Brain US 大脑解剖分割图像数据集

Brain US 全称 Brain Ultrasound，是关于大脑解剖学分割的图像数据集，共包含来自 20 位受试者的 1,629 张 vivo B-mode US 图像，其中 1,300 张用于训练，329 张用于测试。每张图像的 ventricles 和 septum pellecudi 都由超声波专家手动分割，图像大小均为 512 X 512 。这些受试者在 2010 年至 2016 年间接受治疗，受试者的年龄差均小于一岁，一部分受试者患有 IVH，另一部分受试者存在患有 IVH 的风险。 US 扫描是使用飞利浦 US 机器收集的，该机器带有 C8-5 宽带弧形阵列换能器，使用冠状和矢状扫描平面。

Brain US, whose full name is Brain Ultrasound, is an image dataset for brain anatomical segmentation. It contains a total of 1,629 in vivo B-mode ultrasound images from 20 subjects, with 1,300 images allocated for training and 329 images for testing. The ventricles and septum pellucidum of each image have been manually segmented by certified ultrasound experts, and all images have a fixed size of 512 × 512. These subjects received clinical treatment between 2010 and 2016, with an age difference of less than one year among all participants. Some of the subjects suffered from IVH, while others were at risk of developing IVH. The ultrasound scans were acquired using a Philips ultrasound machine equipped with a C8-5 broadband curved array transducer, utilizing coronal and sagittal scan planes.

搜集汇总

背景与挑战

背景概述

Brain US 大脑解剖分割图像数据集是一个专注于大脑解剖学分割的医学图像数据集，包含1,629张来自20位受试者的vivo B-mode US图像，所有图像均由专家手动标注了ventricles和septum pellecudi结构，图像尺寸统一为512x512。该数据集主要用于训练和测试图像分割模型，特别针对婴儿脑部超声图像分析，受试者年龄相近且部分涉及IVH疾病或风险，数据采集使用飞利浦US机器在特定扫描平面下完成，适用于医学影像研究和人工智能应用开发。