Deep residual network based brain tumor segmentation and detection with MRI using improved invasive bat algorithm

A brain tumor is the mass of abnormal and unnecessary cells growing in the brain and it is also considered a life-threatening disease. Hence, segmentation and detection of such tumors at an early stage with Magnetic Resonance Image (MRI) is more significant to save the life. MRI is very effective to find persons with brain cancer such that the detection rate of this modality is moderately higher rather than considering other imaging modalities. Due to the size, shape, and appearance variations, the detection of brain tumors is a major complex task in the system of medical imaging. Hence, an efficient brain tumor detection technique is designed using the proposed Improved Invasive Bat (IIB)-based Deep Residual network model. Accordingly, the proposed IIB algorithm is derived by incorporating the Improved Invasive Weed Optimization (IWO) and Bat algorithm (BA), respectively. The segmentation of tumors with MR images has a great impact on detecting the brain tumor at the beginning stage. The deep learning-based method effectively generated better detection results with MR images. With segmentation results, features are acquired from the tumor regions that are further used to make the detection process with the Deep Residual network. However, the proposed method achieved higher performance in terms of the measures, such as accuracy, sensitivity, and specificity by computing the values of 0.9256, 0.9003, and 0.9146, respectively.