Use of Artificial Intelligence for Blind Spots During Sedated Versus Unsedated Upper Endoscopies
Vanessa M. Shami, MD, FASGE reviewing Chen D, et al. Gastrointest Endosc 2019 Sep 18.
High-resolution esophagogastroduodenoscopy (EGD) is commonly used in diagnosis and treatment of esophageal, gastric, and duodenal diseases. Per Japanese protocol and European Society of Gastrointestinal Endoscopy guidelines, it is recommended that 26 views should be obtained during an EGD to ensure blind spots are not missed. Moreover, the quality of an EGD exam is influenced by the availability of sedation. This study utilized artificial intelligence (AI) to delineate blind-spot rates during an upper endoscopy of 3 groups: unsedated ultrathin transoral endoscopy (U-TOE), unsedated conventional EGD (C-EGD), and sedated C-EGD.
In this single-blind, 3-parallel-group, randomized, single-center trial, 437 patients were randomized to unsedated U-TOE, unsedated C-EGD, or sedated C-EGD groups. The blind-spot rate with the assistance of AI was significantly lower for the sedated C-EGD group compared to the rates for the U-TOE and unsedated C-EGD groups (3.42% vs 21.77% vs 31.23%; P<0.05). Regardless of the type of EGD or sedation, the AI blind-spot rate was consistently lower than that of the control subgroups (sedated C-EGD: 3.42% vs 22.46%, P<0.001; U-TOE: 21.77% vs 29.92%, P<0.001; unsedated C-EGD: 31.23% vs 42.46%, P<0.001).
Comment:
The blind-spot rate of sedated conventional EGD was the lowest among the 3 types of EGD, with the maximal effect of AI on this subset. While this study suggests AI can improve the quality of endoscopy, further studies are required to assess the true impact of AI on endoscopic procedures.
Note to readers: At the time we reviewed this paper, its publisher noted that it was not in final form and that subsequent changes might be made.
Vanessa M. Shami, MD, FASGE
Citation(s):
Chen D, Wu L, Li Y, et al. Comparing blind spots of unsedated ultrafine, sedated, and unsedated conventional gastroscopy with and without artificial intelligence: a prospective, single-blind, 3-parallel-group, randomized, single-center trial. Gastrointest Endosc 2019 Sep 18. (Epub ahead of print) (https://doi.org/10.1016/j.gie.2019.09.016)
