MIT's new AI dramatically speeds up medical image annotation
Researchers from the Massachusetts Institute of Technology (MIT), Massachusetts General Hospital (MGH) and Harvard Medical School have developed a new artificial intelligence. It has been named ScribblePrompt.
The researchers claim that the new AI reduces annotation time by 28%. The purpose of the development is to help medical professionals analyze images. ScribblePrompt is an interactive environment that highlights anatomical structures in various medical images such as MRIs, X-rays. During the AI training, the research team modeled the process of annotating more than 50,000 scans, including ultrasounds and photos of cells and tissues.
The ScribblePrompt interface allows users to simply swipe or click on the area they want to segment. The tool will highlight the entire structure or background as requested. ScribblePrompt also provides the ability to make corrections based on user feedback and to finalize areas, making it a preferred tool among neuroimaging researchers at MGH.
Using algorithms, they were able to mimic user actions, allowing ScribblePrompt to efficiently process new types of images without the need for additional data collection. According to the authors of the project, with the launch of AI, medical professionals will have a powerful tool that can significantly improve the diagnosis and treatment of patients.