STimage-1K4M

STimage-1K4M是由北卡罗来纳大学教堂山分校创建的一个新颖数据集，旨在通过提供子图像的基因组特征来弥补现有医学图像-文本数据集中文本描述不足的问题。该数据集包含1,149张来自空间转录组学数据的空间转录图像，每张图像被分解成更小的子图像块，每个子图像块配有15,000至30,000维的基因表达数据。STimage-1K4M包含4,293,195对子图像和基因表达数据，提供了前所未有的粒度，为多模态数据分析和计算病理学中的创新应用铺平了道路。该数据集主要应用于研究组织的空间组织和理解细胞结构、基因活性与疾病和健康相关结果之间的复杂关系，旨在解决医学图像分析中详细描述不足的问题。

STimage-1K4M is a novel dataset developed by the University of North Carolina at Chapel Hill, designed to address the insufficient textual description problem in existing medical image-text datasets by providing genomic features of sub-images. This dataset contains 1,149 spatial transcriptomic images sourced from spatial transcriptomics data, each split into smaller sub-image patches, with each patch paired with gene expression data ranging from 15,000 to 30,000 dimensions. STimage-1K4M includes 4,293,195 pairs of sub-image patches and their corresponding gene expression data, offering unprecedented granularity and paving the way for innovative applications in multimodal data analysis and computational pathology. This dataset is primarily utilized to study the spatial organization of tissues and decipher the complex correlations between cellular structures, gene activity, and disease or health-related outcomes, aiming to resolve the gap of insufficient detailed descriptions in medical image analysis.