ترکیب شبکه عصبی کانولوشن (CNN) و Grad-CAM برای پیش بینی و تفسیر پذیری بصری بیماری پارکینسون

نوع مقاله : مقاله پژوهشی

نویسندگان

گروه مهندسی کامپیوتر، دانشکده مهندسی، دانشگاه شهید چمران اهواز، اهواز، ایران

10.22091/jemsc.2024.10828.1180

چکیده

بیماری پارکینسون یکی از انواع بیماری‌های عصبی است که در اثر تخریب سلول‌های مغزی تولید کننده دوپامین ایجاد می‌شود. تشخیص زودهنگام بیماری پارکینسون عامل مهمی در کاهش سرعت پیشرفت بیماری است. در این مطالعه، از یک شبکه عصبی کانولوشن (CNN) به نام ConvNet برای طبقه‌بندی افراد سالم و افراد مبتلا به بیماری پارکینسون بر اساس تصاویر توموگرافی رایانه‌ای تک فوتونی (SPECT) از پایگاه داده PPMI استفاده شده است. از آنجایی که این مجموعه داده محدود است، پس از یک مرحله پیش پردازش داده‌ها، از دو تکنیک افزایش داده کلاسیک برای جلوگیری از بیش برازش و عملکرد بهتر مدل استفاده شده است. در نهایت از تکنیک Grad-CAM جهت تفسیر پیش بینی‌های انجام شده توسط شبکه عصبی کانولوشن پیشنهادی استفاده شده است. برای ارزیابی روش پیشنهادی از معیارهای متنوعی همچون دقت، حساسیت و f1-score استفاده شده است. نتایج شبیه سازی برحسب معیارهای ارزیابی نشان می‌ دهد که با استفاده از تکنیک افزایش داده کلاسیک می‌ توان طبقه بندی موثرتری انجام داد، به گونه‌ای که روش پیشنهادی از نظر دقت طبقه بندی به دقت 98.5% دست یافت.

کلیدواژه‌ها

موضوعات


عنوان مقاله [English]

Combining Convolutional Neural Network (CNN) and Grad-CAM for Parkinson’s Disease Prediction and Visual Explanation

نویسندگان [English]

  • Reyhaneh Dehghan
  • Marjan Naderan
  • Seyed Enayatallah Alavi
Department of Computer Engineering, Faculty of Engineering, Shahid Chamran University of Ahvaz, Ahvaz, Iran
چکیده [English]

Parkinson's disease is one of the types of neurological diseases that is caused by the destruction of brain cells that produce dopamine. Early detection of Parkinson's disease is an important factor in slowing the progression of the disease. In this study, a Convolutional Neural Network (CNN) namely ConvNet, is used to discriminate Parkinson's patients based on Single Photon Emission Computed Tomography (SPECT) images acquired from the PPMI database. Since the dataset is limited, after a pre-processing stage, two data augmentation techniques are used. Finally, the Grad-CAM technique is used to obtain visual interpretation from the predictions of the proposed CNN. To evaluate the proposed method, different measures such as accuracy, sensitivity (recall) and f1-score are used. Simulation results according to the measures shows that when the classic data augmentation method is used accuracy is increased to 98.50% and more efficient classification is performed.

کلیدواژه‌ها [English]

  • Parkinson's disease (PD)
  • Convolutional Neural Network (CNN)
  • SPECT images
  • data augmentation
  • Grad-CAM
  1. Appineni, A. and Gupta, A. (2023). Preemptive Diagnosis of Parkinson's Disease through DaT Scans using InceptionV3-based Convolutional Neural Networks. https://doi.org/10.20944/preprints202312.2268.v1
  2. Faull, R. and Laverty, R. (1969). Changes in dopamine levels in the corpus striatum following lesions in the substantia nigra. Experimental neurology, 23(3): p. 332-340. https://doi.org/10.1016/0014-4886(69)90081-8
  3. Poewe, W., et al., (2017). Parkinson disease. Nature reviews Disease primers, 3(1): p. 1-21. https://doi.org/10.1038/nrdp.2017.13
  4. Wang, W., et al., (2020). Early detection of Parkinson’s disease using deep learning and machine learning. IEEE Access, 8: p. 147635-147646. https://doi.org/10.1038/nrdp.2017.13
  5. Wile, D.J., et al., (2017). Serotonin transporter binding and motor onset of Parkinson's disease in asymptomatic LRRK2 mutation carriers: a cross-sectional study. The Lancet. Neurology, 16(5): p. 351. https://doi.org/10.1016/S1474-4422(17)30056-X
  6. Marek, K., et al., (1996). [sup 123 I] beta-CIT/SPECT imaging demonstrates bilateral loss of dopamine transporters in hemi-Parkinson's disease. Neurology, 46(1): p. 231-237. https://doi.org/10.1212/WNL.46.1.231
  7. Mohammed, F., He, X. and Lin, Y. (2021). Retracted: An easy-to-use deep-learning model for highly accurate diagnosis of Parkinson's disease using SPECT images. Computerized Medical Imaging and Graphics. https://doi:10.1016/j.compmedimag.2020.101810
  8. Govindu, A. and Palwe, S. (2023). Early detection of Parkinson's disease using machine learning. Procedia Computer Science, 218: p. 249-261. https://org/10.1016/j.procs.2023.01.007
  9. Tuncer, T., Dogan, S. and Acharya, U.R. (2020). Automated detection of Parkinson's disease using minimum average maximum tree and singular value decomposition method with vowels. Biocybernetics and Biomedical Engineering, 40(1): p. 211-220. https://doi.org/10.1016/j.bbe.2019.05.006
  10. Magesh, P.R., Myloth, R.D. and Tom, R.J. (2020). An explainable machine learning model for early detection of Parkinson's disease using LIME on DaTSCAN imagery. Computers in Biology and Medicine, 126: p. 104041. https://doi.org/10.1016/j.compbiomed.2020.104041
  11. Hathaliya, J., et al., (2022). Convolutional neural network-based Parkinson disease classification using SPECT  imaging data. Mathematics, 10(15): p. 2566. https://doi.org/10.3390/math1015256
  12. Marek, K., et al., (2011). The Parkinson progression marker initiative (PPMI). Progress in neurobiology, 95(4): p. 629-635. https://10.1016/j.pneurobio.2011.09.005
  13. Martinez-Murcia, F.J., et al., (2018). Convolutional neural networks for neuroimaging in Parkinson’s disease: Is preprocessing needed?, International journal of neural systems, 28(10): p. 1850035. https://doi.org/10.1142/S0129065718500351
  14. Mash, R., Borghetti, B. and Pecarina, J. (2016). Improved aircraft recognition for aerial refueling through data augmentation in convolutional neural networks. in Advances in Visual Computing: 12th International Symposium, ISVC 2016, Las Vegas, NV, USA, December 12-14, Proceedings, Part I 12. Springer. https://doi.org/10.1007/978-3-319-50835-1_11
  15. Dehghan, R., Naderan, M. and Alavi, S.E. (2022). Detection of Parkinso’s disease using Convolutional Neural Networks and Data Augmentation with SPECT images. in 2022 12th International Conference on Computer and Knowledge Engineering (ICCKE). IEEE. 17-18 Nov., Mashhad, Iran. https://doi.org/10.1109/ICCKE57176.2022.9960085
  16. Zhou, B., et al. (2016). Learning deep features for discriminative localization. in Proceedings of the IEEE conference on computer vision and pattern recognition. https://doi.org/10.1109/CVPR.2016.319
  17. Selvaraju, R.R., et al. (2017). Grad-cam: Visual explanations from deep networks via gradient-based localization. in Proceedings of the IEEE international conference on computer vision. 22-29 October, Venice, Italy. https://doi.org/10.1109/ICCV.2017.74
  18. Krstinić, D., et al., (2020). Multi-label classifier performance evaluation with confusion matrix. Computer Science & Information Technology, 1: p. 1-14. https://10.5121/csit.2020.100801
CAPTCHA Image