UniBiomed
收藏资源简介:
UniBiomed是一个通用的基础模型,用于基于图像的医学图像解释。该模型基于多模态大型语言模型(MLLM)和分割任意模型(SAM)的创新整合,能够有效地统一临床文本生成和对相应生物医学对象的分割,从而实现基于图像的解释。为了开发UniBiomed,研究人员构建了一个大规模数据集,包含超过2700万个图像、注释和文本描述的三元组,涵盖10种成像模式。该数据集是目前为止最大的、最全面的生物医学图像解释数据集。
UniBiomed is a general-purpose foundational model dedicated to image-based medical interpretation. Based on the innovative integration of multimodal large language models (MLLMs) and the Segment Anything Model (SAM), this model can effectively unify clinical text generation and segmentation of corresponding biomedical objects, thereby achieving image-based medical interpretation. To develop UniBiomed, researchers constructed a large-scale dataset containing over 27 million triplets of images, annotations and textual descriptions, covering 10 imaging modalities. This dataset is the largest and most comprehensive biomedical image interpretation dataset to date.




