Vision Transformer based approach for accurately detecting cervical cancer

Globocan 2022 has reported 6,60,000 cervical cancer diagnosis and 3,50,000 cancer deaths worldwide, making it the fourth most common cancer worldwide. It is caused by the infection of the human papillomavirus (HPV), and this virus is present in over 99% of cervical cancers. Although the pap smear test is frequently used to identify cervical cancer, medical professionals evaluate its accuracy. To increase speed and accuracy, diagnosis systems make use of deep learning and medical image analysis. This paper investigates the potential of vision transformer models in differentiating between healthy and cancerous cells. The publicly accessible SIPAKMED dataset is used to evaluate the experimental study. Six deep learning models, included three distinct CNNs and three vision transformer models, are compared in terms of performance. It is found that the average test accuracy of the ViT models has outperformed the average test accuracy of the CNN models. This proposes the fact that, the role of vision transformers in medical imaging classification tasks is quite promising.