Oppositional based Seagull Optimization Algorithm based 3D DCNN for Medical Image Semantic Segmentation

In this paper, oppositional based Seagull optimization Algorithm Based 3D Deep Convolution Neural Network (OSOA-3DCNN) is developed to semantic segmentation of the medical images. Initially, the lung and brain tumor images can be gathered from the online sources. The projected semantic segmentation process is a combination of 3D Deep Convolution Neural Network (3DCNN) and oppositional based Seagull optimization Algorithm (OSOA). In the 3DCNN, to improve the accuracy and efficiency, the hyperparameters is chosen based on OSOA. The proposed method contains the two major advantages such as: Initially, it reduces the necessity for consuming polling operation in the network. After that, it empowers the resolution by managing the number of training features. Furthermore, memory consumption of the network is reduced. The projected technique is reduced computation cost and memory consumption by managing a preferable validation. The proposed method is utilized to segment the tumor portions of lung as well as brain. The projected technique can be executed in MATLAB in addition performances are validated through performance metrics of similarity measurements as well as quality measurements. To justify the presentation of the projected technique, it can be contrasted by traditional techniques which are Fuzzy C-means Clustering, Convolutional Neural Network (CNN), Particle Swarm Algorithm based Convolutional Neural Network (PS O-CNN) and Deep Convolutional Neural Network-Jellyfish Algorithm (DCNN-JA) respectively.