Hybrid Approach for COVID-19 Segmentation: Integrating ResNet-Darknet19 Based Transfer Learning with Radiomics Features

The COVID-19 pandemic, caused by the SARS-CoV-2 virus, has underscored the critical need for accurate and timely diagnosis to mitigate its spread and optimize patient management. Chest imaging, specifically computed tomography (CT) scans, has emerged as a valuable diagnostic tool due to its sensitivity in detecting COVID-19-related abnormalities. This paper presents a hybrid approach using ResNet-Darknet19 architecture and radiomics features to improve COVID-19 lesion segmentation accuracy. The method outperforms deep learning networks in COVID-19 lesion segmentation, achieving 94.56% accuracy compared to manually defined annotations. PyRadiomics was used for image-based feature extraction, achieving 98% accuracy. The model's generalizability is assessed using a separate test dataset, offering promising prospects for aiding in the diagnosis and treatment of COVID-19 patients.