分享
编辑者： Svitlana Velhush
Use Case # 998: @cognita CXR, a GenAI-powered radiology model developed by Cognita and announced by Mosaic Clinical Technologies, has received Breakthrough Device Designation from the U.S. Food and Drug Administration (@US_FDA) genaigazette.com/cognita-genai-…
FDA授予Cognita CXR生成式AI模型突破性设备地位。这是放射学领域首个获得此称号的生成式模型。
该模型由Cognita公司开发，Cognita是Mosaic Clinical Technologies的人工智能部门，隶属于Radiology Partners。Cognita CXR是一种视觉语言模型，可分析胸部X光片并生成初步报告，供放射科医生随后审核和定稿。
根据初步的内部临床数据，使用该模型可将影像解读效率提高18%。同时，对各种发现的检出率也有改善，提升幅度在16%至65%之间。
“我们认为这是将AI融入放射学日常实践的重要一步，使医生能够发挥最大潜力，并扩大全球范围内获得高质量影像检查的机会，”Mosaic Clinical Technologies的首席医疗AI官、医学博士Nina Kottler表示。
突破性设备地位加速了与FDA的互动，旨在解决诊断吞吐量问题。这将如何影响诊所中放射科医生的实际工作量？
该模型集成到现有工作流程中：放射科医生获得全面的初步评估，并可专注于最终解读。公司强调，这是首批获得监管机构认可的此类工具之一。
Generative Vision Language Model for Chest X-Ray Gets FDA’s Breakthrough Device Designation | Diagnostic Imaging
JUST IN: UN votes 164-1 to adopt Equal Earth map over Mercator, revealing Africa's true size. US sole opponent. This visual comparison highlights why the vote matters—Africa appears far larger and more accurate on the new projection.
Circuit boards made from living bacteria. Strains are engineered to act as N- or P-type transistors. Other strains pass around the signals, which are molecules instead of electrons. These strains are then printed on a plate, using a liquid handler, to build any logic you want.
