Quantitative Assessment of Lung Cancer on Computed Tomography Images using Mate Lap - Imaging Processing

Abstract

Background: Lung cancer remains one of the major global health problems and one of the most important causes of cancer deaths worldwide. The early and accurate diagnosis of the disease is important for better treatment and better survival. Traditional clinical interpretation, based on reading CT images, is often a time-consuming process and is subject to observer variability. Hence, there exists an urgent need for CAD systems using advanced image processing techniques for automated diagnosis. Aim: Design and evaluate a robust image-processing framework for early detection of lung cancer from CT-scan images while focusing on optimum trade-off between diagnostic accuracy and computational efficiency. Materials and Methods: The approach encompasses the following steps: the image enhancement using Gabor, Mean, and Median filters in order to improve the contrast and suppress the noise in the image; lung segmentation using the Marker-Controlled Watershed algorithm that provides effective delineation of the Region Of Interest; and feature extraction using a pixel-based analysis for characterizing possible tumor regions. To this aim, two techniques have been compared, namely Binarization and Masking. These were compared in terms of their performances using different measures like accuracy and computational time. Results: Experimental results indicate a very high True Acceptance Rate (TAR), thus showing great detection capability. Masking resulted in higher accuracy as compared to Binarization but with a corresponding increase in computational time. Conclusion: The developed framework enhances diagnostic reliability, reduces false-positive rates, and provides radiologists with effective CAD support for early lung cancer detection. Future work will focus on optimizing algorithmic efficiency and integrating machine learning techniques for improved automated classification.