Of all cancers, the world’s prominent cause of death is lung cancer as it is in the beginning asymptomatic and characteristically identified at advanced phases. If lung cancer is identified at a prior stage when it is minor and before it has spread, people have a better option of living longer. Though, identifying lung cancer at the precise time is a perplexing task due to high dimensional database space. The main objective of this paper is to diagnose lung cancer’s conditions like A Denocarcinomas (AD), SQuamous cell carcinomas (SQ), Carcinoids (COID), and Normal Lung (NL) for given query inputs. This paper proposes an optimal feature reduction and classification model for diagnosing lung cancers. For feature reduction, the various optimization techniques like Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Monarch Butterfly Optimization (MBO) and Modified Monarch Butterfly Optimization (MMBO) are applied. After feature reduction, for classifying the disease, Deep Neural Network (DNN) is utilized. Experimental results have shown that MMBO-DNN method outperforms MBO-DNN, GA-DNN and PSO-DNN in terms of accuracy, sensitivity, specificity and F1-score.

OFRCM; Cancer; COID; GA; PSO-DNN; DNN