World Journal of Engineering Research and Technology

Urvashi B. Deshmukh* and Dr. Prapti D. Deshmukh

ABSTRACT

Feature extraction plays a crucial role in medical image analysis, enabling the identification of informative patterns for improved diagnosis and prognosis. This study introduces a comprehensive framework for extracting feature from CT scan of lung diseases, including lung cancer, Covid-19 and pneumonia categorize into mild, moderate and severe cases. The Proposed methodology combines wavelet packet decomposition (WPD) with statistical, texture, shape, edge detection, and Gray level co-occurrence matrix (GLCM) features to capture both frequency-based and spatial characteristics of theimages. WPD decomposes the image into multiple frequency subbands, while the additional descriptors analyse pixel intensity distributers, structural properties and texture patterns. For each subband decomposition, we calculate various statistical features such as Mean, Variance, Energy and Entropy. Result demonstrate the efficiency of this framework in extracting meaningful features that represents the complexity of lung disease images, providing a solid foundation for further analysis and classification. This approach contributes to enhancing automated diagnostic system for medical imaging by leveraging diverse image characteristics across varying severity levels.

[Full Text Article] [Download Certificate]