Table 1. Summary of the major classification studies on colon cancer.
| Authors in | Dataset used | CNN architecture |
Accuracy | Using pretrained either feature extraction/fine- tuning |
|---|---|---|---|---|
| Stoean (2020) | colorectal in Stoean et al. (2016) |
CNN model from scratch | 92% | Fine-tune: only kernel size and number of kernels in CNN using EA method |
| Popa (2021) | colorectal in Stoean et al. (2016) |
AlexNet and GoogleNet | 89% | feature extractor |
| Postavaru et al. (2017) | colorectal in Stoean et al. (2016) |
CNN model from scratch | 91% | The number of filters and the kernel size |
| Lichtblau & Stoean (2019) | colorectal in Stoean et al. (2016) |
AlexNet | 87% | Feature extractor with ensemble learning |
| Ohata et al. (2021) | colorectal in Kather et al. (2016) |
Set of pretrained models (VGG16, Inception, Resent) |
92.083% | Feature extraction |
| Rachapudi & Lavanya Devi (2021) | colorectal in Kather et al. (2016) |
CNN architecture | 77% | Fine-tune CNN model |
| Dif & Elberrichi (2020a) | colorectal in Kather et al. (2016) |
Pretrained Resnet121 | 94% | Feature extraction |
| Boruz & Stoean (2018) | colorectal in Stoean et al. (2016) |
Contour low-level image features |
92.6% |