Brain tumor diagnosis using a step-by-step methodology based on courtship learning-based water strider algorithm

•New pipeline methodology to automatic diagnosis of the cancer images from the MRI.•Feature selection and classification are established by a new improved metaheuristic.•The metaheuristic is designed based on courtship learning-based algorithm.•The method is fed on the “Brain-Tumor-Progression” database for validation. Medical imaging plays an essential function in the management of brain tumors for diagnosis and assortment. Today, MRI images are used to diagnose brain tumors because they show the structure of a normal brain in great detail. Brain tumor segmentation from MRI is a challenging procedure that has certain ups and downs. The tumor to be diagnosed has a flexible and complex structure in the image, it is completely different in size and location and from disease to patient, and accordingly, various algorithms were suggested. In this paper, an automated method is introduced to achieve higher speed and appropriate accuracy in the diagnosis of brain tumors. The present study proposed a new pipeline technique to automatic diagnosis of the brain cancer images from the MRI. Extraction of a feature was conducted to the input images followed by preprocessing to reduce the complexity of the system. The features are then entered into an optimal ANN to provide an efficient diagnosis system. Both feature selection and classification are established by an improved metaheuristic, named courtship learning-based water strider algorithm. The proposed method is then enforced on the “Brain-Tumor-Progression” database and the outcomes have been validated by comparing with some formerly published methods. Simulation consequences indicated the higher efficiency of the suggested method against the other analyzed procedures.