tone when the gaze fell within a predefined area and a low-pitched warning tone when the gaze wandered outside the area. Conversely, no area of interest was defined in the control group, and a continuous high-pitched tone was heard throughout withdrawal. All endoscopists received a brief information session, during which they were instructed on how to adjust their gaze toward the peripheral area using the eye-tracking and feedback (ETF) system.

There were 198 endoscopists in the intervention group and 199 in the control group. The adenoma detection rate (ADR) was 53.3% in the intervention group and 39.2% in the control group (P < .007). Adenomas per colonoscopy were 1.34 in the intervention group and 0.95 in the control group (P = .046). Polyp detection rate was 66.5% in the intervention group and 47.2% in the control group (P < .001); and peripheral gaze rate (PGR) was 33.7% in the intervention group compared with 25.9% in the control group (P < .001).

However, no improvement was observed in the advanced adenoma detection rate or the sessile serrated lesion detection rate. The authors noted that a subsequent large-scale trial involving endoscopists across diverse settings, particularly low-detector endoscopists, is needed.

“Crucially, the benefit was observed regardless of experience level,” the authors wrote. “Among the 10 endoscopists with baseline PGR < 26%, three were nonexperts. Among them, one nonexpert achieved a significant improvement in PGR, accompanied by a substantial increase in adenoma detection in colonoscopies with improved PGR.”

Limitations of the study included the use of a proprietary ETF system, which may have limited immediate global generalizability. Still, the strong link between baseline PGR and ADR, along with the rapid improvement in PGR after using the ETF system, “indicates that the act of looking at the periphery itself, rather than the device, underlies improved detection. Therefore, the core message — that intentional peripheral gaze is critical for lesion detection — remains valid,” the authors wrote.

Auditory Feedback Easier on Endoscopists

The auditory cues are important in this intervention, Venkata S. Akshintala, MD, PhD, told Medscape Medical News. Akshintala, who is an associate professor of medicine in the Division of Gastroenterology and Hepatology at Johns Hopkins Medicine in Baltimore, noted that guidance tools that involve looking at another screen would interrupt the physician’s focus, whereas warning sounds allow consistent focus.

“This is an important study, and it’s well done,” he said, adding that there hasn’t been good technology yet to assess how well endoscopists are looking at all the quadrants other than measuring withdrawal time. “This is a much more nuanced way to ensure that the endoscopist is looking at all the quadrants,” he said.

The technology addresses a natural habit to focus on the center, he said. “It’s like flying a helicopter, you focus on the center. But with colonoscopy, the entity is not in the center, it’s along the wall.” As procedure room monitors have gotten larger, it has become even more difficult for endoscopists to focus on the periphery, Akshintala said. At Hopkins, he said, a decision was made to return to a smaller screen after experiencing the periphery challenges of larger monitors.

The technology described in the paper “levels the playing field. It lifts everybody to high standards,” he said.

‘Phenomenal’ Potential

“This tech would be phenomenal for those in training or for those who do not have endoscopy as a major part of their practice,” Christina Seo, MD, a colon and rectal surgeon at Holy Name Medical Center in Englewood, New Jersey, told Medscape Medical News. “Having multiple modes of feedback is helpful, and the sound reminders can help train the physician to make peripheral visualization more routine.”

She said the increased polyp detection numbers were “impressive,” adding that though the smaller polyps “may not be dangerous, the overall number of polyps found can change the risk of colon cancer for a patient or can indicate a genetic predisposition for colon cancer or colon cancer syndromes. It does not necessarily help detect more cancers, simply because they tend to be more obvious, but it could help classify patients who need closer surveillance or need genetic testing.”

Seo said that such a system would be a novel technology in the United States. “The closest thing available is an AI system that points out areas on the screen that could be considered a polyp, though it tends to overread,” she said. “I would be interested in trying this tech, if only to see how my routine compares to what this program would alert me to.”

This work was supported by JSPS KAKENHI and Kowa Life Science Foundation. 

Ishibashi and co-author Sho Suzuki, MD, received honoraria for lectures from Fujifilm Corporation and Olympus Corporation. The other authors declared no relevant financial relationships. Seo and Akshintala declared no relevant financial relationships. 

Marcia Frellick is an independent, Chicago-based healthcare journalist and a regular contributor to Medscape.