TABLE 6. Comparison of Proposed 3DDCNN With State-of-the-Art on Clinical Dataset Using Different Statistical Metrics, Namely, Mean Detection Error Rate (Mean 
), Mean Classification Error Rate (Mean 
), Variance Detection Error Rate (Var 
), Variance Classification Error Rate (Var 
), Standard Deviation Detection Error Rate (Std 
), Standard Deviation Classification Error Rate (Std 
), Mean Average Precision (m-AP) and Processing-Time (Time).
| Methods | Score | Mean 
|
Mean 
|
Var 
|
Var 
|
Std 
|
Std 
|
m-AP (%) | Time (min) |
|---|---|---|---|---|---|---|---|---|---|
| Mask R-CNN [35] | 83.63±1.34 | 2.20±1.2 | 2.38±4.5 | 2.3±0.5 | 1.23±0.27 | 4.18±3.30 | 4.3±1.8 | 54.3±2.5 | 11 |
| RetinaNet [36] | 83.68±4.10 | 3.94±0.71 | 1.93±2.8 | 4.6±1.2 | 1.62±0.30 | 3.40±4.31 | 3.9±3.0 | 56.4±2.8 | 7–8 |
| Retina U-Net [37] | 89.4±.91 | 2.64±2.3 | 1.84±2.3 | 1.1±0.7 | 1.79±0.95 | 2.82±2.32 | 4.2±1.7 | 62.9±3.4 | 5–6 |
| Fast R-CNN [38] | 79.5±3.70 | 5.69±3.94 | −0.04±4.4 | 5.8±1.3 | 1.97±0.90 | 8.58±8.02 | 9.9±2.2 | 44.9±3.0 | 21.7 |
| Faster R-CNN [24] | 76.3±2.31 | 6.07±2.92 | 1.04±6.8 | 1.7±0.6 | 1.61±0.99 | 7.41±4.31 | 9.31±5.1 | 52.8±10.7 | 17.2 |
| Proposed 3DDCNN | 84.2±1.20 | 3.32±2.8 | 2.96±1.3 | 1.2±2.35 | 1.24±0.51 | 3.69±3.87 | 4.01±5.2 | 55.8±4.1 | 4–6 |
| Proposed 3DDCNN (Using mRPN) | 87.8±1.35 | 3.30±0.58 | 1.28±0.19 | 1.4±2.48 | 1.10±0.26 | 2.88±3.61 | 4.5±3.8 | 59.7±3.2 | 2–4 |