Early detection of Coronavirus Disease 2019 , an infectious disease caused by the SARS-CoV-2 virus, is crucial in minimizing the risk of mortality and limiting its spread, particularly among asymptomatic individuals. Computed tomography (CT) scans of the chest are commonly employed for diagnosing this condition, necessitating the development of segmentation techniques for analyzing these images effectively. However, segmenting COVID-19 CT images poses considerable challenges due to the indistinct boundaries between gray and white matter, as well as the homogeneous and ambiguous structures within the regions. To address these issues, we propose a hybrid approach that combines Undecimated Wavelet Transform (UWT), Fuzzy Clustering (FC), and Anisotropic Diffusion Filter (ADF). Our method involves utilizing UWT to denoise CT images in the frequency domain, followed by an advanced fuzzy clustering technique based on texture features and local gray value entropy for autonomous segmentation of CT images. The segmented images are then processed with ADF to eliminate uncertainty and noise. The performance of our proposed method was evaluated visually and through similarity measurements using an open-source dataset. A comparative analysis with alternative segmentation methods was conducted using multiple metrics, including Dice, Jaccard, Precision, Accuracy, Sensitivity, F-measure, MCC, and Specificity. Our results demonstrate that the proposed hybrid approach significantly enhances the detection of COVID-19 from CT images.

No abstract