EUVIMEDEUVIMED entdecken
Crossref

Use of artificial intelligence device in colorectal cancer screening colonoscopies. Experience in a second-level hospital

E. González de Castro, L. F. Aguilar Argeñal, D. Robles de la osa, P. U. Claudia, E. P. Pablo, C. M. Luis, Á.M. Montero Moretón, L. Pérez Citores, M. Cimavilla Román, M. Curieses Luengo, J. Santos Fernández, F. Santos Santamarta und 5 weitere

Endoscopy · 01.04.2024 · 56 (S 02 S 02) · S406

DOI: 10.1055/s-0044-1783713 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Endoscopy, erschienen am 01.04.2024. Verfasst von E. González de Castro und 16 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Aims

Colonoscopies often miss lesions, a fact associated with interval cancer. Artificial Intelligence (AI) devices could enhance this performance. To compare differences in the adenoma detection rate (ADR) between conventional colonoscopies and AI-assisted ones. Secondary objectives included comparing ADR differences by segments, the mean number of adenomas per colonoscopy (APC) globally and by segments, and the mean exploration time (MET).

Methods

Prospective and randomized study over six months. Inclusion criteria: colonoscopy indicated after a positive FOBT and Boston≥6. Randomization into two groups: conventional colonoscopy (G1) or AI-assisted colonoscopy (G2). An AI device was used to highlight mucosal alterations visually compatible with polyps.

Results

227 patients included: 115 in G1 and 112 in G2. ADR in G1 was 60.9%, in G2 57.1%, with no statistically significant differences (p=0.568). APC was 1.42 (SD: 1.947) in G1, 1.23 (SD: 1.495) in G2, with no statistically significant differences (p=0.423). By segments, ADR was: right colon 26.1% in G1, 27.7% in G2 (p=0.787); transverse colon 15.7% in G1, 11.6% in G2 (p=0.375); left colon 44.3% in G1, 43.8% in G2 (p=0.928). APC was: right colon 0.43 in G1, 0.42 in G2 (p=0.956); transverse colon 0.23 in G1, 0.18 in G2 (p=0.529); left colon 0.70 in G1 and 0.78 in G2 (p=0.661). MET was 22 minutes in G1 and 21.94 minutes in G2, with no statistically significant differences (p=0.960).

Conclusions

In our setting, the use of AI in screening colonoscopies did not modify ADR or APC. One explanation could be that the device used does not distinguish between mucosal patterns (NICE). Nevertheless, not extending the exploration time could make it a useful tool, although more studies are needed. Publication History Article published online: 15 April 2024 © 2024. European Society of Gastrointestinal Endoscopy. All rights reserved. Georg Thieme Verlag KG Rüdigerstraße 14, 70469 Stuttgart, Germany

Zitieren

APA

Castro, E. G. D., Argeñal, L. F. A., osa, D. R. D. L., Claudia, P. U., Pablo, E. P., Luis, C. M., Moretón, Á. M., Citores, L. P., Román, M. C., Luengo, M. C., Fernández, J. S., Santamarta, F. S., Silva, B. A. M. D., Antolín, S. M., Medina, F. J. R., Laguna, J. B., & Millán, A. G. P. (2024). Use of artificial intelligence device in colorectal cancer screening colonoscopies. Experience in a second-level hospital. Endoscopy, 56(S 02 S 02), S406. https://doi.org/10.1055/s-0044-1783713

Vancouver

Castro EGD, Argeñal LFA, osa DRDL, Claudia PU, Pablo EP, Luis CM, et al. Use of artificial intelligence device in colorectal cancer screening colonoscopies. Experience in a second-level hospital. Endoscopy. 2024;56(S 02 S 02):S406. doi:10.1055/s-0044-1783713

BibTeX
@article{Castro2024,
  author = {Castro, E. González de and Argeñal, L. F. Aguilar and osa, D. Robles de la and Claudia, P. U. and Pablo, E. P. and Luis, C. M. and Moretón, Á.M. Montero and Citores, L. Pérez and Román, M. Cimavilla and Luengo, M. Curieses and Fernández, J. Santos and Santamarta, F. Santos and Silva, B. A. Moreira Da and Antolín, S. Maestro and Medina, F. J. Rancel and Laguna, J. Barcenilla and Millán, A. G. Pérez},
  title = {Use of artificial intelligence device in colorectal cancer screening colonoscopies. Experience in a second-level hospital},
  journal = {Endoscopy},
  year = {2024},
  volume = {56},
  number = {S 02 S 02},
  pages = {S406},
  doi = {10.1055/s-0044-1783713}
}
Kostenlos starten →