Deep Learning-Driven COVID-19 Lesion Segmentation in CT Scans with R-EDNet

Accurate segmentation of COVID-19 lesions in CT images is critical for effective clinical diagnosis and treatment planning. This study presents R-EDNet, a deep learning framework based on the ResNet50 architecture within an encoderdecoder structure for precise segmentation of COVID-19 lesions. The proposed model integrates convolutional and decoder blocks with selected pre-trained ResNet50 layers, optimizing feature extraction and segmentation accuracy. R-EDNet achieved significant results, including a Dice coefficient of 93.87%, demonstrating its capability to effectively delineate COVID-19 lesions. Additional metrics, such as a specificity of 99.84% and a sensitivity of 92.09%, further validate its robustness. Comparative analysis highlights R-EDNet's performance, achieving a Dice coefficient of 93.87%, surpassing several state-of-the-art models. These results establish R-EDNet as a reliable and efficient tool for automated lesion segmentation, with the potential to enhance diagnostic accuracy, clinical decision-making, and patient care. The framework demonstrates promise for broader applications in medical imaging and contributes to advancements in automated segmentation techniques.