ADHD Dataset

该数据集包含了776个主题的记录，这些主题被标记为注意力缺陷多动障碍（ADHD）的真实患者和正常对照组。原始数据集不平衡，为了性能评估，随机抽取了100名ADHD患者和100名正常对照组。所有扫描均已裁剪并调整至96×96×96的尺寸；使用MNI 152与AAL图谱作为注册和分割的模板。规模上，共有200次扫描（160次用于训练，20次用于验证，以及20次用于测试）。任务包括脑成像分析，涵盖提取、注册、分割、分区和分类。This dataset contains records for 776 subjects, labeled as real patients with attention-deficit/hyperactivity disorder (ADHD) and normal control participants. The original dataset was imbalanced; thus, 100 ADHD patients and 100 normal control subjects were randomly sampled for performance evaluation. All scans were cropped and resized to a dimension of 96×96×96, with the MNI 152 template and AAL atlas serving as templates for registration and segmentation. The finalized dataset includes 200 scans in total: 160 allocated for training, 20 for validation, and 20 for testing. The associated tasks cover brain imaging analysis, including feature extraction, registration, segmentation, parcellation, and classification.

搜集汇总

背景与挑战

背景概述

该数据集是UniBrain模型的一部分，主要用于脑成像分析，包含ADHD 200和ABIDE等公共数据集，支持脑提取、配准、分割、分区、网络生成和分类等多种任务。