Yüz Görüntüleri ile Parkinson Teşhisi için Çoğunluk Oylamalı Topluluk Derin Öğrenmesi

Abstract

Parkinson hastalığı (PD), dopamin üreten beyin hücrelerinin kaybı veya hasar görmesi nedeniyle ortaya çıkan ilerleyici bir nörodejeneratif bozukluktur. Erken teşhis büyük önem taşımaktadır, çünkü zamanında uygulanan tedavi, hastaların yaşam kalitesini artırabilir ve hastalığın ilerlemesini yavaşlatabilir. PD teşhisi için beyin görüntüleme, nörolojik testler, el yazısı ve ses analizi, yüz görüntüsü değerlendirmesi ve fiziksel muayene gibi çeşitli yöntemler kullanılmaktadır. Bu çalışmada, yüz görüntülerini kullanarak PD teşhisi koymak için çoğunluk oylamasına dayalı bir sınıflandırma sistemi öneriyoruz. Modelimiz, Derin Öğrenme (DL) yöntemi olan Konvolüsyonel Sinir Ağı (ESN) çerçevesinde üç farklı özellik seçme tekniğini—Korelasyona Dayalı Özellik Seçimi (CFS), Pearson Korelasyon Katsayısı (PCC) ve En Küçük Mutlak Büzülme ve Seçim Operatörü (LASSO)—entegre etmektedir. Bu üç özellik seçme yöntemi, farklı bakış açıları oluşturmak için kullanılmakta ve ardından çoğunluk oylaması yöntemiyle birleştirilerek sınıflandırma doğruluğu artırılmaktadır. Çalışmada kullanılan veri seti, bir nöroloji uzmanı tarafından etiketlenmiş yüz görüntülerinden oluşmaktadır. Deneysel sonuçlar, önerilen topluluk modelinin bireysel zayıf sınıflandırıcılardan daha yüksek doğruluk oranı elde ettiğini göstermektedir. Bu model, tıbbi uzmanlara PD teşhisinde daha verimli ve doğru bir şekilde yardımcı olma potansiyeline sahip olup, hasta bakımını ve tedavi süreçlerini iyileştirebilir.

Keywords

• Parkinson hastalığı
• Yüz Görüntüleri
• ESN
• Çoğunluk Oylaması
• Sınıflandırma

Ensemble Deep Learning with Majority Voting for Parkinson’s Diagnosis via Facial Images

Abstract

Parkinson's disease (PD) is a progressive neurodegenerative disorder caused by the loss or damage of dopamine-producing brain cells. Early diagnosis is crucial, as timely treatment can enhance patients' quality of life and slow disease progression. Various methods, including brain imaging, neurological tests, handwriting and voice analysis, facial image assessment, and physical examination, are used for PD diagnosis. In this study, we propose a majority voting-based classification system for diagnosing PD using facial images. Our model integrates three different feature selection techniques—Correlation-Based Feature Selection (CFS), Pearson Correlation Coefficient (PCC), and Least Absolute Shrinkage and Selection Operator (LASSO)—within a Convolutional Neural Network (CNN) framework, a deep learning (DL) method. These three feature selection approaches contribute to the design of distinct views, which are then combined through majority voting to enhance classification accuracy. The dataset comprises facial images labeled by a neurology expert. Experimental results indicate that the proposed ensemble model outperforms individual weak classifiers, achieving higher classification accuracy. This model has the potential to assist medical professionals in diagnosing PD more efficiently and accurately, ultimately improving patient care and treatment outcomes.

Keywords

• Facial Images
• CNN
• Majority Voting
• Classification
• Parkinson's Disease

Supporting Institution

TUBITAK

Ethical Statement

Ethical consent for the study was obtained from the ethics committee of Sakarya University of Applied Sciences with the decision dated January 01, 2023 and numbered 70850. All participants provided informed consent before taking part in the study.

Thanks

This study is supported by TUBITAK within the scope of 2209-A University Students Research Projects Support Program.

References

1. Bianchini, E., Rinaldi, D., Alborghetti, M., Simonelli, M., D’Audino, F., Onelli, C., ..., Pontieri, F. E., 2024. The Story behind the Mask: A Narrative Review on Hypomimia in Parkinson’s Disease. Brain Sciences, 14(1), 109.
2. Budak, H., 2018. Feature Selection Methods and a New Approach. Süleyman Demirel University Journal of Natural and Applied Sciences, 22, 21-31.
3. Bukhari, S. N. H., & Ogudo, K. A. (2024). Ensemble Machine Learning Approach for Parkinson’s Disease Detection Using Speech Signals. Mathematics, 12(10), 1575.
4. Chuquimarca, L. E., Vintimilla, B. X., Velastin, S. A., 2024. A review of external quality inspection for fruit grading using CNN models. Artificial Intelligence in Agriculture.
5. Cossini, F., Cuesta, C., Román, K., Zambrano, S., Rubinstein, W., & Politis, D. (2024). Relationship between severity of hypomimia and basic emotion recognition in Parkinson's disease. Revista de neurologia, 79(3), 71-76.
6. Goetz, C. G., 2011. The history of Parkinson's disease: early clinical descriptions and neurological therapies. Cold Spring Harbor perspectives in medicine, 1(1), a008862.
7. Hireš, M., Drotár, P., Pah, N. D., Ngo, Q. C., Kumar, D. K., 2023. On the inter-dataset generalization of machine learning approaches to Parkinson's disease detection from voice. International Journal of Medical Informatics, 179, 105237.
8. Hou, X., Zhang, Y., Wang, Y., Wang, X., Zhao, J., Zhu, X., & Su, J. (2021). A markerless 2d video, facial feature recognition–based, artificial intelligence model to assist with screening for parkinson disease: development and usability study. Journal of medical Internet research, 23(11), e29554.

9. Huang, W., Zhou, Y., Cheung, Y. M., Zhang, P., Zha, Y., & Pang, M. (2022). Facial Expression Guided Diagnosis of Parkinson's Disease via High-Quality Data Augmentation. IEEE Transactions on Multimedia, 25, 7037-7050.
10. Jakubowski, J., Potulska-Chromik, A., Białek, K., Nojszewska, M., Kostera-Pruszczyk, A., 2021. A study on the possible diagnosis of Parkinson’s disease on the basis of facial image analysis. Electronics, 10(22), 2832.
11. Jin, B., Qu, Y., Zhang, L., Gao, Z., 2020. Diagnosing Parkinson disease through facial expression recognition: video analysis. Journal of medical Internet research, 22(7), e18697.
12. Kang, J., Derva, D., Kwon, D. Y., & Wallraven, C. (2019). Voluntary and spontaneous facial mimicry toward other’s emotional expression in patients with Parkinson’s disease. PloS one, 14(4), e0214957.
13. Kannojia, S. P., Jaiswal, G., 2018. Effects of varying resolution on performance of CNN based image classification: An experimental study. International Journal of Computer Sciences and Engineering, 6(9), 451-456.
14. Langevin, R., Ali, M. R., Sen, T., Snyder, C., Myers, T., Dorsey, E. R., Hoque, M. E., 2019. The PARK framework for automated analysis of Parkinson's disease characteristics. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 3(2), 1-22.
15. Nour, M., Senturk, U., Polat, K., 2023. Diagnosis and classification of Parkinson's disease using ensemble learning and 1D-PDCovNN. Computers in biology and medicine, 161, 107031.
16. Özdemir, E. Y., Özyurt, F., 2025. Elasticnet-Based Vision Transformers for early detection of Parkinson’s disease. Biomedical Signal Processing and Control, 101, 107198.
17. Pak, A., Rad, A. K., Nematollahi, M. J., Mahmoudi, M., 2025. Application of the Lasso regularisation technique in mitigating overfitting in air quality prediction models. Scientific Reports, 15(1), 547.
18. Pereira, C., Barros, P., Rodrigues, J., Araújo, P., Borges, R., Almeida, K., & Veras, R. (2024). Parkinson’s Identification in Facial Images Using Pre-Trained Deep Learning Models. In Proceedings of the 12th Regional School on Computing of Ceará, Maranhão, and Piauí, (pp. 169-178). Porto Alegre: SBC.
19. Putro, I. H., Ahmad, T., 2024. Feature Selection Using Pearson Correlation with Lasso Regression for Intrusion Detection System. 12th International Symposium on Digital Forensics and Security (ISDFS), 29-30 April 2024, San Antonio, TX, USA, pp. 1-6.
20. Rajnoha, M., Mekyska, J., Burget, R., Eliasova, I., Kostalova, M., Rektorova, I., 2018. Towards identification of hypomimia in Parkinson's disease based on face recognition methods. 10th International Congress on Ultra-Modern Telecommunications and Control Systems and Workshops (ICUMT), 5-9 November 2018, Moscow, Russia, pp. 1-4.
21. Rameswari, R., Kumar, S. N., Aananth, M. A., Deepak, C., 2021. Automated access control system using face recognition. Materials Today: Proceedings, 45, 1251-1256.
22. Williams-Gray, C. H., Worth, P. F., 2016. Parkinson's disease. Medicine, 44(9), 542-546.
23. Zhou, Y., Pang, M., Huang, W., & Wang, B. (2024, April). Early Diagnosing Parkinson's Disease Via a Deep Learning Model Based on Augmented Facial Expression Data. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1621-1625). IEEE.