– Data published in Clinical Gastroenterology & Hepatology indicate that the trial successfully shows the efficacy and safety of Wision AI’s artificial intelligence-based computer-aided polyp detection system during colonoscopy –
SHANGHAI, China, Sept. 29, 2021 (GLOBE NEWSWIRE) -- Wision AI, a leader in developing computer-aided diagnostic algorithms and real-time systems to improve the accuracy and effectiveness of diagnostic imaging, today announced the publication of data demonstrating positive results and competitive advantages of using its artificial intelligence computer-aided polyp detection (CADe) system during colorectal cancer screening in a U.S. patient population. In this world’s first external independent randomized controlled trial (RCT) of AI in the medical field conducted in the United States (NCT03925337), 232 patients were enrolled for screening and surveillance colonoscopy in four leading U.S. academic institutions. Results indicated that the Wision AI‘s CADe device, EndoScreener, significantly reduced the adenoma miss rate (31.25% vs. 20.12%, p=0.0247) and increased adenoma detected per colonoscopy (0.90 vs. 1.19, p=0.0323). The trial data have been published in Clinical Gastroenterology & Hepatology.i
“Our study demonstrates that computer-aided polyp detection has the potential to decrease variability in colonoscopy quality among providers by reducing the miss rate even for experienced physicians,” said senior study author Tyler M. Berzin, MD, director of the Advanced Endoscopy Fellowship at Beth Israel Deaconess Medical Center. “While traditional colonoscopy already detects precancerous polyps earlier and more effectively than stool DNA tests and "virtual" CT scan colonography, the addition of AI polyp detection tools will further support the role of colonoscopy as the gold standard for colorectal cancer screening and prevention.”
This is the world’s first randomized controlled trial of AI technology to conclusively show generalizability by independent, external validation. Patients in the trial were randomly assigned to groups that received either AI-assisted colonoscopy or routine colonoscopy, followed immediately by the other procedure. Institutions involved in the dataset training were excluded from the trial. Investigators evaluated the efficacy of a colon polyp detection system derived from training data sets developed in China and demonstrated high performance of the algorithm in a U.S. patient population undergoing colorectal cancer screening and surveillance. Results indicated that the performance and efficacy of EndoScreener in the trial were consistent with the results of previous RCTs conducted in the originated institution.ii,iii,iv,v
“AI-based systems hold great promise in their ability to improve diagnostic imaging and ensure more accurate detection of adenomas and polyps missed during standard colonoscopy,” said Dr. Sophie Xiao, Chief Scientist of Wision AI. “However, the mathematical essence of deep learning algorithms may lead to the lack of generalization and robustness of AI systems that cannot be cured by increasing training data. The total development dataset of our algorithm contains less than 6000 images that are 100% from a single hospital in west China.vi The EndoScreener system was designed to overcome these challenges, and the data reported in the trial demonstrate that there was no observed performance discrepancy or observed ceiling effect of the system in four leading hospitals in the U.S. We believe that a well designed and rigorously conducted external independent RCT is an ultimate and necessary validation for medical AI applications and we are very pleased to see the positive results of the EndoScreener system in this trial.”
After the exclusion of nine patients, a total of 223 patients entered the study, of whom 59.6% were undergoing primary colorectal cancer (CRC) screening and 40.4% were undergoing post-polypectomy surveillance. Patients were randomized to either CADe (n=113) or HDWL colonoscopy (n=110) first, after which they immediately underwent colonoscopy using the other method. Key findings of the study include:
- A 20.12% AMR in the CADe-first group, compared with 31.25% in the HDWL-first group (p=0.0247).
- A 20.70% polyp miss rate (PMR) in the CADe-first group, compared with 33.71% in the HDWL-first group (p=0.0007).
- A 7.14% sessile serrated lesion (SSL) miss rate in the CADe-first group, compared with 42.11% in the HDWL-first group (p=0.0482).
- An APC of 1.9 in the CADe-first group, compared with 0.90 in the HDWL-first group (p=0.0323).
- The first-pass adenoma detection rate (ADR) was higher with observable difference in the CADe-first group (50.44%) vs. the HDWL-first group (43.64%) (p=0.309).
- Multivariate analysis identified randomization to HDWL-first, age ≤65 years old, and right colon vs. other location and factors significantly associated with missed adenomas (p=0.0214, p=0.0451, and p=0.0436, respectively).
About Wision AI
Wision AI has extensive expertise in mathematics and algorithm development. The company integrates medical knowledge into flexible and scalable models that leverage cutting-edge, convolutional neural networks and general-purpose computing to achieve stable detection efficacy in diagnostic imaging.
References
i Glissen Brown JR, Mansour NM, Wang P, et al. Deep Learning Computer-Aided Polyp Detection Reduces Adenoma Miss Rate: A U.S. Multi-Center Randomized Tandem Colonoscopy Study. Clin Gastrol & Hepatol. September 13, 2021. DOI: https://doi.org/10.1016/j.cgh.2021.09.009
ii Wang P, Berzin TM, Glissen Brown JR, et al. Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a prospective randomised controlled study. Gut. 2019;68(10):1813-1819. doi:10.1136/gutjnl-2018-317500
iii Wang P, Liu X, Berzin TM, et al. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study [published correction appears in Lancet Gastroenterol Hepatol. 2020 Apr;5(4):e3]. Lancet Gastroenterol Hepatol. 2020;5(4):343-351. doi:10.1016/S2468-1253(19)30411-X
iv Liu P, Wang P, Glissen Brown JR, et al. The single-monitor trial: an embedded CADe system increased adenoma detection during colonoscopy: a prospective randomized study. Therap Adv Gastroenterol. 2020;13:1756284820979165. Published 2020 Dec 15. doi:10.1177/1756284820979165
v Wang P, Liu P, Glissen Brown JR, et al. Lower Adenoma Miss Rate of Computer-Aided Detection-Assisted Colonoscopy vs Routine White-Light Colonoscopy in a Prospective Tandem Study. Gastroenterology. 2020;159(4):1252-1261.e5. doi:10.1053/j.gastro.2020.06.023
vi Wang, P., Xiao, X., Glissen Brown, J.R. et al. Development and validation of a deep-learning algorithm for the detection of polyps during colonoscopy. Nat Biomed Eng 2, 741–748 (2018). https://doi.org/10.1038/s41551-018-0301-3
Media Contact
Erich Sandoval, Lazar FINN
Erich.sandoval@finnpartners.com
+1 (917) 497-2867
