Developing a Computational Tool for Lung Cancer Prediction with Data Mining Techniques

Machine learning is commonly used to automate disease classification by extracting features from segmented objects, but automated segmentation can be difficult. Model proposed uses average values to compute features based on the whole image, eliminating the need to perform additional steps. Results obtained by the algorithm were encouraging, but implementing it requires considerable computational expertise. Lung cancer detection is a very important field of medical processing. We have included two processes named feature extraction and feature selection processes and evaluated and compared the performance of feature extraction and feature extracting and selection processes. In addition, some benchmark sets can be used to compare the performance of the proposed work model. PCA, ICA, and SURF were used to develop the friendly disease prediction model. The feature selection process is applied after the feature extraction process to select the relative features and will compare with the feature extraction criteria. The feature selection is done using the HPSO-CSO Algorithm. The proposed method in which the work has been done using canny edge operator and feature selection has shown good result values in comparison to feature extraction. The proposed techniques aim to minimise classification errors and maximise accuracy. From the results, it has been concluded that accuracy was achieved at 97.94%.