Automated lung field detection in chest X-ray (CXR) images is critical for medical imaging as it is a significant aspect of analyzing and diagnosing different pulmonary disorders. This study employs different deep learning-based U-net architectures for automated lung segmentation using an encoder-decoder to obtain pixel-level accuracy. The methods are compared in terms of their accuracy, precision, recall, and F1-score, as well as the Jaccard similarity and Dice coefficients. The proposed ResNet-based U-Net had the best prediction performance with an accuracy of 98.37% and a Dice Coefficient of 96.87%, which signifies that the model had better feature extraction and generalization ability. The results were further analyzed using loss curves for training and validation. The analysis shows that state-of-the-art deep learning techniques can provide accurate lung segmentation, which can help create practical diagnostic applications in medical imaging. This work will help healthcare professionals make better, faster diagnoses of lung diseases using automated segmentation tools.