This paper introduces a novel method for segmenting and identifying lung tumors in medical images called FMN-SAL. The proposed approach com-bines the strengths of semi-supervised learning and fuzzy clustering, enabling effective tumor segmentation and identification even with limited labeled data. By leveraging feature membership functions based on spatial and pixel intensity, the model captures nonlinear relationships in medical images, ensuring robust performance. Key contributions include: (1) Incorporation of semi-supervised and active learning to enhance tumor segmentation accuracy; (2) Use of piecewise-linear membership functions to handle uncertainties in in-tensity and spatial data; (3) A novel tumor identification framework refining segmentation outputs for lung tumor localization; and (4) Comprehensive evaluation on lung tumor datasets, demonstrating superiority over existing techniques. This research provides a powerful diagnostic tool for lung tumor detection, enhancing clinical decision-making and treatment outcomes.