Hybdeefu-Feaseg: A Comprehensive Framework For Lung Cancer Detection Combining Deep Learning, Fuzzy Logic, And Optimized Feature Selection

Lung carcinoma is one of the greatest threatening and life-taking diseases in the globe, with the highest mortality
rate. Early diagnosis and treatment can save lives. Among all of the diseases being investigated, lung cancer requires more focus because it impacts both men and women and raises the rate of death. Although MRI is the greatest scanning technology in the healthcare industry, it is difficult for doctors to comprehend and identify cancer from MRI (Magnetic Resonance Imaging) scans. In order to precisely detect malignant cells, computer-aided diagnostics might be beneficial for clinicians. Several studies and applications of computer-aided methods employ image processing and machine learning. Early and accurate diagnosis of malignant tumors can significantly enhance both the effectiveness of treatment and the persistence rate of carcinoma patients. Nevertheless, the conventional strategies' limited sensitivity, high cost, and disruptive characteristics
constrain the applicability. Standard lung tumor forecasting techniques could not sustain precision because of low picture quality, interfering with the segmentation process. This research proposes a Hybrid Deep Learning and Fuzzy Logic Integration (HYBDEEFU-FEASEG) approach for Lung Cancer Detection. The projected model will include five major phases: The gathered raw MRI image was pre-processed via Gaussian filtering. The segNet method was proposed for segmentation. Haralick Texture Features (contrast, correlation, energy, and entropy), Local Phase Feature (LPF), and Shape feature (Hu moments, Zernike moments) techniques were used to extract texture features from segmented images. Artificial Gorilla Troops Optimization (AGTO) and Seagull Optimization Algorithm (SOA) approaches were developed for feature selection. The final detected outcomes (presence/absence of lung cancer) will be acquired from fuzzy logic. This method was used to classify and
detect lung cancer more accurately. Therefore, our research targets to increase the efficiency of the existing model.