PCG Classification Using a Neural Network Approach
| Autor: | Katarzyna Stepien, Jacek Rosinski, Anna Perka, Jan Gieraltowski, Michal Lepek, Iga Grzegorczyk, Mateusz Soliński, Joanna Rymko |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Artificial neural network
business.industry Computer science 0206 medical engineering Process (computing) Pattern recognition 02 engineering and technology Machine learning computer.software_genre 020601 biomedical engineering 03 medical and health sciences 0302 clinical medicine Ordinate Frequency domain Segmentation Time domain Sensitivity (control systems) Artificial intelligence Hidden Markov model business computer 030217 neurology & neurosurgery |
| Zdroj: | CinC |
| ISSN: | 2325-887X |
| DOI: | 10.22489/cinc.2016.323-252 |
| Popis: | Phonocardiography (PCG) is the one of noninvasive ways to diagnose condition of human heart. The mechanics of heart muscle contractions and closure of the heart valves generates vibrations audible as sounds and murmurs, which can be analysed by qualified cardiologists. Developing an accurate algorithm to determine whether patients' heart works properly or should be referred to an expert for further diagnosis would significantly improve the quality of healthcare system. It would allow to perform less unnecessary, expensive and time consuming examinations. The analysed data consisted of PCG recordings from the training set provided by the organizers of the PhysioNet Challenge 2016. Its length variedfrom several to 120 seconds. We propose the machine learning algorithm based on neural networks. The segmentation of the PCG signals is performed with algorithm based on Hidden Markov Model. Whereas, the features necessary to define whether the signal looks normal or should be further analysed were carefully chosen by our team and belonged to time domain, ordinate axis or frequency domain group. The great emphasis was put on the statistical features representing the characteristics of the signal. Their optimal values were found during the process of learning of our algorithm. The best overall score we achieved in the official phase of the PhysioNet Challenge 2016 is 0.79 with specificity 0.76 and sensitivity 0.81. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|