Ostrava-based researchers and doctors are developing a tool that could shorten colon cancer surgeries and reduce the risk of complications for patients. Using artificial intelligence and a supercomputer, a 3D model of blood vessels is being created, allowing doctors to see the precise location of arteries and veins in the abdominal cavity before surgery. The model's development is a joint effort between experts from the Faculty of Electrical Engineering and Informatics (FEECS), the National Supercomputing Center IT4Innovations at VSB – Technical University of Ostrava and the University Hospital Ostrava, as part of the LERCO project.
Every year, doctors at University Hospital Ostrava perform around 340 colon cancer surgeries. In these procedures, knowing the exact position of the blood vessels in the abdominal cavity is crucial. “The vascular system in the abdomen is highly variable and differs from person to person. When a surgeon knows its course, they can plan the operation more effectively and reduce the risk of complications. In the case of colon cancer, this also allows us to perform the surgery in a way that is oncologically radical enough to be as effective as possible for the patient,” explains Lubomír Martínek, Head of the Surgical Department at University Hospital Ostrava.
Scientists and doctors are therefore working together to ensure that surgeons have a detailed 3D model of the vascular system available prior to the procedure. “We selected 60 patients, whose CT scans we manually annotated to highlight the blood vessels. It was extremely challenging, as the abdominal cavity contains dozens of arteries and veins. Manual segmentation, in which all vessels larger than 2 mm are identified and marked, took us five to six hours per patient,” explains surgeon Jan Hrubovčák.
The result of the Ostrava medical team’s efforts is a unique dataset now being used by researchers from the Faculty of Electrical Engineering and Computer Science and IT4Innovations. “Our task was to develop a 3D segmentation algorithm based on deep learning. The algorithm learns to recognise vascular structures based on the patterns identified by the doctor,” says Jan Kubíček from FEECS.
The algorithm is being trained on the IT4Innovations supercomputer. “We are developing a neural network that can automatically determine the vascular system in the abdominal area from the patient’s imaging data. While a doctor might spend several hours segmenting the vessels, the current model can do it for arteries in 2.5 minutes, and the same amount of time is needed for veins,” adds Petr Strakoš from IT4Innovations.
The first results look very promising. A 3D model of the arteries is already complete, and the team is now focusing on modelling the veins and developing a specialised model for critical blood vessels located near tumours. The goal is to create a web-based application where doctors can upload a specific patient’s data and receive an accurate 3D model of the vascular system within minutes.
This tool could bring a range of benefits to both doctors and patients. “More precise surgical planning, shorter procedure times, and reduced risk of pre- and post-operative complications directly impact prognosis and a patient’s chances of recovery. If the algorithm could segment blood vessels in real time, it could have wide-ranging applications not only in research but also in everyday clinical practice,” adds Professor Martínek.