haifan-gong/LFC
收藏资源简介:
LFC数据集是一个精心策划的3D胎儿脑MRI数据集,专门设计用于胎儿小脑地标检测和胎儿小脑生物测量。它支持胎儿脑发育、胎儿小脑形态学、医学地标检测和3D医学图像分析的研究。该数据集包含300个去标识化的胎儿MRI病例,每个病例提供重建的3D胎儿脑MRI体积和专家评审的胎儿小脑地标注释。每个MRI体积包含六个胎儿小脑地标,对应三个胎儿小脑生物测量的端点:横向小脑直径(TCD)、蚓部高度直径(HDV)和蚓部前后直径(ADV)。数据集的孕周范围为24至38周,每个孕周包含20个病例,并提供了官方的训练/验证/测试划分(120个训练病例、60个验证病例、120个测试病例)。MRI数据为T2加权序列,通过多视角厚层扫描重建为各向同性的3D体积,空间分辨率为0.6×0.6×0.6 mm³。地标注释由放射科医生标注并经过另一位经验丰富的放射科医生检查修订达成共识。该数据集旨在用于学术和非商业研究,包括胎儿脑MRI分析、医学图像地标检测、领域泛化等。
The LFC Dataset is a curated 3D fetal brain MRI dataset designed for fetal cerebellum landmark detection and fetal cerebellum biometric measurement. It supports research in fetal brain development, fetal cerebellum morphology, medical landmark detection, and 3D medical image analysis. The dataset includes 300 de-identified fetal MRI cases, each providing reconstructed 3D fetal brain MRI volumes with expert-reviewed fetal cerebellum landmark annotations. Each MRI volume contains six fetal cerebellum landmarks, corresponding to the two endpoints of three fetal cerebellum biometric measurements: Transverse Cerebellar Diameter (TCD), Height Diameter of the Vermis (HDV), and Anteroposterior Diameter of the Vermis (ADV). The gestational age range of the dataset is 24 to 38 weeks, with 20 cases per gestational week, and it provides an official train/validation/test split (120 training cases, 60 validation cases, 120 testing cases). The MRI data are T2-weighted sequences, reconstructed from multi-view thick-slice scans into isotropic 3D volumes with a spatial resolution of 0.6×0.6×0.6 mm³. Landmark annotations were first annotated by a radiologist and then checked and revised by another experienced radiologist to reach consensus. The dataset is intended for academic and non-commercial research, including fetal brain MRI analysis, medical image landmark detection, domain generalization, and more.




