An automated brain tumor segmentation framework using a novel fruit fly UNet

Brain image analysis and segmentation are the most difficult tasks in medical image processing because of image complexity. Moreover, MRI images are mostly utilized to predict different brain-based diseases; if the images are complex, the disease prediction accuracy is very low. To overcome this problem, the current research has planned to design a novel fruit fly-based UNet (FFbU) framework to detect the Tumor accurately. Moreover, the fitness of the fruit fly was upgraded in the UNet pooling module that has tended to gain the finest results. Initially, the standard datasets were gathered from the net source and trained to the system. Consequently, the training error is removed in the primary layer of FFbU then the error-cleared data is entered into the UNet dense layer for tumor detection and segmentation. Finally, the proposed model is executed in a MATLAB environment, and the proficiency of the designed FFbU model is estimated in terms of accuracy, recall, precision, Dice, and Jaccard. In addition, the planned novel FFbU model has the ability to predict and segment different tumor types.