Early Diagnosis of Lung Infection via Deep Learning Approach

Abstract

The rapid global spread of COVID-19 and RT-PCR tests are insensitive in early infection phases, according to hospitals. To find Covid-19, a fast, accurate test is needed. CT scans have shown diagnostic accuracy. CT scan processing using a deep learning architecture may improve illness diagnosis and treatment. A deep learning system for COVID-19 detection was derived using CT scan features. Using and comparing numerous transfer-learning models, fine-tuning, and the embedding process yielded the best infection diagnostic results. All models' diagnostic effectiveness was assessed using 2482 CT scan images. The optimized model demonstrated encouraging outcomes by significantly enhancing the sensitivity metric (86.26±1.72), a critical factor in accurately detecting COVID-19 infection. Additionally, the resulting model demonstrated elevated values for accuracy (81.15±0.17), specificity (77.90±1.33), precision (76.79±0.80), F1_score (81.24±0.37), and AUC (81.88±0.2). Deep learning methodologies have been effectively employed to detect COVID-19 in chest CT scan images. In the future, the suggested approach may be employed by clinical practitioners to study, identify, and effectively mitigate a greater number of pandemics.