Intel Labs and the Perelman School of Medicine at the University of Pennsylvania (UPenn) have come together to develop a technology to train AI models to identify brain tumors while prioritizing preserving privacy. The technique using a distributed ML approach is called federated learning, which enables organizations to collaborate on deep learning projects without disclosing patient data.
The Information Technology for Cancer Research (ITCR) program of the National Cancer Institute (NCI) is funding Penn Medicine’s work through a three-year, $1.2 million grant.
Intel labs and Penn Medicine claimed to be the first to publish a paper on medical imaging federated learning.
The companies first published the research at the International Conference of Medical Image Computing and Computer Assistant Intervention (MICCAI) 2018. The model trained with a federated learning method was over 99% accurate to a model trained under traditional, non-private methods.
Both companies claim the work built on this new method will leverage Intel hardware and software to provide impenetrable privacy protection to both the model and the data.
The two companies are coordinating with 29 healthcare and research institutions belonging to seven countries.
Jason Martin, principal engineer, Intel Labs, said, “AI could be helpful in early detection of brain tumor, but will require more data to reach its full potential than any single medical center holds.”
AI has always been a significant contributor to the Healthcare industry. Babylon Health believes that it can appropriately triage patients in 85% cases, and Microsoft disclosed details of a $40 million worth “AI for Health” project.
#AIMonks #IntelLabs #UPenn #ArtificialIntelligence #AI #ITCR #NCI # ML # MachineLearning #DeepLearning #Healthcare #FederatedLearning #Research
0 Comments