胸腹部疾病诊断结构化文本与影像数据集

该数据为国药东风总医院在开展临床诊疗、放射影像检查等医疗服务过程中产生，包含胸腹部平扫CT原始DICOM格式影像及配套放射科诊断报告，辅以增强CT、增强MRI等其他模态影像数据，以及胃肠镜、病理等佐证报告，覆盖胸部、腹部多类疾病谱系；数据均经合规脱敏处理（含假名化加密、隐私属性删除、年龄/时间信息泛化等），保留影像扫描参数、病灶位置/大小/形态、疾病 ICD编码、肿瘤TNM分期等完整有效元数据，可用于第三方大模型厂商训练医疗影像诊断、疾病智能识别、临床病理分析、肿瘤分期评估等医疗垂类人工智能模型，为医疗AI模型的研发、优化与产业化落地提供高质量的临床标注数据支撑。

This dataset was generated during the provision of clinical diagnosis, treatment and radiological imaging services by Sinopharm Dongfeng General Hospital. It contains raw DICOM-format chest and abdominal plain-scan CT images and accompanying radiological diagnosis reports, supplemented by other modal imaging data such as contrast-enhanced CT and contrast-enhanced MRI, as well as supporting corroborative reports including gastrointestinal endoscopy reports and pathological reports, covering a broad spectrum of diseases affecting the chest and abdomen. All data have undergone compliant de-identification processing, including pseudonymization encryption, deletion of privacy-related attributes, and generalization of age and temporal information, among other measures, while retaining complete and valid metadata including imaging scan parameters, lesion location, size and morphology, ICD codes for diseases, and tumor TNM staging. This dataset can be utilized by third-party large language model (LLM) vendors to train medical vertical AI models for tasks including medical image diagnosis, intelligent disease recognition, clinical pathological analysis and tumor staging assessment, providing high-quality clinically annotated data support for the R&D, optimization and industrial implementation of medical AI models.

搜集汇总

背景与挑战

背景概述

该数据集由国药东风总医院提供，包含胸腹部CT、MRI等多模态影像及配套诊断报告，经脱敏处理后保留完整医学元数据，适用于医疗AI模型的训练与研发。