DEEP LEARNING DETECTION OF RESPIRATORY ANOMALIES
DOI:
https://doi.org/10.24867/16BE41TomicKeywords:
CNN, ICBHI, Convolutional Neural NetworksAbstract
Respiratory diseases cause huge health, economic and social burden and are the third leading cause of death worldwide and a significant burden on the public health system. Therefore, significant research efforts have been made to improve the early diagnosis and routine monitoring of patients with respiratory diseases to enable timely interventions. Respiratory anomalies collected with a stethoscope are classified as discontinuous (cracking) or continuous (wheezing). Such anomalies in the format of recorded sound can also be displayed visually in the form of spectrograms. As convolutional neural networks are dominant in image classification, it is possible to train them to detect respiratory anomalies. The approach presented in this paper based on convolutional neural networks represents a significant improvement over previous approaches and the results obtained among the best on the trained data set.
References
[2] Russakovsky, Olga, et al. "Imagenet large scale visual recognition challenge." International journal of computer vision 115.3 (2015): 211-252.
[3] Piirila P, Sovijarvi AR. Crackles: recording, analysis and clinical significance. Eur Respir J. Eur Respiratory Soc; 1995;8(12):2139–48.
[4] Sarkar M, Madabhavi I, Niranjan N, Dogra M. Auscultation of the respiratory system. Ann Thorac Med. Medknow Publications; 2015;10(3):158.
[5] Sovijarvi ARA. Characteristics of breath sounds and adventitious respiratory sounds. Eur Respir Rev. 2000;10:591–6.
[6] Jakovljević, Nikša, and Tatjana Lončar-Turukalo. "Hidden markov model based respiratory sound classification." International Conference on Biomedical and Health Informatics. Springer, Singapore, 2017.
[7] Kochetov, Kirill, et al. "Wheeze detection using convolutional neural networks." EPIA Conference on Artificial Intelligence. Springer, Cham, 2017.
[8] Demir, Fatih, Abdulkadir Sengur, and Varun Bajaj. "Convolutional neural networks based efficient approach for classification of lung diseases." Health information science and systems 8.1 (2020): 1-8.
[9] https://www.kaggle.com/c/tensorflow-speech-recognition-challenge
[10] https://hyperopt.github.io/hyperopt/
[11] https://bhichallenge.med.auth.gr/
