Radboudumc in Nijmegen will receive three million euros for research into an AI application for the treatment of cancer. With this new algorithm, AI is promoted from valued help to serious fellow doctor.
“Ai should not mimic what human experts can do, but deliver performance that goes much further.”
–
According to the university, this is an ‘ambitious project’ for the development of an algorithm and a different use of AI than is currently the case. ‘In medicine, AI has so far been mainly used to ease the work of the doctor. With AI, for example, tumors and metastases can be automatically detected and measured, according to Radboudumc. ‘Ai can do that better and faster. The advantages also: AI does not get tired, delivers consistent results and provides extra capacity in the absence of the number of available specialists.’
Much more is possible when AI is applied in a broader and more inventive way, says researcher Geert Litjens of the Pathology department at Radboudumc. According to him, their new algorithm to be built has major consequences for the way AI is used. Litjens: ‘Ai should not only mimic what human experts can do, but also try to deliver performance that goes much further. In the age of personalized medicine, we need to use computational technologies such as machine learning to improve assessment systems designed by pathologists.”
Algorithm
If successful, researchers can, for example, discover new biomarkers (measurable indicators of some biological state or condition) with major implications for clinical oncology and cancer research. ‘To be able to take this big step, a new methodological approach is essential,’ says Litjens.
Litjens’ research group has developed a new algorithm for this innovative approach. It enables full integration of local details and global context of complete images into a single machine learning model. By enriching the algorithm with the latest techniques from computational language processing, it should be possible to make algorithms that are transparent and explainable. The researchers promise a transparent model. Especially because the criticism is often that an algorithm is a black box.
The new applications of AI go beyond detecting and measuring tumors using AI. As a result, for the first time they can also independently take the step towards diagnosis and prognosis. ‘As a result, Ai changes from valued help to serious fellow doctor’, says Litjens.
–
