MedImg is a medical imaging database intended to provide the data training the medical artificial intelligence models.
MedImg is a medical image database designed to provide the necessary training and validation data required for the development of medical artificial intelligence (AI) models. As deep learning continues to advance, AI models are becoming increasingly complex and demand larger volumes of training data. In the medical field, datasets are generally limited in size. MedImg addresses this challenge by integrating numerous medical image datasets comprising 103 datasets. These data cover multiple imaging modalities including XR, CT, MRI, ultrasound, endoscopy, OCT, histopathology slide, ECG, EEG, and device photo, multiple body parts including lung, brain, heart, skin, knee, eye, kidney, liver, and prostate, and multiple data types. All these data are stored in a structured way in the MedImg database. In summary, MedImg serves as a valuable resource for training AI models and can provide a benchmark for evaluating algorithmic performance.
Department of Biomedical Informatics,
School of Basic Medical Sciences,
Peking University, Beijing, China
Download the full database without ACROBAT2022 (3 parts): (Medage)
Download the ACROBAT2022 (15 parts): ACROBAT2022