 |
| Stefan Posch Photo Credit: Uni Halle / Markus Scholz |
"Artificial Professor" is the nickname for a new research project at the University Hospital Leipzig and at the Martin Luther University Halle-Wittenberg (MLU). A team of doctors and bioinformatics wants to use self-learning software to significantly improve the therapy of lymphatic cancer (Hodgkin's lymphoma) in children. The second phase of the multi-year project recently began with 40,000 euros in funding from the Mitteldeutsche Kinderkrebsforschung Foundation.
Children affected by lymph gland cancer can now be cured with modern treatment methods such as chemotherapy and radiation in 95 percent of cases. However, intensive treatment in childhood often leads to late damage. Irradiation in particular increases the risk of developing a second cancer later. Long-term studies show massive over-mortality due to second diseases, such as cancer or heart diseases in adulthood.
Therefore, the primary goal of the medical profession is: only as little treatment as necessary. The data analysis developed in the project, based on artificial intelligence, is intended to help and optimize therapy for each individual patient. In the first phase, the researchers first prepared and prepared a unique data set for the big data analysis: a network of 270 child cancer clinics from 21 countries sent the data from the imaging PET examinations anonymously to Leipzig for years. The three-dimensional image series shows how well individual therapies work and how the tumor tissue develops over time.
These picture series were developed by a team led by the bioinformatician Prof. Dr.-Ing. Stefan Posch from the MLU segmented and a method was developed that automatically distinguishes the tumor tissue from the various non-tumor-related enrichments in the various organs and can thus extract the tumor tissue from the images. "For success, close cooperation with the medical professionals was necessary to bring the different language worlds and scientific approaches into line," explains Posch.
So, phase two has now started. "In this, self-learning software searches for patterns in the previously segmented image data," describes project manager Dr. Thomas Georgi from the University Hospital Leipzig the goal. Because the "artificial professor" with his artificial intelligence is able to learn characteristic peculiarities in the huge amount of image data that the human eye can no longer determine. If successful, the young patients could be treated with even more precise therapies tailored to them and are more likely to be cured permanently.
Source/Credit: Martin Luther University Halle-Wittenberg
ai090622_01