Table 1.
Summary of brain observation approach reviewed in this study
| References | Detection | Task | Data Set | Accuracy | Sensitivity | Specificity |
|---|---|---|---|---|---|---|
| Mosarrat Rumman et al. [10] |
123I-Ioflupane SPECT images |
Their proposed ANN Architecture | 100 PD,100 HC | 94% | 100% | 88% |
| Thakur et al. [22] | SPECT images | DenseNet-121 Architecture with Soft attention block | 1160 PD, 230 HC | 99.2% | 99.2% | 99.4% |
| Kurmi et al. [21] | DaTscan images | 4 CNN and Fuzzy Rank Level Fusion | 432 PD,213 HC | 98.45% | 98.84% | 97.67% |
| Magesh et al. [20] | DaTscan images | VGG 16 with transfer learning | 430 PD,212 HC | 95.2% | 97.5% | 90.9% |
| Prashant et al. [12] | SPECT images | SVM RBF using SBR values | 369 PD,179 HC | 96.14% | 96.55% | 95.03% |
| Prashant et al.[13] | SPECT images | SVM | 427 PD,208 HC,80 SWEDD | 97.29% | 97.37% | 97.18% |
| Ortiz et al. [11] | DaTscan images |
LeNet AlexNet |
158 PD,111 HC |
(1) 0.95 ± 0.03 (2) 0.95 ± 0.03 |
(1) 0.94 ± 0.04 (2) 0.95 ± 0.05 |
(1) 0.95 ± 0.04 (2) 0.95 ± 0.04 |
| Chakraborty et al. [23] | MRI images | 3D CNN | 203 PD, 203 HC | 95.29% | – | 0.9430 |
| Kaur et al. [25] | MRI images | GAN-based transfer learning AlexNet and data augmentation | 67 PD, 85 HC | 89.23% | 90.27% | 89.03% |
| Sivaranjini and Sujatha [29] | MRI images | AlexNet and Transfer learning, data augmentation | 82 PD, 100 HC | 88.9% | 89.3% | 88.4% |
| Solana-Lavalle and Rosas-Romero [30] | MRI images |
VBM,7 classifiers (1) Men (2) Women |
330 PD,150 HC |
(1) 99.01% (2) 96.97% |
(1) 99.35% (2) 100% |
(1) 100% (2) 96.15% |
| Yuvaraj et al. [43] | EEG signals | PD diagnosis index based on Higher-Order Spectra feature, SVM with RBF kernel | 20 PD, 20 HC | 99.62% | 100% | 99.25% |
| Lee et al. [37] | EEG signals |
CRNN based model |
1st dataset: 600 PD, 630 HC 2nd dataset: 1200 PD,1260 HC |
99.2% | – | – |
| Khare et al. [39] | EEG signals |
PDCNNet model SPWVD & 2-D CNN |
Dataset 1: 15 PD,16 HC Dataset 2: 20 PD,20 HC |
Dataset 1: 100% Dataset 2: 99.97% |
Dataset 2: 100% |
Dataset 2: 99.94% |
| Oh et al. [36] | EEG signals | CNN based CAD system | 20 PD,20 HC | 88.25% | 84.71% | 91.77% |
| Majid Nour [44] | EEG signals | Ensemble Approach | 15 PD,16 HC | 99.31% | – | – |