Auscultation is a standard physical examination method used by physicians and is widely accepted by doctors and patients because of its simplicity, repeatability, and non-invasiveness. Auscultation can be classified as direct or indirect. Direct auscultation attaches the ear directly to the body wall of the examined person, while indirect auscultation uses a stethoscope. Because the sound heard by direct auscultation is weak and inconvenient to operate, indirect auscultation has gradually replaced direct auscultation since the invention of acoustic stethoscope (Sarkar et al., 2015).

With its low price and ease of use, the acoustic stethoscope is still the most common auscultation tool despite 200 years of evolution and development (Vyshedskiy et al., 2009; Bank et al., 2016). However, several limitations of the acoustic stethoscope are evident, such as interference from environmental noise, low volume, and the lack of storage and playback functions. Given these defects, the electronic stethoscope emerged (Tavel, 2006; Bank et al., 2016; Rennoll et al., 2021). In 1956, Airsonic Limited (London) produced a new electronic stethoscope that included a miniature battery-operated amplifier, a contact microphone, and conventional ear-pieces. It overcame the poor sound quality of previous electronic stethoscopes and had a good environmental noise reduction effect. The stethoscope only produced noise when its diaphragm was in contact with the body (Taylor and Fothergill, 1956). So far, there is no report that the modern electronic stethoscope can filter this noise, even though it has undergone some evolution and improvement.

Nowadays, the application of electronic stethoscope in cardiopulmonary sound auscultation is increasing gradually. Further, their audio recording characteristics are used for intelligent recognition of cardiopulmonary sounds. However, noise similar to a fine crackle, referred to as a fake crackle, is generated during use, which may interfere with the doctors’ clinical diagnosis of patients. A crackle is a discontinuous, explosive, non-musical adventitious lung sound (Murphy, 1981; Sarkar et al., 2015), with a frequency between 200 and 1200 Hz. It is often detected in the respiratory phase because it is generated by the sudden opening of the airway during inspiration (Forgacs et al., 1971; Forgacs, 1978; Vyshedskiy et al., 2009). Crackles are commonly observed in chronic obstructive pulmonary disease, bronchiectasis, pneumonia, pulmonary edema, and other respiratory diseases, as well as heart failures (Bettencourt et al., 1994; Piirilä and Sovijärvi, 1995; Tilkian and Conover, 2009; Bohadana et al., 2014). Crackles can be roughly divided into coarse crackles and fine crackles. The coarse crackles have a loud intensity and low pitch, with a duration of about 15 m and a typical frequency of about 350 Hz. The fine crackles have a quieter intensity and a higher pitch, with a duration of about 5 m and a typical frequency of about 650 Hz (Bohadana et al., 2014; Sarkar et al., 2015).

At the end of the 20th century, some scholars explored the automatic detection of lung sounds (Kaisla et al., 1991; Sovijärvi et al., 1998; Vannuccini et al., 1998). Due to the separation of the recording process from the auscultation process, the collection efficiency of the lung sounds is low, and the sample size is too small for research. Hence, the resulting lung sound recognition model is not widely used.

The electronic stethoscope can simultaneously collect lung sounds and stethoscope, which simplifies the collection process of lung sounds. Therefore, some studies on intelligent lung sound recognition have used an electronic stethoscope, such as the 3M Littmann 3200, to collect data (Grzywalski et al., 2019; Kevat et al., 2020; Hsu et al., 2021; Nguyen and Pernkopf, 2022). However, the recorded lung sounds still need to be heard by annotators and labeled manually before they can be used in the training algorithm. The fake crackles may be labeled as crackles, affecting the crackle feature extraction and the recognition rate of the pulmonary sound classification model. This study aims to reveal the causes, characteristics, and occurrence rules of fake crackles and establish a reference for improving the reliability of the electronic stethoscope during pulmonary auscultation.

Of the 56 participants in the study (29 males and 27 females), 34 heard a fake crackle when they were auscultated with an electronic stethoscope. We collected 504 audio recordings from 56 participants, of which 4 were excluded because of operating errors that caused the duration to be less than 5 s or respiratory sounds not being recorded. The remaining 500 were included in the study.

Fake crackles were found in 61 of 500 recordings, of which 56 (91.80% of the total number of fake crackle audio recordings, posterior same) were heard in the inspiratory phase, 2 (3.28%) in the expiratory phase, and 3 (4.92%) in the respiratory biphasic phase. There were 261 audio recordings for men, 33 of which had fake crackles, and 239 audio recordings for women, 28 of which had fake crackles. There was no significant difference in fake crackles generated using an electronic stethoscope in males and females (p = 0.751).

The minimum and maximum frequencies of fake crackles in each audio recording are presented in Supplementary Table S1. After spectrum analysis of each audio recording with a fake crackle, we found that the minimum frequency of the fake crackle was between 250 and 1150 Hz, the maximum frequency band was 1050–1950 Hz, the middle frequency band was 700–1375 Hz, and the mean middle frequency was 1052.05 ± 99.40 Hz (1 standard deviation). The frequencies of fake crackles at different locations are depicted in Figure 3. Although they differed slightly, there was a common frequency band for all locations of 500–1600 Hz.

The numbers of fake crackle audio recordings in each auscultation location are presented in Table 1 and account for the total number of fake crackle audio recordings, as depicted in Figure 4. The left lower lung (L3) and right lower lung (R3) have the most fake crackles, accounting for 19.67% and 21.31% of the total number of fake crackle audio recordings. There was no significant difference in fake crackles between the nine auscultation locations (p = 0.061). We then combined L1 and R1 into the upper lung, middle lung (R2 and L2), lower lung (R3 and L3), and infrascapular region (R4 and L4). The overall difference between fake crackles at the five locations was significant (p < 0.001). A pairwise comparison of each location revealed significant differences between lower lung and larynx (p < 0.001), lower lung and upper lung (p = 0.005), lower lung and middle lung (p = 0.005), and lower lung and infrascapular region (p = 0.027). In contrast, there was no significant difference in the pairwise comparison of other locations. The results of the pairwise comparison of each location are depicted in Figure 5.

Our study proposed (to the best of our knowledge) for the first time that fake crackles may be generated by an electronic stethoscope during auscultation. In this study, we researched the cases and occurrence rules of fake crackles and analyzed their frequency spectrum. Our findings suggest that fake crackles are most likely to be heard when using an electronic stethoscope to auscultate the bilateral lower lungs and that the frequency of a fake crackle overlaps with a crackle, reducing the reliability of electronic stethoscopes during pulmonary auscultation.