Welcome to the Qinsun Instruments Co., LTD! Set to the home page | Collect this site
The service hotline


Related Articles

Product Photo

Contact Us

Qinsun Instruments Co., LTD!
Address:NO.258 Banting Road., Jiuting Town, Songjiang District, Shanghai

Your location: Home > Related Articles > Intel helps develop new AI to train and discover brain tumors without infringing on privacy

Intel helps develop new AI to train and discover brain tumors without infringing on privacy

Author:QINSUN Released in:2024-01 Click:97

Intel and the University of Pennsylvania School of Medicine are researching a massive, cross institutional artificial intelligence that will help identify brain tumors without exceeding strict medical privacy rules. This cross institutional artificial intelligence will use a technology called "federated learning" as it spans 29 different medical and research institutions.

Training artificial intelligence using disease datasets to enable a large number of cases to act as filters has shown effective effectiveness in many aspects. However, its disadvantage is that in order to achieve effective performance, it requires a considerable dataset. A single medical institution or research laboratory may find it difficult to provide all the necessary information for the machine learning being developed.

Dr. Spyridon Bakas from the Center for Biomedical Image Computing and Analysis (CBICA) at the University of Pennsylvania explains, "Machine learning training requires a large and diverse amount of data. Although technically not a challenge, the reality is that health privacy laws (whether HIPAA, GDPR, or other laws) limit the content that can be shared, which is a bottleneck in big data processing."“

Intel and Pennsylvania Medical School's answer to this is joint learning. Instead of sharing individual patient records, it is better to distribute an encrypted machine learning model to each participating institution. It decrypts in a secure zone on each computer and receives local data training. Subsequently, only the model updates will be shared with the organization responsible for summarizing the models. Due to the fact that patient data never leaves a separate institution, it is more private, and the retrained model data is smaller, making data transmission more efficient.

The Pennsylvania School of Medicine and 29 medical and research institutions from the United States, Canada, the United Kingdom, Germany, the Netherlands, Switzerland, and India will collaborate to develop a method for identifying brain tumors using AI using this joint learning system running on Intel hardware. This year, the alliance will begin developing algorithms to identify brain tumors from a significantly expanded version of the Brain Tumor Segmentation (BraTS) dataset. This alliance will allow medical researchers to access a large amount of medical data while protecting its security.

The American Brain Tumor Association stated that nearly 80000 people will be diagnosed with brain tumors in 2020. After training on raw MRI data, the AI model developed by Intel and the Pennsylvania Medical Association can achieve 99% accuracy when identifying glioblastoma brain tumors from scans.