Table 1.
The state-of-the-art machine and deep-learning techniques applied on awake EEG for PD diagnosis, staging and biomarkers identification.
Fig 1.
Proposed deep-learning framework.
Fig 2.
(a) N3-CWT for PD-NC (b) N3-CWT for PD-MCI (c) N3-VMD for PD-NC (d) N3-VMD for PD-MCI.
Fig 3.
Proposed CNN for (a) N3 and REM CWT data (b) N2 CWT data (c) VMD data.
Table 2.
Components of the proposed CNN model applied on the CWT data.
Table 3.
Components of the proposed CNN model applied on the VMD data.
Table 4.
CWT dataset description.
Table 5.
Proposed CNN training parameters.
Table 6.
Performance of the proposed CWT-based CNN.
Fig 4.
ROC of the proposed classifier applied on (a) N2 (C3) (b) N3 (F3) (c) REM (C3).
Table 7.
Confusion matrix of the proposed CWT-based CNN on N2 (C3).
Table 8.
VMD dataset description.
Table 9.
Model training parameters.
Table 10.
Performance of the proposed VMD-based CNN.
Fig 5.
ROC of the proposed classifier applied on (a) N2 (C3) (b) N3 (F3) (c) REM (C3).
Table 11.
Confusion matrix of the proposed CWT-based CNN on N2 (C3).
Table 12.
Comparison of the computational-time for both the CWT-based and VMD-based CNN frameworks.
Fig 6.
VMD gray-scale images and their corresponding heat maps where the rows from top to bottom represent the N2 (C3), N3 (F3) and REM (C3) respectively.
Also, the first and last two columns represent the VMD images for PD-MCI and PD-NC respectively.
Table 13.
Comparison of the proposed approach and the-state-of-the-art architectures for PD detection.