Convolution network optimized with LO metaheuristic algorithm for automatic brain tumor classification

Undoubtedly, the brain, as the most sensitive organ of the body, controls the basic and important functions of the human body. A brain tumor is a serious cancer that is caused by the uncontrolled and abnormal division of cells. Because the incorrect classification of brain tumor can lead to bad consequences, the correct selection of tumor type and grade plays an important role in determining the appropriate treatment plan. For this reason, automatic brain tumor classification plays a vital and efficient role in accelerating the treatment process, planning and increasing the survival rate of patients. In order to address this issue, a new approach called convolution network optimized with meta-heuristic algorithm (LO CNN) has been developed. This approach involves preprocessing brain MRI images to reduce false tumor detection rates. Then, using line segments to preserve hidden edge details, a candidate region process is applied to identify the tumor region. Various features are extracted from the segmented region, which is classified using a convolutional neural network (CNN). The proposed LO CNN system is evaluated using pixel accuracy, error rate, precision, specificity and sensitivity criteria. This system achieves 99% accuracy on the Kaggle dataset.