AI-based algorithm for clinical decision support system in colonoscopy

AIM: to estimate the implementation of the original method that uses artificial intelligence (AI) to detect colorectal neoplasms.

MATERIALS AND METHODS: we selected 1070 colonoscopy videos from our archive with 5 types of lesions: hyperplastic polyp, serrated adenoma, adenoma with low-grade dysplasia, adenoma with high-grade dysplasia and invasive cancer. Then 9838 informative frames were selected, including 6543 with neoplasms. Lesions were annotated to obtain data set that was finally used for training a convolution al neural network (YOLOv5).

RESULTS: the trained algorithm is able to detect neoplasms with an accuracy of 83.2% and a sensitivity of 77.2% on a test sample of the dataset. The most common algorithm errors were revealed and analyzed.

CONCLUSION: the obtained data set provided an AI-based algorithm that can detect colorectal neoplasms in the video stream of a colonoscopy recording. Further development of the technology probably will provide creation of a clinical decision support system in colonoscopy.

1. Kaprin A.D., Starinsky V.V., Shakhzadova A.O. Malignant neoplasms in Russia in 2021 (morbidity and mortality). Moscow: P.A. Herzen Moscow State Medical Research Institute − Branch of the Federal State Budgetary Institution «NMIC of Radiology» of the Ministry of Health of Russia. 2022; 252 p. (in Russ.).

2. Winawer SJ, Zauber AG. Incidence reduction following colonoscopic polypectomy. Am J Gastroenterol. 2011;106:370. PMID: 21301465. doi: 10.1038/ajg.2010.387

3. Tribonias G, Theodoropoulou A, Konstantinidis K, et al. Comparison of standard vs high-definition, wide-angle colonoscopy for polyp detection: a randomized controlled trial. Colorectal Dis. 2010;12:e260–e266.

4. Hassan C, Piovani D, Spadaccini M, et al. Variability in adenoma detection rate in control groups of randomized colonoscopy trials: a systematic review and meta-analysis. Gastrointest Endosc. 2023;97(2):212–225.e7. doi: 10.1016/j.gie.2022.10.009

5. van Rijn JC, Reitsma JB, Stoker J, et al. Polyp miss rate determined by tandem colonoscopy: a systematic review. Am J Gastroenterol. 2006 Feb;101(2):343–50. PMID: 16454841. doi: 10.1111/j.1572-0241.2006.00390.x

6. Kashiwagi K, Inoue N, Yoshida T, et al. Polyp detection rate in transverse and sigmoid colon significantly increases with longer withdrawal time during screening colonoscopy. PLoS One. 2017 Mar 22;12(3):e0174155. PMID: 28328936; PMCID: PMC5362195. doi: 10.1371/journal.pone.0174155

7. Sui Y, Wang Q, Chen HH, et al. Comparison of adenoma detection in different colorectal segments between deep-sedated and unsedated colonoscopy. Sci Rep. 2022 Sep 12;12(1):15356. PMID: 36097050; PMCID: PMC9468171. doi: 10.1038/s41598-022-19468-y

8. Wang Y, He X, Nie H, et al. Application of artificial intelligence to the diagnosis and therapy of colorectal cancer. Am J Cancer Res. 2020;10(11):3575–98

9. Perone CS, Cohen-Adad J. Promises and limitations of deep learning for medical image segmentation. J Med Artif Intel. 2019;2.

10. Lui TKL, Leung WK. Is artificial intelligence the final answer to missed polyps in colonoscopy? World J Gastroenterol. 2020 Sep 21;26(35):5248–5255. PMID: 32994685; PMCID: PMC7504252. doi: 10.3748/wjg.v26.i35.5248

11. Luo Y, Zhang Y, Liu M, et al. Artificial Intelligence-Assisted Colonoscopy for Detection of Colon Polyps: a Prospective, Randomized Cohort Study. J Gastrointest Surg. 2021 Aug;25(8):2011–2018. Epub 2020 Sep 23. PMID: 32968933; PMCID: PMC8321985. doi: 10.1007/s11605-020-04802-4

12. Wallace MB, Sharma P, Bhandari P, et al. Impact of Artificial Intelligence on Miss Rate of Colorectal Neoplasia. Gastroenterology. 2022 Jul;163(1):295–304.e5. Epub 2022 Mar 15. PMID: 35304117. doi: 10.1053/j.gastro.2022.03.007

13. Xu H, Tang RSY, Lam TYT, et al. Artificial Intelligence-Assisted Colonoscopy for Colorectal Cancer Screening: A Multicenter Randomized Controlled Trial. Clin Gastroenterol Hepatol. 2023 Feb;21(2):337–346.e3. Epub 2022 Jul 19. PMID: 35863686. doi: 10.1016/j.cgh.2022.07.006

14. Areia M, Mori Y, Correale L, et al. Cost-effectiveness of artificial intelligence for screening colonoscopy: a modelling study. Lancet Digit Health. 2022 Jun;4(6):e436–e444. Epub 2022 Apr 13. PMID: 35430151. doi: 10.1016/S2589-7500(22)00042-5

15. Zavyalov D.V., Kashin N.V., Nesterov P.V., et al. Algorithm of clarifying diagnostics and intraluminal endoscopic removal of epithelial neoplasms of the colon. Koloproktologia. 2021;20(1):17–22. (in Russ.). doi: 10.33878/2073-7556-2021-20-1-17-22

16. Bakulin I.G., Rasmagina I.A., Skalinskaya M.I., et al. The use of artificial intelligence for the analysis of endoscopic images in inflammatory bowel diseases. Therapy. 2022;7:7–17. (in Russ.). doi: 10.18565/therapy.2022.7.7-14

17. Wan J, Chen B, Yu Y. Polyp Detection from Colorectum Images by Using Attentive YOLOv5. Diagnostics (Basel). 2021 Dec 3;11(12):2264. PMID: 34943501; PMCID: PMC8700704. doi: 10.3390/diagnostics11122264