Automatic Classification of Adventitious Respiratory Sounds: A (Un)Solved Problem?

Abstract

(1) Background: Patients with respiratory conditions typically exhibit adventitious respiratory sounds (ARS), such as wheezes and crackles. ARS events have variable duration. In this work we studied the influence of event duration on automatic ARS classification, namely, how the creation of the Other class (negative class) affected the classifiers’ performance. (2) Methods: We conducted a set of experiments where we varied the durations of the other events on three tasks: crackle vs. wheeze vs. other (3 Class); crackle vs. other (2 Class Crackles); and wheeze vs. other (2 Class Wheezes). Four classifiers (linear discriminant analysis, support vector machines, boosted trees, and convolutional neural networks) were evaluated on those tasks using an open access respiratory sound database. (3) Results: While on the 3 Class task with fixed durations, the best classifier achieved an accuracy of 96.9%, the same classifier reached an accuracy of 81.8% on the more realistic 3 Class task with variable durations. (4) Conclusion: These results demonstrate the importance of experimental design on the assessment of the performance of automatic ARS classification algorithms. Furthermore, they also indicate, unlike what is stated in the literature, that the automatic classification of ARS is not a solved problem, as the algorithms’ performance decreases substantially under complex evaluation scenarios.

Benchmarking of eight recurrent neural network variants for breath phase and adventitious sound detection on a self-developed open-access lung sound database—HF_Lung_V1

“…of normal breathing sounds and various types of adventitious sounds [38,[46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63][64][65]. The models in most of these studies are developed on the basis of an open-access ICBHI database [20,21].…”

Pilot study on nocturnal monitoring of crackles in children with pneumonia

“…It has also been shown that the use of spectrograms can improve the classification of wheezes and crackles in an educational setting [ 23 ]. On the other hand, the automatic classification of adventitious respiratory sounds is still an ongoing problem [ 24 ]. Currently there are no public databases available to evaluate and compare new algorithms, but the creation and public release of such databases will be useful to solve the respiratory sound classification problem [ 25 ].…”

Monophonic and Polyphonic Wheezing Classification Based on Constrained Low-Rank Non-Negative Matrix Factorization

“…ss5: Update the estimated wheezing activations matrix A W using Equation (16). ss6: Update the estimated respiratory activations matrix A R using Equation (17). ss7: Repeat Steps 3-6 until the algorithm converges (or until the maximum number of iterations M is reached).…”

“…Continuous Adventitious Sounds (CASs) are characterized by a long duration of more than 100 ms, such as rhonchi, stridor, and wheezing [14]. In recent years, several works have been published that carried out an exhaustive review of lung acoustic measurements [15] and signal processing methods [16,17] applied to adventitious sounds, most of them focused on detection [16][17][18][19][20][21][22][23] and classification tasks [16,17,[24][25][26][27][28][29].…”