Searched over 200M research papers
10 papers analyzed
These studies suggest that wheezing and crackling sounds when breathing out are caused by the opening of airways in deflated lung regions and can be effectively detected and classified using advanced technologies like digital stethoscopes, AI, and deep learning methods, aiding in the diagnosis and treatment of respiratory diseases.
20 papers analyzed
Wheezing and crackling are common adventitious lung sounds that can indicate various respiratory conditions. Wheezing is characterized by a high-pitched, whistling sound during expiration, often associated with airway obstruction. Crackling, on the other hand, is typically due to the explosive opening of airways in lung regions that have been deflated to residual volume.
Wheezing occurs when air passes through narrowed airways at high velocity, producing a high-pitched sound. This sound is largely independent of the airway's caliber and length but is determined by the linear velocity of the air jet. Clinically, wheezing is often associated with conditions such as asthma and chronic obstructive pulmonary disease (COPD). It is important to note that wheezing is not synonymous with bronchospasm and can result from other causes of airway obstruction.
Crackling sounds are brief, discontinuous, and typically heard during inspiration. They are often associated with conditions such as pneumonia and bronchitis. Crackles are generated by the sudden opening of small airways and alveoli that were previously collapsed. In pediatric populations, digital stethoscopes have been shown to be more sensitive than traditional auscultation in detecting crackles.
Recent advancements in deep learning and artificial intelligence (AI) have significantly improved the accuracy of detecting and classifying respiratory sounds. For instance, convolutional neural networks (CNNs) have been utilized to classify respiratory sounds with high accuracy, achieving an overall accuracy of 85.7% and a mean area under the ROC curve (AUC) of 0.92. Similarly, optimized S-transform and deep residual networks (ResNets) have shown excellent performance in multi-classification of respiratory sounds, with accuracy, sensitivity, and specificity up to 98.79%, 96.27%, and 100%, respectively.
Digital stethoscopes, combined with AI algorithms, have demonstrated high accuracy in detecting pathological breath sounds. For example, AI algorithms trained to detect wheezes and crackles achieved positive percent agreement (PPA) and negative percent agreement (NPA) of up to 0.95 and 0.99, respectively, for crackle detection, and 0.90 and 0.97, respectively, for wheeze detection. These technologies offer a reliable alternative to traditional auscultation, which often suffers from poor inter-observer reliability.
A study involving 4033 individuals aged 40 years or older found that 28% had wheezes or crackles. The prevalence of wheezes was higher in women (18.6%) compared to men (15.3%), while crackles were also more common in women (10.8%) than in men (9.4%). Significant predictors of wheezing included age, female gender, self-reported asthma, and current smoking. For crackles, predictors included age, current smoking, and decreased lung function.
Wheezing and crackling sounds during breathing out are important clinical indicators of various respiratory conditions. Advances in digital stethoscopes and AI have enhanced the accuracy and reliability of detecting these sounds, providing valuable tools for clinicians. Understanding the prevalence and predictors of these sounds in the general population can aid in early diagnosis and management of respiratory diseases.
Most relevant research papers on this topic