The future 
of colonoscopy

PolypAID™ - a next generation colonoscopy guidance system. A medical device intended to assist endoscopists by aiding the detection of adenomas, sessile serrated lesions, and hyperplastic polyps in the colon.

Product

Save lives and improve life quality

Cancer takes thousands of lives and reduces quality of life for many survivors. Improved polyp detection can reduce the burden for patients, doctors and society.

Reduce the cost of care

Early detection of colorectal cancers leads to much less aggressive interventions and treatments, leading to better outcomes and a reduced long-term cost of care.

Intelligent real-time video analytics

With video analytics and artificial intelligence, we aim to improve survival rates of colorectal cancer, one of the most common and deadly cancer forms with 2 million annual cases. Our system analyses the entire image, in every frame of the colonoscopy examination. This means that the system can detect polyps that are easily missed due to human factors. The system can identify a polyp within 0.1s.

Unique dataset

We have 941 colonoscopy examination with pathological results collected at Oslo University Hospital. The dataset has been interpreted and annotated by experienced physicians and divided into a machine leaning dataset and a separate test dataset.

Machine learning on colonoscopies

Our model is trained on annotated high quality colonoscopy videos. The technology was born out of leading real-time multimedia systems and has 12 years of research going into it resulting in more than 30 publications. The focus for the machine learning has been to detect hard to discover polyps.

Graphical overlay

PolypAID™ connects to the live video stream between the output from the video processor and the display. The colonoscopy examination passes through the machine with no processing and the AI results are presented as an overlay on the original video signal. Any detected polyps will be outlined by the four corner markings of the boundary box while a green sector indicator lights up in the corner of the colonoscopy image indicating where in the image the polyp has been indicated.

 

 

The sector indicator remain lit for 2 seconds after the initial detection thus serve as a potential missed polyp indicator. This has proved to be particular helpful for polyps that are visible for only a short period of time making it difficult for the operator to spot the marking of the polyp or the polyp itself.

Comparative study

PolypAID™ has high performance and detects relevant polyps with few false positives and thus presents substantial advantages to the other products1.

  • Three expert colonoscopists reviewed and classified 552 video clips, with 375 containing no polyps.
  • Sensitivity – the proportion of correctly identified polyps (positive).
  • Specificity – the proportion of correctly identified non-polyp (negative) clips.
  • Accuracy – the share of correct identifications of both polyp and non-polyp observations.

1. A comparative study benchmarking colon polyp detection with CADe software N. Papachrysos, P. H. Smedsrud, K. Anonsen, T. J. Berstad, H. Espeland, A. Petlund, P. He denström, P. Halvorsen, J. Varkey, H. Hammer, M. Riegler, T. de Lange Endoscopy 2024; 56(S 02): S92-S93 DOI: 10.1055/s-0044-1782889

PolypAID™ can significantly increase detection rates

The best doctors detects prestage polyps in 40% of the patients while others discover prestage polyps in only 10% of the patients1. A 1% increase in ADR has showed to reduce the risk for developing cancer with 3%, thus improving detection rates is critical.

1. Bretthauer et al. Population-Based Colonoscopy Screening for Colorectal Cancer, JAMA (2016).

Compatible with the latest Olympus systems

Currently our device supports the latest Olympus video processors, X1 CV-1500 and EXERA III CV-190. In the future, we aim to connect to any vendor's equipment, be it Olympus, Pentax or Fujifilm to highlight lesions that endoscopists miss.

Real time feedback

Augere gives the endoscopist a real time indication of early-stage lesions being in the field of view.

Transparent AI manifest

Augere Medical is dedicated to providing transparent AI for our patients and clinicians by adhering to these voluntary norms:

  • We will use AI to enhance the overall patient experience and outcomes
  • We will use AI to augment, not replace, the skills and judgment of human clinicians
  • We will provide proper training for clinicians using our AI tools
  • We will provide clear explanations of how our AI algorithms make decisions
  • We ensure that all data used to train our AI systems are ethically sourced and obtained with consent
  • We will test AI systems for safety and efficacy in real-world clinical settings
  • We implement rigorous data security and privacy protection measures
  • We commit to ongoing research and development to enhance our AI technologies

Based on these norms, all data which have been used for machine learning and algorithm testing for our PolypAID™ system have been collected with consent from the patients and the hospital where the colonoscopy was done. All patient information has been kept confidential by Augere Medical from the point of data collection.

The data annotation has been carried out by medical staff hired by Augere Medical.

And finally, all aspects of data gathering, model training, and workforce conditions are available for inspection by customers, stakeholders, and regulators on request.