101 Dalmatians: a multimodal naturalistic fMRI dataset in typical development and congenital sensory loss

We present an fMRI dataset of 50 participants, including typically developed and sensory-deprived (congenitally blind and deaf) individuals, exposed to audiovisual, auditory, or visual versions of the live-action movie 101 Dalmatians. The dataset includes auditory and visual features from computational models (e.g., VGGish, VGG-19), semantic embeddings from GPT-4, and human-tagged annotations of movie events, all formatted in standardized BIDS.

我们发布了一项包含50名参与者的功能磁共振成像（fMRI）数据集，参与者涵盖典型发育个体与感觉剥夺人群——包括先天失明及先天失聪者。所有参与者均接受了真人版电影《101忠狗》的多模态视听、纯听觉或纯视觉版本刺激。该数据集包含来自计算模型的听觉与视觉特征（如VGGish、VGG-19）、GPT-4生成的语义嵌入，以及针对电影事件的人工标注，所有数据均采用标准化脑成像数据结构（Brain Imaging Data Structure，简称BIDS）格式进行组织。