Pulmonary (Lungs) Sound

Data is collected and ground truth classification are performed by Fortis Hospital, Vasant Kunj, New Delhi, India. The pulmonary sound dataset is arranged in two sets: a normal set (11.28%) representing recordings obtained from healthy subjects and an abnormal set (88.72%) containing recordings from pulmonary patients. The lung sound was recorded with an e-stethoscope connected to a laptop via an amplifier. This setup contains e-stethoscope having a chest piece which is contacted to subject having a microphone to record signal (sampling rate 44,100 Hz) connected to a signal amplifier. The signal amplifier kit can specifically amplifies the sound range (70–2000 Hz) for - lung sounds (having amplifier controller, frequency con- troller) connected to earpiece (for listening live recordings) and laptop. A laptop is used to record the signal where the E-stethoscope is connected using an audio input. The dataset contains audio samples which are in a.wav audio format having a single channel with 16 bit per sample, 44,100 Hz sampling rate and 705kbps Bitrate. Each file is 10 s long.
The data can be used for research purpose only. It is not for commercial purpose.
Anyone using this data for any Publication need to cite the article:
Baghel, N., Nangia, V. & Dutta, M.K. ALSD-Net: Automatic lung sounds diagnosis network from pulmonary signals. Neural Comput & Applic 33, 17103–17118 (2021). https://doi.org/10.1007/s00521-021-06302-1