Detection of Parkinson's Disease Based on Deep Learning and Feature Selection

Abstract

Parkinson's disease is a serious long-term neurodegenerative disease affecting the motor system. It progresses slowly and causes the degeneration of its cells over time. This disease is difficult to diagnose and is one of the common diseases in society. Due to the deficiency of dopamine cells in the brain, it causes motor and non-motor (speech, smell) defects in the body. It is known that most patients with Parkinson's have voice disorders. Speech signals in Parkinson's patients differ greatly compared to normal people. In this study, a method based on deep learning by using the acoustic features of speech signals is proposed for the classification of Parkinson's disease. In the first step, acoustic features are passed through genetic algorithm and effective features are selected. In addition, ReliefF feature selection algorithm is used to compare the performance of genetic algorithm. These features are given as input to the Convolutional Neural Network (CNN) architecture designed in the second step. Experiments are done with a dataset widely used in the literature. This dataset consists of two classes and is taken from the UCI Machine Learning repository. The average accuracy was 89.67% without feature selection, and 94.23% on average with feature selection.

Keywords

• Parkinson
• Deep learning
• Genetic algorithm

Özellik seçimi ve Derin Öğrenmeye Dayalı Parkinson Hastalığı Tespiti

Öz

Parkinson hastalığı, motor sistemini etkileyen uzun süreli ciddi bir nörodejeneratif hastalıktır. Yavaşça ilerler ve zamanla hücrelerinin dejenerasyonuna neden olur. Bu hastalığın teşhisi zordur ve toplumdaki yaygın hastalıklardan biridir. Beyindeki dopamin hücrelerinin yetersizliği nedeniyle, vücutta motor ve motor olmayan (konuşma, koku alma) kusurlara yol açar. Parkinson hastalarının çoğunda ses bozukluğu olduğu bilenmektedir. Parkinson hastalarındaki konuşma sinyalleri, normal insanlara kıyasla büyük farklılıklar göstermektedir. Bu araştırmada, Parkinson hastalığının sınıflandırılması için konuşma sinyallerinin akustik özellikleri kullanılarak derin öğrenmeye dayalı yöntem önerilmektedir. İlk adımda akustik özellikler genetik algoritmadan geçirilerek etkin özellikler seçilmiştir. İkinci adımında tasarlanan Konvolüsyonel Sinir Ağı (KSA) mimarisine bu özellikler girdi olarak verilmiştir. Deneyler literatürde yaygın kullanılan veri seti ile yapılmıştır. Bu veri seti iki sınıftan oluşmaktadır ve UCI Makine Öğrenimi deposundan alınmıştır. Özellik seçimi olmadan ortalama %87,66, özellik seçim ile ise ortalama olarak %88,98 doğruluk elde edilmiştir.

Anahtar Kelimeler

• Parkinson
• Derin öğrenme
• Genetik algoritma

References

1. W. Poewe et al., “Parkinson disease,” Nat. Rev. Dis. Prim., vol. 3, no. 1, 2017, doi: 10.1038/nrdp.2017.13.
2. A. Benba, A. Jilbab, and A. Hammouch, “Detecting Patients with Parkinson’s disease using Mel Frequency Cepstral Coefficients and Support Vector Machines,” Int. J. Electr. Eng. Informatics, vol. Volume 7, pp. 297–307, Jul. 2015, doi: 10.15676/ijeei.2015.7.2.10.
3. A. Reeve, E. Simcox, and D. Turnbull, “Ageing and Parkinson’s disease: why is advancing age the biggest risk factor?,” Ageing Res. Rev., vol. 14, no. 100, pp. 19–30, Mar. 2014, doi: 10.1016/j.arr.2014.01.004.
4. J. E. Arena and A. J. Stoessl, “Optimizing diagnosis in Parkinson’s disease: Radionuclide imaging,” Parkinsonism Relat. Disord., vol. 22, pp. S47–S51, 2016, doi: 10.1016/j.parkreldis.2015.09.029.
5. L. F. Parra-Gallego, T. Arias-Vergara, J. C. Vásquez-Correa, N. Garcia-Ospina, J. R. Orozco-Arroyave, and E. Nöth, “Automatic Intelligibility Assessment of Parkinson’s Disease with Diadochokinetic Exercises,” Communications in Computer and Information Science. Springer International Publishing, pp. 223–230, 2018, doi: 10.1007/978-3-030-00353-1_20.
6. N. Hosseini-Kivanani, J. C. Vásquez-Correa, M. Stede, and E. Nöth, “Automated Cross-language Intelligibility Analysis of Parkinson’s Disease Patients Using Speech Recognition Technologies,” Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop. Association for Computational Linguistics, 2019, doi: 10.18653/v1/p19-2010.
7. B. E. Sakar et al., “Collection and Analysis of a Parkinson Speech Dataset With Multiple Types of Sound Recordings,” IEEE J. Biomed. Heal. Informatics, vol. 17, no. 4, pp. 828–834, 2013, doi: 10.1109/jbhi.2013.2245674.
8. C. D. Rios-Urrego, J. C. Vásquez-Correa, J. F. Vargas-Bonilla, E. Nöth, F. Lopera, and J. R. Orozco-Arroyave, “Analysis and evaluation of handwriting in patients with Parkinson’s disease using kinematic, geometrical, and non-linear features,” Comput. Methods Programs Biomed., vol. 173, pp. 43–52, 2019, doi: 10.1016/j.cmpb.2019.03.005.

9. N. H. Trinh and D. O’Brien, “Pathological Speech Classification Using a Convolutional Neural Network,” in in Proc. IMVIP, Ireland, 2019.
10. M. Little, P. McSharry, E. Hunter, J. Spielman, and L. Ramig, “Suitability of dysphonia measurements for telemonitoring of Parkinson’s disease,” Nat. Preced., 2008, doi: 10.1038/npre.2008.2298.1.
11. I. Bhattacharya and M. P. S. Bhatia, “SVM classification to distinguish Parkinson disease patients,” Proceedings of the 1st Amrita ACM-W Celebration on Women in Computing in India - A2CWiC ’10. ACM Press, 2010, doi: 10.1145/1858378.1858392.
12. C. O. Sakar et al., “A comparative analysis of speech signal processing algorithms for Parkinson’s disease classification and the use of the tunable Q-factor wavelet transform,” Appl. Soft Comput., vol. 74, pp. 255–263, 2019, doi: 10.1016/j.asoc.2018.10.022.
13. H. Gunduz, “Deep Learning-Based Parkinson’s Disease Classification Using Vocal Feature Sets,” IEEE Access, vol. 7, pp. 115540–115551, 2019, doi: 10.1109/access.2019.2936564.
14. L. Parisi, N. RaviChandran, and M. L. Manaog, “Feature-driven machine learning to improve early diagnosis of Parkinson’s disease,” Expert Syst. Appl., vol. 110, pp. 182–190, 2018, doi: 10.1016/j.eswa.2018.06.003.
15. L. Ali, C. Zhu, M. Zhou, and Y. Liu, “Early diagnosis of Parkinson’s disease from multiple voice recordings by simultaneous sample and feature selection,” Expert Syst. Appl., vol. 137, pp. 22–28, 2019, doi: 10.1016/j.eswa.2019.06.052.
16. L. Chen, C. Wang, J. Chen, Z. Xiang, and X. Hu, “Voice Disorder Identification by using Hilbert-Huang Transform (HHT) and K Nearest Neighbor (KNN),” J. Voice, 2020, doi: 10.1016/j.jvoice.2020.03.009.
17. S. Sivaranjini and C. M. Sujatha, “Deep learning based diagnosis of Parkinson’s disease using convolutional neural network,” Multimed. Tools Appl., vol. 79, no. 21–22, pp. 15467–15479, 2019, doi: 10.1007/s11042-019-7469-8.
18. Y. Fu and C. Aldrich, “Flotation froth image recognition with convolutional neural networks,” Miner. Eng., vol. 132, pp. 183–190, 2019, doi: 10.1016/j.mineng.2018.12.011.
19. A. GÜLCÜ and Z. KUŞ, “Konvolüsyonel Sinir Ağlarında Hiper-Parametre Optimizasyonu Yöntemlerinin İncelenmesi,” Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji, vol. 7. Gazi Üniversitesi, pp. 503–522, 2019, doi: 10.29109/gujsc.514483.
20. D. C. Cireundefinedan, U. Meier, J. Masci, L. M. Gambardella, and J. Schmidhuber, “Flexible, High Performance Convolutional Neural Networks for Image Classification,” in Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence - Volume Volume Two, 2011, pp. 1237–1242.
21. T.-C. Wu, K.-C. Hung, J.-H. Liu, and T.-K. Liu, “Wavelet-based ECG data compression optimization with genetic algorithm,” J. Biomed. Sci. Eng., vol. 06, no. 07, pp. 746–753, 2013, doi: 10.4236/jbise.2013.67092.
22. J. R. Koza, “Genetic programming as a means for programming computers by natural selection,” Stat. Comput., vol. 4, no. 2, pp. 87–112, 1994, doi: 10.1007/BF00175355.
23. Y. Xiong and Y. Lu, “Deep Feature Extraction From the Vocal Vectors Using Sparse Autoencoders for Parkinson’s Classification,” IEEE Access, vol. 8, pp. 27821–27830, 2020, doi: 10.1109/access.2020.2968177.