Transforming Lung Cancer Diagnosis with Twin Vision Transformers

Lung cancer, a major global cause of death, underscores the need for early detection to improve survival rates. Computer-Aided Detection (CAD) systems are pivotal in speeding up lung cancer identification, potentially saving lives. In artificial intelligence, deep learning is increasingly applied in medical contexts. This paper addresses lung cancer challenges, particularly Non-Small Cell Lung Cancer (NSCLC), proposing a Twin Transformer Architecture for accurate CT scan classification into NSCLC subtypes (Squamous Cell lung cancer, Adenocarcinoma, Large Cell Carcinoma), achieving notable accuracy 94.3% and precision 94.49%. TwinViT, distinct with twin and parallel processing, offers efficiency and scalability, enhancing performance by combining independently acquired representations.