Hybrid approach for lung cancer detection based on deep learning/machine learning

The incidence of Lung Cancer (LC) is rising in India. LC has been diagnosed and detected numerous times utilizing numerous data processing and identification strategies. Since the underlying origin of LC is still unknown, treatment is hopeless, making early diagnosis of lung tumors the only viable treatment option. So, a Machine Learning (ML) and Deep Learning (DL) based system is utilized to categorize CT scans for the existence of LC. The Visual Geometry Group (VGG-16) and Multi-Class Support Vector Machine (VGG-16+MSVM) technique is proposed in this research. Non-Local Means (NLM) Filter and Bi-Histogram Equalization (Bi-HE) are used, respectively; to filter out unwanted background noise in raw data samples and improve image quality. To isolate tumors in the raw data, the K-Means Clustering (KMC) technique is applied. The Gray Level Co-Occurrence Matrix (GLCM) is employed to generate features from the segmented data. The proposed approach is optimized with the use of a Genetic Algorithm (GA) that selects optimal feature subsets to maximize its performance. Combining ML and DL methods in Medical Image Processing is the most effective approach to detecting LC and its stages with the hope of achieving more precise findings. When accuracy is assessed and compared to other procedures, it becomes clear that the suggested methodology is more accurate (95%).