RadImageGAN: Multi-Modal Generative AI for Medical Imaging
, Sr. Machine Learning Engineer, NVIDIA
, Chief Medical Information Officer and Head of Musculoskeletal Interventional Services, East River Medical Imaging
, Postdoctoral Fellow, Mount Sinai
, PhD Student, Mount Sinai
高度评价
We introduce RadImageGAN, a new multi-modal (CT/MR/ultrasound) generative AI for radiology capable of generating 165 distinct classes with various pathologies over 14 anatomical regions. This work uses StyleGAN-XL, and was trained on 1.3 million images from the RadImageNet dataset. This was a collaboration between NVIDIA, Mt. Sinai, and East River Imaging.