Parkinson’s Identification in Facial Images Using Pre-Trained Deep Learning Models

• Caio B. A. A. Pereira UFPI
• Pedro H. X. R. Barros UFPI
• José N. A. Rodrigues IFPI
• Paulo H. R. Araújo UFPI
• Rodrigo N. Borges UFPI
• Kelson J. Almeida UFPI
• Rodrigo M. S. Veras UFPI

DOI: https://doi.org/10.5753/ercemapi.2024.243751

Abstract

This article evaluated the use of pre-trained deep learning models in the classification of facial images to differentiate between healthy individuals and patients with Parkinson’s Disease (PD). We utilized a dataset consisting of 340 images of PD patients and 358 images of healthy individuals, applying techniques such as group cross-validation with 5 folds, data augmentation, and fine-tuning. The best performance was achieved with the DenseNet-201 model, which showed an average accuracy of 92.77%, F1-score of 92.48%, and a Kappa index of 85.33%. These results suggest that pre-trained CNNs are promising for the detection of Parkinson’s Disease in facial images.

References

Hou, X., Zhang, Y., Wang, Y., Wang, X., Zhao, J., Zhu, X., and 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.

Huang, W., Zhou, Y., Cheung, Y.-m., Zhang, P., Zha, Y., and Pang, M. (2023). Facial expression guided diagnosis of parkinson’s disease via high-quality data augmentation. IEEE Transactions on Multimedia, 25:7037–7050.

Jiang, R., Chazot, P., Pavese, N., Crookes, D., Bouridane, A., and Celebi, M. (2022). Private facial prediagnosis as an edge service for parkinson’s dbs treatment valuation. IEEE Journal of Biomedical and Health Informatics, 26(6):2703–2713.

Landis, J. R. and Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1):159–174.

Lee, H. M. and Koh, S. B. (2015). Many faces of parkinson’s disease: Non-motor symptoms of parkinson’s disease. Journal of Movement Disorders, 8(2):92–97.

Mumuni, A. and Mumuni, F. (2022). Data augmentation: A comprehensive survey of modern approaches. Array, 16:100258.

Pegolo, E., Volpe, D., Cucca, A., Ricciardi, L., and Sawacha, Z. (2022). Quantitative evaluation of hypomimia in parkinson’s disease: A face tracking approach. Sensors, 22(4).

Powers, D. M. W. (2020). Evaluation: from precision, recall and f-measure to roc, informedness, markedness and correlation.

Rajnoha, M., Mekyska, J., Burget, R., Eliasova, I., Kostalova, M., and Rektorova, I. (2018). Towards identification of hypomimia in parkinson’s disease based on face recognition methods. In 2018 10th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT), pages 1–4.

Tajbakhsh, N., Shin, J. Y., Gurudu, S. R., Hurst, R. T., Kendall, C. B., Gotway, M. B., and Liang, J. (2016). Convolutional neural networks for medical image analysis: Full training or fine tuning? IEEE transactions on medical imaging, 35(5):1299–1312.