Lung Segmentation using a ConvMixer-Based Framework

Accurate lung segmentation is crucial for diagnosing various lung diseases from chest X-ray images. U-Net and its variations have been successfully applied in the medical field, but they have limitations in capturing global context information due to the local nature of convolution operations. In this paper, we propose a novel Conv-Mixer based model for lung segmentation. This model is inspired by the ConvMixer model to effectively capture local and global features in lung images. The proposed model is evaluated on two public chest X-ray datasets: Shenzhen and Montgomery; and achieves superior performance compared to the state-of-art segmentation methods. Specifically, it achieves an accuracy of 97.52% and an IoU of 92.71%. The proposed model is a promising method for lung segmentation and has the potential to aid in the diagnosis of various lung diseases.