Segmed, a medical data curation platform, has partnered with NVIDIA and RadImageNet to offer synthetic medical imaging data via their self-serve platform, Segmed Insight. The synthetic data is in addition to their existing 60M+ de-identified real-world imaging records.
Segmed Offers Synthetic Medical Imaging Data to Researchers and Companies
Segmed’s self-serve medical data curation platform, Segmed Insight, now offers synthetic medical imaging data provided by state-of-the-art generative imaging models. Researchers and companies can license the data for medical research purposes, in combination with the existing real-world data accessible through the platform.
The generative models are capable of generating synthetic data for CT, MRIs, Ultrasound, and Endoscopic surgery. These models create over 160 pathologic classifications and synthetic segmentations on top of the synthetic image frames. This data improves the availability of training data, while still protecting patient privacy.
State-of-the-Art Generative AI Models Enhance Medical AI Algorithm Refinement
Segmed, NVIDIA, and RadImageNet’s partnership focuses on advancing medical AI algorithms to provide better accuracy and consistency in medical diagnoses. The synthetic data generated using generative AI models will augment downstream AI model training, helping to expand the scope and variability of patient datasets.
The achieved results will be utilized to classify modality, body part, and reconstruction plane. The synthetic data created will closely mimic real-world data, ultimately leading to better patient outcomes.
Segmed, NVIDIA, and RadImageNet Collaborate to Democratize Healthcare Imaging Data
NVIDIA’s representative is excited about working with Segmed and RadImageNet to take generative AI models for imaging to the next level. This collaboration is a significant step towards enhancing datasets used for research and creating democratized healthcare imaging data for medical diagnosis and treatment.
Rachel Ma, spokesperson for Segmed said, “Supplementing the real-world data Segmed already provides with synthetic data can further increase the robustness and adaptability of our customers’ AI algorithms and models.”
RadImageNet provides a radiologic foundation for radiology artificial intelligence, and their synthetic RadImageNet-RadImageGan {GAN} model pre-training weights have created new database and model options for AI research in radiology.
Segmed’s latest partnership has yielded synthetic medical imaging data for researchers and companies looking to improve medical AI algorithms. With this, the availability of effective data is likely to increase, augmenting the scope and variability of patient datasets, while still ensuring privacy. Ultimately, this collaboration will enhance the accuracy and consistency of medical diagnoses and improve patient outcomes.
