Robust Pre-Processing Strategies for Early Lung Cancer Diagnosis with Low-Dose CT Scans

Cancer that starts in the lungs due to the growth of abnormal cell in uncontrolled way is known as lung cancer. Tumors that prevent the lung from supplying blood with oxygen may develop because of it. So, it is important to detect it in its early stages. Low-Dose Computed Tomography (LDCT) scans play very important role to detect it in early stages. This work represents the various pre-processing techniques which are applied to the dataset of LDCT scan images to enhance the efficacy and accuracy of the system to detect Lung cancer. The several pre-processing techniques, including Gaussian Filter, TV (Total Variation) Method, Bilateral Method, Wavelet Method, Invariant Wavelet Method, NLM (Non-Local Means) Method and BM3D (Block-Matching & 3D filtering) Methods are used in this study. These methods are able to perform positive results for image enhancement, noise reduction, and feature extraction. The results of the study show that these pre-processing techniques not only improve the clarity of LDCT scan images but also enable more precise identification of suspected tumor locations. These pre-processing methods also improve specificity and sensitivity of the system. Through the ability to intervene earlier and increase the chances of patient survival, this discovery holds the potential to save lives. It also has a great deal of potential to advance the field of early lung cancer detection.