Attention-ResUNet and EfficientSASM-UNet: UNet-Based Frameworks for Lung and Nodule Segmentation

Lung cancer has been one of the major threats across the world, with the highest mortalities. Computer-aided detection (CAD) can aid early detection and thus increase the survival rate. Accurate lung parenchyma segmentation (to include the juxta-pleural nodules) and lung nodule segmentation, the primary symptom of lung cancer, play a crucial role in the overall accuracy of the Lung CAD pipeline. Lung nodule segmentation is quite challenging due to the diverse nodule types and other inhibiting structures within the lung lobes. Traditional machine/deep learning methods suffer from generalization and robustness. Recent Vision Language Models/Foundation Models perform well at the anatomical level but struggle on fine-grained segmentation tasks, and their semi-automatic nature limits their effectiveness in real-time clinical scenarios. In this paper, we propose two novel, fully automatic methods for accurate 3D segmentation of lung parenchyma and lung nodules. The proposed architecture is an attention-based, fully convolutional UNet with residual blocks at each encoder-decoder state. Dilated convolutions at each encoder-decoder stage allow the model to capture broader context without increasing computational cost. For Lung Nodule Segmentation, an Efficient-SASM (Self-Attentive Similarity Module)- based 3D-UNet is proposed, optimized for easy deployment. The proposed methods have been extensively evaluated on one of the largest publicly available datasets, namely LUNA16, and compared with recent notable work in the domain using standard performance metrics such as Dice score and IOU. The results show that the proposed Lung Segmentation method outperforms state-of-the-art methods, while the proposed Nodule Segmentation technique achieves comparable results. The source code, datasets, and pre-processed data can be accessed using the link: https://github.com/EMeRALDsNRPU/Attention-Based-3D-ResUNet.