August 26, 2022 – An algorithm developed by researchers from Helmholtz Munich, the Technical University of Munich (TUM) and its university hospital rechts der Isar, University Hospital Bonn (UKB) and the University of Bonn is able to learn independently across different institutions. The main characteristic is that it is “self-learning”, i.e. it does not require long and time-consuming discoveries or markings by radiologists in the MRI images. This federated algorithm was trained on more than 1,500 MRI scans of healthy study participants from four institutions while maintaining data confidentiality. The algorithm was then used to analyze more than 500 MRIs of patients to detect diseases such as multiple sclerosis, vascular disease and various forms of brain tumors that the algorithm had never seen before. This opens up new possibilities for developing efficient federated AI-based algorithms that learn autonomously while protecting privacy. The study has just been published in the journal Nature Machine Intelligence.
Healthcare is currently being revolutionized by artificial intelligence. With accurate AI solutions, physicians can be aided in diagnosis. However, such algorithms require a considerable amount of data and the results of associated radiological specialists for training. The creation of such a large and central database, however, imposes special requirements in terms of data protection. In addition, creating results and annotations, for example marking tumors in an MRI image, is time-consuming. To overcome these challenges, a multidisciplinary team from Helmholtz Munich, University Hospital Bonn and the University of Bonn collaborated with clinicians and researchers from Imperial College London and TUM and its University Hospital rechts der Izar. The goal was to develop an AI-based medical diagnostic algorithm for brain MRI images, without any data being annotated or processed by a radiologist. In addition, this algorithm had to be trained “at federal level”: In this way, the algorithm “comes to the data”, so that medical image data requiring special protection can remain in the respective clinic and have no not to be collected centrally.
Learn from multiple institutes without data exchange
In their study, the researchers were able to show that the federated AI algorithm they developed outperformed any AI algorithm trained using only data from a single institution. “In his ‘The Wisdom of Crowds’, James Surowiecki argued that large groups of people are smarter, no matter how smart an individual may be. Basically, that’s how our federated AI algorithm works” , says Professor Shadi Albarqouni, Research Professor of Computational Medical Imaging in the Department of Diagnostic and Interventional Radiology at University Hospital Bonn and Head of the Helmholtz AI Junior Research Group at Helmholtz Munich. To pool knowledge about brain MRI images, the research team trained the AI algorithm in different independent medical institutions without violating data privacy or collecting data centrally. “Once this algorithm learns what MRI images of the healthy brain look like, it will be easier for it to detect disease. Achieving this requires smart computing aggregation and coordination between participating institutes,” says Professor Albarqouni. PD Dr. Benedikt Wiestler, Chief Physician at TUM rechts der Isar University Hospital and also involved in the study, adds: “Training the model on data from different centers contributes significantly to the fact that our algorithm detects diseases much more robustly than other algorithms that are only trained with data from a single center.”
Towards affordable collaborative AI solutions
By protecting patient data while reducing the workload of radiologists, the researchers believe their federated AI technology will significantly advance digital medicine. “AI and health care must be affordable, and that is our goal. With our study, we have taken a step in this direction,” says Professor Dr Albarqouni. “Our main goal is to develop AI algorithms, trained collaboratively in different decentralized medical institutes, including those with limited resources.”
For more information: https://www.helmholtz-munich.de/en/helmholtz-zentrum-muenchen/index.html