Tritention U-Net: A Modified U-Net Architecture for Lung Tumor Segmentation

Abstract

Lung tumor segmentation in computed tomography (CT) images is a critical task in medical image analysis. It aids in the early detection and diagnosis of lung cancer, which is one of the primary causes of cancer deaths around the world. However, because of the variable sizes, uncertain shapes of lung nodules, and complex internal lung structure, lung tumor segmentation is a difficult problem. In this study, we propose a novel Tritention U-Net as an efficient model for solving that problem. It is integrated with the Tritention Gate on the contracting path between the encoder and decoder to highlight task-relevant salient features. The proposed Tritention U-Net model is trained and evaluated using the Medical Image Decathlon dataset - Task06_Lung, which requires the model to segment a small portion of a lung image. Our model achieved a Dice score of 91.80% and was compared to well-known models to demonstrate the improvement.