An Enhanced Fuzzy C-Means-Based Framework for Automated Lung Tumor Segmentation and Cancer Detection from Medical Images

Abstract

Lung cancer is one of the major causes of cancer-related mortality worldwide, and early identification of suspicious pulmonary lesions plays an important role in improving clinical decision-making and treatment planning. Medical image processing techniques provide valuable support for identifying and characterizing abnormal regions in lung images. However, variations in intensity, heterogeneous tumor appearance, noise, and unclear boundaries between healthy and pathological tissues make automated tumor segmentation a challenging task. This study proposes an enhanced Fuzzy C-Means (FCM)-based framework for automated lung tumor segmentation and cancer detection from medical images. The proposed framework consists of image pre-processing, lung-region enhancement, fuzzy clustering, tumor-region identification, post-processing, and classification stages. FCM is employed to exploit the degree of membership of image pixels to different tissue clusters, thereby providing flexible segmentation of regions with uncertain or overlapping intensity characteristics. The enhanced framework incorporates pre-processing and segmentation refinement strategies to improve the identification of suspicious tumor regions and reduce the influence of noise and intensity variations. The segmented regions can subsequently be characterized using relevant image features and used for automated cancer detection. The framework is designed to provide a computer-aided approach for consistent analysis of lung medical images and to support radiological assessment. Performance can be evaluated using segmentation and classification measures such as Dice Similarity Coefficient, Jaccard Index, sensitivity, specificity, accuracy, precision, and F1-score. The proposed approach provides a potential computational framework for automated lung tumor analysis and may contribute to the development of efficient computer-aided diagnostic systems.