Carolina Mangas-Sanjuan, MD, PhD, discusses findings from her research on the use of artificial intelligence computer-aided detection in colonoscopies in this Q&A with HCPLive.
A recent study involving more than 3200 patients from 6 sites in Spain is providing clinicians with important insight about the viability of artificial intelligence computer-aided detection (CADe) in colonoscopies.1
Results of the parallel, controlled randomized trial found no difference in the detection of advanced colorectal neoplasia in colonoscopies with and without CADe, highlighting the technology’s shortcomings when it comes to detecting advanced lesions.1
“The current findings are a snapshot of what these systems can currently offer and what can be expected from them. Detecting more advanced lesions still lies in the hands of experienced endoscopists who can recognize the lesions and achieve adequate mucosal exposure,” wrote investigators.1
Named the CADILLAC trial, the study enrolled patients with a positive fecal immunochemical test to evaluate the contribution of CADe to colonoscopic detection of advanced colorectal neoplasias, adenomas, serrated polyps, and nonpolypoid and right-sided lesions. Participants were part of a Spanish colorectal cancer screening program and were randomly assigned to a colonoscopy with or without computer-aided detection during the withdrawal phase of the procedure.1
Upon analysis, investigators found no significant difference between the intervention group and the control group for advanced colorectal neoplasia detection rate (34.8% vs 34.6%; adjusted risk ratio, 1.01; 95% Confidence Interval [CI], 0.92-1.10), mean number of advanced colorectal neoplasias detected per colonoscopy (0.54 [Standard Deviation [SD], 0.95] vs 0.52 [SD, 0.95]; adjusted rate ratio, 1.04; 99.9% CI, 0.88-1.22), or adenoma detection rate (64.2% vs 62.0%; adjusted risk ratio, 1.06; 99.9%, 0.91-1.23).1
CADe was associated with a slight increase in the mean number of nonpolypoid lesions and proximal adenomas detected per colonoscopy. Investigators observed similar results in lesions ≤5 mm including polyps, adenomas, and serrated lesions.1
To learn more about the study and the future of CADe in colonoscopies, HCPLive reached out to study investigator Carolina Mangas-Sanjuan, MD, PhD, of the Digestive System Service of the General University Hospital of Alicant, for more information about the findings of her investigation as they pertain to the use of artificial intelligence in clinical practice.1
HCPLive: What are some of the larger implications for the results of this study when it comes to using AI in the field?
Carolina Mangas-Sanjuan: The main implications of our study is that, for the first time, we have demonstrated that AI does not improve advanced colorectal lesions detection in a very specific context which is FIT-based colorectal cancer screening program. Our results show that it is necessary to continue improving this technology by using broader image databases to train computer-aided detection devices to recognize these types of lesions.
HCPLive: What responsibilities can AI be trusted with in clinical practice today?
Carolina Mangas-Sanjuan: AI has been shown to improve the adenoma detection rate (percentage of colonoscopies in which we identify an adenoma), and also the average number of adenomas detected in each colonoscopy. This is a very relevant aspect, since ADR has been shown to be inversely related to colorectal cancer incidence and mortality. However, this tends to occur in settings where there are lower detection rates and at the expense of small lesions with less clinical relevance.
HCPLive: What tasks might this technology not yet be ready to handle?
Carolina Mangas-Sanjuan: There is significant room for improvement for this promising technology and there are many questions to be answered. First of all, it is necessary to evaluate whether these AI devices allow the colorectal cancer incidence reduction, which is, ultimately, what we want for our patients. On the other hand, whether these devices are more useful in other groups of endoscopists with lower detection rates, as well as whether they are of help in groups of patients with a higher risk of cancer, such as those with hereditary cancer syndromes. On the other hand, which role these devices play in improving colonoscopy quality indicators, such as colon cleansing or cecal intubation. Likewise, also the role of AI to improve characterization and optical diagnosis of lesions.
HCPLive: Looking ahead, where do you see room for improvement in this technology to make it a more viable resource in clinical settings in the future?
Carolina Mangas-Sanjuan: Artificial intelligence is here to stay. It is not a matter of making an indiscriminate use of these devices, but we must continue to look at what aspects and in which contexts it can be most useful, and I am sure we will succeed.
HCPLive: Aside from improving the technology, how might clinicians themselves need to adapt in order to integrate AI into their workflow and make it work in practice?
Carolina Mangas-Sanjuan: This is an area where we still have a lot to learn. Above all, what I think is most important is how we integrate these devices in trainees. Specifically, regarding computer-aided detection devices, it is essential that we learn to look for lesions, because if we do not show the mucosa to the device, it will never be able to see those lesions.
1. Mangas-Sanjuan C, de-Castro L, Cubiella J, et al. Role of Artificial Intelligence in Colonoscopy Detection of Advanced Neoplasias : A Randomized Trial [published online ahead of print, 2023 Aug 29]. Ann Intern Med. 2023;10.7326/M22-2619. doi:10.7326/M22-2619