Lung CT Image Segmentation Using VGG-16 Network with Image Enhancement Based on Bounded Turning Mittag-Leffler Function

Automated segmentation of diseases considered a necessary initial step in routine diagnosis. Lung diseases that affect the lungs, such as pneumonia or lung collapse can result in areas of consolidation or atelectasis, where lung tissue becomes denser or collapses. It can be challenging to accurately segment such areas, as they may exhibit similar characteristics to adjacent structures. This study proposes a lung CT image segmentation method which includes two steps: (1) a new image enhancement model that uses the bounded turning Mittag-Leffler function to improve the CT images for better segmentation outcomes. (2) a new modified VGG-16 for infection of lung segmentation based on expanding the original VGG-16 network. The dilated convolutional layers are added to the original VGG-16 network to create a new lung CT image segmentation method. The experimental results showed that the proposed method can accurately segment the infected region in lung CT scans. The results led to Accuracy, Dice Coefficient and Jaccard Index values of 96.3%, 91.2%, and 82.3% respectively. The proposed method is accurate and suitable for implementation in real-world applications. Following result computation, seven related studies are compared with the recommended methodology. This demonstrated how well this study had performed in comparison to many earlier studies. Despite the fact that segmenting lung CT images requires a lot of work. Obtaining a suitable level of accuracy was quite difficult.