Table 1 Comparison of methodologies used in relevant research papers on brain tumor radiogenomic classification
Study | Data | Problem | Classifier | GAN method | No augmentation | Classic/affine augmentation | Gan augmentation |
|---|---|---|---|---|---|---|---|
[37] | Brain MRI | Tumor Detection | ResNet-50 | PGGAN | Accuracy: 90.06 Sensitivity: 85.27 Specificity: 97.04 | Accuracy: 90.70 Sensitivity: 88.70 Specificity: 93.62 | Accuracy: 91.08 Sensitivity: 86.60 Specificity: 97.60 |
[40] | Pancreas CT | Lesion Detection | Faster RCNN | TMP-GAN | - | Precision: 82.64 Recall: 81.26 F1 Score: 81.94 | Precision: 86.28 Recall: 85.89 F1 Score: 86.08 |
[41] | Brain CT | Cerebrospinal Fluid (CSF) segmentation | UNet | PCGAN | Accuracy: 88.1 | - | Accuracy: 89.3 |
[42] | Brain MRI | Multi-class brain tumor classification | CNN | MSG-GAN | Accuracy: 90.3 | - | Accuracy: 93.1 |
[43] | Brain MRI | Medical Image Generation | No Classifier | DCGAN, WGAN | - | - | Accuracy for DCGAN: 77 Accuracy for WGAN: 66 |
[44] | Brain MRI | Tumor Classification | CNN | DCGAN | Accuracy: 91.70 Sensitivity: 90.16 Specificity: 95.58 Precision: 91.17 F1-score: 90.54 | - | Accuracy: 93.01 Sensitivity: 92.01 Specificity: 96.39 Precision: 92.28 F1-score: 92.10 |
[45] | Brain MRI | MRI Generation and Classification | CNN MobileNetV2 ResNet152V2 | DCGAN | - | Accuracy: 95.84-99.09 Precision: 96.48-99.12 Recall: 95.68-99.08 | Accuracy: 93.27-97.94 Precision: 93.24-96.91 Recall: 92.19-96.91 |
[46] | Brain MRI | Tumor Classification | Multi-stream 2D CNN | PGGAN | Accuracy: 82.45 Sensitivity: 70.51 Specificity: 87.4 | - | Accuracy: 88.82 Sensitivity: 81.81 Specificity: 92.17 |
[47] | Liver CT | Liver Lesion Classification | CNN | DCGAN, ACGAN | Accuracy: 57 | Sensitivity: 78.6 Specificity: 88.4 | Sensitivity: 85.7 Specificity: 92.4 |