AI IN GI: THE BASICS

This article is a summary of a physician podcast discussion around AI in healthcare. The contents and conclusions of the following article are solely those of the speaker, unless otherwise cited. The speaker received funding from Covidien LP, a Medtronic company, for this engagement.

When we talk about artificial intelligence (AI) in gastroenterology, we are talking about “smart” machines and software that can act as a secondary observer for healthcare professionals. One application is to help increase the ability to detect and diagnose suspicious lesions. But what are these different systems, and how do they work?

Computer-aided detection (CADe) and computer-aided diagnosis (CADx) are part of a class of AI systems that help physicians in the analysis of medical images. One of AI’s distinct strengths is the ability to sift through and analyze data with incredible efficiency. The goal of CADe and CADx systems is to improve the accuracy of physicians by reducing the amount of time it takes to interpret images.

For AI-assisted colonoscopy, CADe systems are designed for the identification of lesions in endoscopic images. Typically, they are powered by neural networks that have been trained and validated using thousands of parameters and millions of images of histologically confirmed polyps. They can output processed images in several microseconds and superimpose a digital box over suspected lesions.1

CADx systems are a bit different in that they are geared toward the classification of lesions. This kind of technology, at least for colonoscopy, is not available in clinical practice. A future opportunity is a functioning CADx system that would be able to classify polyps. This is based on robust data with variation of lesion characteristics.2

If AI can already detect lesions effectively, and diagnosis is on the horizon, is the role of the physician being diminished? Well, not so fast. There are some important things to keep in mind if this is a concern:

  • AI is best as a complementary tool for healthcare professionals to increase their efficiency and effectiveness; the two work much better together than individually.
  • Humans think and work in a non-linear manner that quickly adapts to changing conditions and environments; this kind of suppleness is less natural for AI.
  • AI, by its very nature, is not human. It cannot be present, empathetic, and interactive in the way humans can.

Another common concern of physicians regarding CADe and CADx systems is the idea that they might be overzealous in detecting lesions or making false-positive diagnoses. AI-powered tools can balance the false negative and false positive rates. Technology should be enhanced with strong testing and evaluation of accuracy before implementation. Early data shows that the accuracy is clinically relevant and false positives are quickly ruled out without adding any significant time to the procedure.3,4

Ultimately, physicians want the best for patients and to deliver the best possible outcomes. AI can be a tool for healthcare professionals to complement their current practices.  

These types of technologies might enhance our abilities as a physician to make increased therapeutic decisions and improve patient outcomes. It is an exciting time to be part of the future of AI.

PHYSICIAN PODCAST SERIES

LISTEN NOW

SIGN UP FOR MORE

Become a member and embark on the future of care.

SIGN UP NOW

AI IN GI HEALTH

Learn how AI is enhancing care and patient outcomes.

LEARN MORE

OUR AI SOLUTIONS

Explore the technology we have right now for AI.

VIEW NOW
1

GI Genius™ intelligent endoscopy module [instructions for use]. Mansfield, MA: Medtronic; 2019.

2

Repici, A., Wallace, M. B., East, J. E., Sharma, P., Ramirez, F. C., Bruining, et al. (2019). Efficacy of Per-oral Methylene Blue Formulation for Screening Colonoscopy. Gastroenterology, 156(8), 2198–2207.e1.

3

Hassan, C., Wallace, M. B., Sharma, P., Maselli, R., Craviotto, V., Spadaccini, M., et al. (2020). New artificial intelligence system: first validation study versus experienced endoscopists for colorectal polyp detection. Gut, 69(5), 799–800.

4

Lee, J. Y., Jeong, J., Song, E. M., Ha, C., Lee, H. J., Koo, J. E., et al. (2020). Real-time detection of colon polyps during colonoscopy using deep learning: systematic validation with four independent datasets. Scientific reports, 10(1), 8379.

Disclaimer: All content from healthcare professionals is their individual conclusions, unless otherwise cited. All speaker or author engagement for content is noted to acknowledge funding from Covidien LP, a Medtronic company, for any consulting engagement.

Sign Up

Sign up for new AI education resources.

Sign Up