Light-Adapted Electroretinogram and Oscillatory Potentials (LEOPs) Dataset for Autism Spectrum Disorder and Typically Developing Individuals

The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides light-adapted electroretinogram (ERG) and oscillatory potentials (OPs) waveforms recorded from children and adolescents diagnosed with Autism Spectrum Disorder (ASD), ASD with co-occurring Attention Deficit Hyperactivity Disorder (ASD + ADHD), and typically developing (TD) controls. Data were collected at two sites: Flinders University, Adelaide, Australia and University College London, United Kingdom, using the RETeval handheld device with skin electrodes.

The dataset comprises 5309 averaged ERG waveforms and 4434 OPs waveforms from 253 participants (157 TD, 75 ASD, 21 ASD + ADHD) across three stimulus protocols:

- 9-step protocol: 4,246 waveforms (173 participants) from a randomised nine-flash-strength series (12–446 Td.s) describing the photopic hill luminance–response function, with 60 averages per flash strength.
- 2-step protocol: 415 waveforms (61 participants) at two flash strengths (113 and 446 Td.s) with an extended recording window (50 ms baseline, 0–170 ms), with 30 averages.
- LA3 protocol: 648 waveforms (217 participants) at the ISCEV standard Light-Adapted 3 cd.s.m⁻² flash (85 Td.s), with 30 averages.

The dataset folder contains three items:

1. LEOPs_dataset.xlsx — Excel spreadsheet with five worksheets. Three waveform sheets (9_step, 2_step, LA3) store per-recording metadata and paired time–amplitude arrays (ms, µV) for each ERG and, where available, OPs waveform. Two participant sheets provide a tabular summary of demographics, stimulus parameters, and time-domain features. Per-waveform metadata includes: diagnostic group, diagnosis confidence, site, age, sex, electrode position, medication status, clinical scores (FSIQ, ASD severity, CARS), iris colour ratio, test date/time, tested eye, flash strength, and ERG features (a-wave and b-wave amplitude and implicit time, summed OPs).

2. jsons/ — 253 structured JSON files (one per participant) designed for machine learning workflows. Each file aggregates all recordings across protocols under a single participant identifier, with participant-level demographics and per-recording waveforms, metadata, and features.

3. ImagesERG/ — 558 PNG eye images (278 right eye, 279 left eye) from 244 participants, showing the RETeval skin electrode position relative to the lower eyelid. Electrode position has been subjectively graded (+1, 0, −1, −2 relative to the manufacturer's recommended height). Images are cross-referenced in both the Excel and JSON files.

Keywords: electroretinogram, ERG, oscillatory potentials, autism spectrum disorder, ASD, ADHD, retina, biomarker, light-adapted, photopic hill, machine learning, RETeval, skin electrode, paediatric

LEOPs（Light-ERG-Oscillatory Potentials，光-ERG-振荡电位）数据集收录了来自确诊为自闭症谱系障碍（Autism Spectrum Disorder, ASD）、合并注意缺陷多动障碍的ASD（ASD + ADHD）以及典型发育（typically developing, TD）对照群体的儿童与青少年的光适应视网膜电图（electroretinogram, ERG）与振荡电位（oscillatory potentials, OPs）波形。本数据集于澳大利亚阿德莱德弗林德斯大学与英国伦敦大学学院两个采集站点，采用RETeval手持设备搭配皮肤电极完成数据采集。

本数据集共包含253名受试者的5309条平均ERG波形与4434条OPs波形，其中典型发育组157人、ASD组75人、ASD+ADHD组21人，涵盖三种刺激范式：
- 9步范式：包含4246条波形（覆盖173名受试者），采用随机化的9级闪光强度序列（12–446 Td.s）以描记明适应希尔亮度-响应函数，每级闪光强度重复采集60次平均波形。
- 2步范式：包含415条波形（覆盖61名受试者），采用两级闪光强度（113与446 Td.s），并配置扩展记录窗口（基线时长50 ms，记录窗口范围0–170 ms），每级闪光强度重复采集30次平均波形。
- LA3范式：包含648条波形（覆盖217名受试者），采用国际临床视觉电生理学会（ISCEV）标准光适应3 cd·s·m⁻²闪光刺激（85 Td.s），每级闪光强度重复采集30次平均波形。

数据集文件夹包含三项内容：
1. LEOPs_dataset.xlsx — 包含5个工作表的Excel表格。其中3个波形工作表（9_step、2_step、LA3）存储了每条记录的元数据，以及每一条ERG（若有配套OPs则一并存储）波形的配对时间-振幅数组（单位：ms, µV）。另外2个受试者工作表提供了人口统计学信息、刺激参数与时域特征的表格汇总。单条波形的元数据包括：诊断分组、诊断置信度、采集站点、年龄、性别、电极位置、用药状态、临床评分（全量表智商Full-Scale Intelligence Quotient, FSIQ、ASD严重程度、儿童孤独症评定量表Childhood Autism Rating Scale, CARS）、虹膜色彩比、检测日期/时间、受试眼、闪光强度，以及ERG特征（a波、b波振幅与潜伏期，以及振荡电位总幅值）。
2. jsons/ — 面向机器学习工作流设计的253个结构化JSON文件（每名受试者对应一个文件）。每个文件以单一受试者标识符整合其所有范式下的记录数据，包含受试者层面的人口统计学信息、单条记录的波形、元数据与特征。
3. ImagesERG/ — 来自244名受试者的558张PNG眼部图像（右眼278张、左眼279张），展示了RETeval皮肤电极相对于下眼睑的位置。电极位置已参照制造商推荐高度进行主观分级（+1、0、-1、-2）。Excel与JSON文件中均已对这些图像完成交叉引用。

关键词：视网膜电图（ERG）、振荡电位（OPs）、自闭症谱系障碍（ASD）、注意缺陷多动障碍（ADHD）、视网膜、生物标志物、光适应、明适应希尔曲线、机器学习、RETeval、皮肤电极、儿科