Lung cancer detection via deep learning-based pyramid network with honey badger algorithm
A. Angel mary, K. K. Thanammal
Abstract
Lung carcinoma is a fatal illness characterized by the unchecked growth of tumors in the respiratory tract. A diagnosis of cancers and their dimensions and forms are often difficult to ascertain from imaging tests because of their irregular form. And high discrepancy in health images is a severe problem in health image investigation. To overcome this issue a novel deep learning-based lung cancer detection (DL-LCD) model is proposed. The LIDC-IDRI and NIH Chest X-ray dataset was used which comprises CT scan and CXR images. Initially, these images are pre-processed using CLAHE (Contrast Limited Adaptive Histogram Equalization) by eliminating noise distortions and enhancing the image clarity. The lung images are segmented using Honey Badger Algorithm. The segmented images are classified into cancerous and non-cancerous. Based on the LIDC-IDRI database, achieves 98.5 % accuracy in normal lungs and 97.44 % accuracy in abnormal lungs. The results of the experiments indicate that the proposed approach outperforms existing approaches.