HLS-CMDS: Heart and Lung Sounds Dataset Recorded from a Clinical Manikin using Digital Stethoscope
Description
This dataset contains 210 recordings of heart and lung sounds captured using a digital stethoscope from a clinical manikin, including both individual and mixed recordings of heart and lung sounds; 50 heart sounds, 50 lung sounds, and 110 mixed sounds. It includes recordings from different anatomical chest locations, with normal and abnormal sounds. Each recording has been filtered to highlight specific sound types, making it valuable for artificial intelligence (AI) research and applications in automated cardiopulmonary disease detection, sound classification, and deep learning algorithms related to audio signal processing. If you use this dataset in your research, please cite the following paper: Torabi, Y., Shirani, S., & Reilly, J. P. (2024), Manikin-Recorded Cardiopulmonary Sounds Dataset Using Digital Stethoscope, arXiv preprint, https://doi.org/10.48550/arXiv.2410.03280 Data Type: Audio files (.wav), Comma Separated Values (.CSV) Each category is accompanied by a corresponding CSV file that provides metadata for the respective audio files. The CSV files (HS.csv, LS.csv, and Mix.csv) contain metadata about the corresponding audio files, including the file name, gender, heart and lung sound type, and the anatomical location where we recorded the sound. Sound Types: Normal Heart, Late Diastolic Murmur, Mid Systolic Murmur, Late Systolic Murmur, Atrial Fibrillation, Fourth Heart Sound, Early Systolic Murmur, Third Heart Sound, Tachycardia, Atrioventricular Block, Normal Lung, Wheezing, Crackles, Rhonchi, Pleural Rub, and Gurgling. Auscultation Landmarks: Right Upper Sternal Border, Left Upper Sternal Border, Lower Left Sternal Border, Right Costal Margin, Left Costal Margin, Apex, Right Upper Anterior, Left Upper Anterior, Right Mid Anterior, Left Mid Anterior, Right Lower Anterior, and Left Lower Anterior. Applications: AI-based cardiopulmonary disease detection, unsupervised sound separation techniques, and deep learning for audio signal processing.
Files
Steps to reproduce
Heart and lung sounds were recorded using a 3M™ Littmann® CORE Digital Stethoscope placed on various anatomical locations of a clinical manikin in a sitting position. An instructor tablet was used for controlling the manikin via WiFi. Auscultation sites included the apex, right and left upper sternal borders, and costal margins for heart sounds, and upper, mid, and lower anterior chest regions for lung sounds. The stethoscope used Bell, Diaphragm, and Midrange filter modes to capture heart, lung, and mixed sounds, respectively. Each recording lasted 15 seconds and was transmitted via Bluetooth to the Eko Software, then uploaded to the Eko Cloud and saved in .wav format. The final dataset was categorized into Heart Sounds (HS.zip), Lung Sounds (LS.zip), and Mixed Heart and Lung Sounds (Mix.zip), recorded in a controlled, noise-free environment to ensure high-quality data.