An Early Detection of Oral Epithelial Dysplasia Based on GoogLeNet Inception-v3

Oral cancer is the sixth most common type of cancer globally, with a high mortality rate, and early diagnosis can increase the survival rate Oral cancer is usually preceded by oral epithelial dysplasia (OED); thus, the detection of OED is critical forthe early diagnosis of oral cancer. The manual diagnosis and classification of OED is a time-consuming and tedious task. Hence is a need for accurate automated detection of OED, and this study armed to develop such detection model based on GoogLeNet Inception-v3 and transfer learning. We utilized data augmentation and transfer learning to overcome the overfitting problems that occur with training GoogLeNet Inception-v3 with a small dataset. Consequently, our proposed model achieved an accuracy, precision, recall and F1-score of 92.5%, 90%, 95.5%, and 92.8%, respectively.