EHMO-Based Deep ResNet for Survival Timeline Prediction of Adenocarcinoma Cancer

Lung cancer is due to the growth of uncontrolled cells in the lungs, and the death rate is high compared with all types of cancer. It is recognized and treated using images of computed tomography (CT). This paper develops the elephant herding magnetic optimization-based deep residual network (EHMO-based Deep ResNet) for survival timeline prediction in adenocarcinoma. Here, preprocessing is performed using a Gaussian filter for the lung CT image. The preprocessed image is subjected to lung lobe segmentation, which is performed by the active contour model. Nodule identification locates nodules in the segmented image, where the process is carried out using a grid-based scheme. After that, feature extraction is carried out to extract intensity, wavelet, tetrolet transform, local optimal oriented pattern (LOOP), and clinical features. Finally, the extracted features are fed to the prediction module, which is based on the Deep ResNet classifier, which is trained by the proposed EHMO optimization algorithm. Here, the developed EHMO combines elephant herding optimization (EHO) and the magnetic optimization algorithm (MOA). The developed adenocarcinoma survival timeline prediction technique exhibits efficient performance in terms of accuracy, 0.955; maximal sensitivity, 0.962; and high specificity, 0.958.