| News | Blog | AI improves colorectal cancer screening

AI improves colorectal cancer screening

The early detection of colorectal cancer plays a crucial role in medicine, as it can significantly increase survival rates. This is where the GI-Vision project comes in, led by Dr. Adrian Krenzer and funded by the bidt. The project's goal was to improve the accuracy and efficiency of colorectal examinations by leveraging cutting-edge artificial intelligence. This blog post provides insight into the project work and its societal relevance.


What is the GI Vision project?

The GI-Vision project developed an advanced AI system that supports gastroenterologists in detecting and classifying polyps – precursors of colorectal cancer – more accurately and quickly. The main task of the AI is to analyse relevant images in real time during the colonoscopy, a camera examination of the bowel. The technology uses neural networks, a form of machine learning, to learn from over 500,000 endoscopic images and thereby to significantly improve the detection of polyps.

Results of the project

The centrepiece of the research, the GI-Vision system, achieved a recognition rate of over 90 percent in the identification of polyps. Small and flat polyps in particular, which can easily be overlooked, were reliably recognised. This high level of accuracy helps to improve diagnosis and to initiate treatment earlier and in a more targeted manner.

An important result was the development of a prototype that is able to process images at a speed of at least 25 images per second and fulfil real-time requirements. This system was successfully integrated into clinical practice at the University Hospital of Würzburg and tested in real patient cases.

Another important feature of the system is the classification of polyps according to size and class. The size of a polyp is a critical factor influencing the treatment strategy, as larger polyps have a higher risk of malignancy and often require immediate removal. The GI-Vision system uses advanced image analysis techniques to predict the size of polyps and provide the treating physician with important information for further action.

Classification of the polyp class is also of great importance, as different types of polyps pose different risks for the development of cancer. The system can distinguish between different classes of polyps that pose a low risk and those that are considered precancerous. This classification helps doctors to better assess the urgency of treatment and create individualised screening plans.

Collaboration and interdisciplinary synergies

The GI Vision project was a prime example of successful interdisciplinary collaboration. Computer scientists from the Julius-Maximilians-Universität Würzburg and doctors from the University Hospital of Würzburg worked hand in hand to overcome technological and clinical challenges. This co-operation not only led to technological innovations, but also ensured that the solutions developed could be used directly in clinical practice.

Social relevance and knowledge transfer

The social relevance of the project is enormous, as colorectal cancer is one of the most common types of cancer worldwide. The early detection and treatment of polyps can significantly reduce the mortality rate. By implementing the GI-Vision system, gastroenterologists can detect polyps more efficiently and accurately, which improves the prognosis for patients and reduces the burden of invasive treatments.

The project has also demonstrated the importance of knowledge transfer between research and clinical application. The results have been actively communicated to healthcare professionals and presented at scientific conferences, leading to wider acceptance and application of the technology.

Future prospects

The technologies and methods developed as part of the GI Vision project have the potential to be transferred to other areas of medical imaging and diagnostics. Future research could focus on developing similar AI-supported systems for the early detection of other types of cancer or chronic diseases.

Conclusion

The GI Vision project shows how the use of artificial intelligence can improve medical diagnostics. The successful collaboration between computer scientists and medical professionals has led to innovative solutions that not only improve diagnostic accuracy, but also increase the efficiency of medical procedures. This project is therefore a shining example of how interdisciplinary research and technological innovation can work together to improve medical practice and ultimately promote public health.

The blog posts published by the bidt reflect the views of the authors; they do not reflect the position of the Institute as a whole.