Automated Lung Cancer Diagnosis Using Vision Transformers (ViT) on CT Scan Data

In this study, it brings into focus, the use of Vision Transformers (ViTs) in the diagnostic of lung cancer with the aid of Computed Tomography (CT) scan data. Taking advantage of the self-attention properties of ViTs the presented model considers global dependencies in imaging data and as a result demonstrates higher diagnostic performance. The training and validation of the proposed model were done using the public dataset, namely the LIDC-IDRI dataset and the classification accuracy is 92%. 5%, a specificity of 88 and the overall accuracy was 91%. 2% for detecting tumours in tissues examined while having a sensitivity of 88. 8%. This shows that ViTs can be made as a substitute to the prior CNN-based models and can be used as a reliable and efficient diagnostic tool in the early stage of lung cancer. These findings may be helpful in making clinical decision and also for increasing survival rate among the patients of lung cancer.