ULS23
收藏资源简介:
ULS23数据集是由荷兰和德国的多个医学研究机构联合创建,专注于胸部-腹部-骨盆CT检查中的3D通用病变分割。该数据集包含38,693个病变,覆盖了胰腺、结肠和骨骼等难以分割的病变类型。数据集的创建过程涉及从临床报告中提取信息,并通过专业人员进行精确的3D分割。ULS23数据集的应用领域主要在于提高癌症治疗评估中对肿瘤表现的测量效率,通过自动化分割加速放射学工作流程,并支持更复杂的分析如放射组学特征分析,以区分不同类型的病变。
The ULS23 dataset was jointly developed by multiple medical research institutions across the Netherlands and Germany, focusing on 3D universal lesion segmentation in chest-abdominal-pelvic CT scans. This dataset contains 38,693 lesions, covering challenging lesion types such as those arising from the pancreas, colon, and skeletal tissues. The development process of the dataset involved extracting information from clinical reports and performing precise 3D segmentation by medical professionals. The primary application areas of the ULS23 dataset include improving the measurement efficiency of tumor performance during cancer treatment assessment, accelerating radiological workflows through automated lesion segmentation, and supporting more complex analyses such as radiomic feature analysis to differentiate between distinct lesion types.



