Table 4 Comparison outcomes of proposed and traditional method validation accuracy based on k-fold cross-validation
From: Multimodal Biomedical Image Segmentation using Multi-Dimensional U-Convolutional Neural Network
Type | Proposed MDU-CNN (%) | Traditional U-Net (%) | Comparative Increase in performance (%) |
|---|---|---|---|
Magnetic Resonance Image | 79.2130 ± 0.8182 | 77.2289 ± 0.6923 | 1.3241 |
Non-invasive dermoscopy | 81.3188 ± 0.4423 | 76.1256 ± 3.9834 | 5.1932 |
Microbes Fluorescence Microscopy | 92.6228 ± 0.9816 | 88.1209 ± 1.9923 | 4.5019 |
Endoscopy | 83.1567 ± 1.6822 | 72.9190 ± 1.3989 | 10.2377 |
Electron Microscopy | 88.8651 ± 0.8012 | 87.9929 ± 0.7717 | 0.8722 |