Image segmentation plays a pivotal role in the early biomedical diagnosis of various diseases including lung cancer. Lung cancer ranks second in the category of the most common cause of cancer-related deaths. This research focuses on enhancing the segmentation of lung nodules in thoracic computed tomography (CT) scans, utilizing the Lung Image Database Consortium Image Collection (LIDC-IDRI) dataset, which is one of the largest publicly available datasets for lung image analysis. In this study, we address the limitations of the widely used TransUNet architecture by introducing a novel approach, TransUNet with a Double Adaptive Attention (DAA) block. The proposed DAA block is designed to mitigate the inherent bias in convolutional operations and efficiently capture both local and global contextual information. Leveraging 2669 instances from the LIDC-IDRI dataset, containing 7371 nodules, our method achieves a dice score of 96.57%. This high segmentation quality is crucial for accurate and early detection of lung cancer, contributing to improved diagnostics and patient outcomes. The results demonstrate the potential of TransUNet with DAA in helping medical professionals for healthcare and diagnostics in the context of lung cancer.