Brain tumor segmentation using double density dual tree complex wavelet transform combined with convolutional neural network and genetic algorithm

<span>Image segmentation is often faced by low contrast, bad boundaries, and inhomogeneity that made it difficult to separate normal abnormal tissue. Therefore, takes long periodto read diagnose brain tumor patients. The aim of this study was applied hybrid methods optimize process magnetic resonance image brain. In study, we divide the images with double density dual-tree complex wavelet transform (DDDTCWT), continued convolutional neural network (CNN), optimized genetic algorithm (GA) 48 combinations yielding excellent results. F-1 score 99.42%, 913 test data. training consist 1397 MRI 302 imaging (MRI) resized 32 x32 pixels. DDDTCWT transforms input into more detail than ordinary transforms, CNNs will recognize pattern output images. Additionally, GA weights biases from first layer layers. parameters used for evaluating were dice similarity coefficient (DSC), positive present value (PPV), sensitivity, accuracy. result showed combination DDDTCWT, CNN, could be generated 95%.</span>