# Noise Holter Database - a multi-device, multi-arrhythmia real-world clinical ECG signal quality benchmark Latest version: 1.0.0. The **Noise Holter Database** is a publicly available, multi-arrhythmia Holter ECG dataset comprising 258 two-lead recordings from distinct subjects. Encompassing approximately 35 hours of data, the database features 6,255 ECG segments with expert-assigned multi-label signal quality annotations. The primary objective of making this resource available is to facilitate the development of software for filtering out noisy segments in ambulatory recordings. The dataset includes a wide spectrum of spontaneous cardiac arrhythmias—including atrial fibrillation, ventricular tachycardia, AV block, and supraventricular tachycardia—captured under real-world ambulatory conditions with authentic environmental noise. The signal quality annotations categorize the segments into three functional levels: * **Clean**: Represents an optimal signal where atrial and ventricular electrical activities (such as normal waveforms or sustained arrhythmia patterns) are clearly readable without impact on the signal interpretation (43.7% of the dataset). * **N1**: Represents an intermediate noise level where the signal is corrupted by artifacts, but the QRS complexes can still be reliably identified. This class is clinically relevant as it represents the threshold at which rhythm analysis remains feasible (36.0% of the dataset). * **N2**: Designates severe noise where atrial and ventricular electrical activities are completely obscured and no longer clearly recognizable. Segments in this class are unsuitable for automated or manual ECG editing (20.3% of the dataset). --- ## Content All ECG waveform data and expert annotations are provided as extracted segments named using a standardized, sequential identifier combining the patient number and the segment index (e.g., `001-001`, `001-002`). Each clinical segment comprises four interconnected open-source files organized: * **ECG signal files (`.dat`)**: 16-bit binary files containing the raw ECG waveform. A segment contains two concurrent leads if both channels shared the exact same signal quality and clinical annotation, or a single lead if the specific annotated event applied to only one channel. * **Header files (`.hea`)**: Standard ASCII header files detailing critical metadata required for signal processing, including the sampling frequency (200 Hz or 256 Hz depending on the device), number of leads, ADC resolution, recording duration, and the specific Holter device type. Global characteristics of each segment are embedded directly within the comment section using a strict string convention: `CLASS: | NOISE: | EVENTS:`. * **Clinical event annotation files (`.atr`)**: Binary annotation files providing the clinical ground truth, where temporal locations of individual beats (expressed as sample indices) are associated with standard WFDB annotation symbols ('N' for normal sinus beats, 'V' for premature ventricular contractions, 'S' for supraventricular premature beats, etc.). The specific temporal context is preserved within the WFDB `aux_note` field. * **QRS detection files (`.qrs`)**: Supplementary binary files containing exclusively the temporal indices of detected QRS complexes.