Enhanced pre-processing for deep learning in MRI whole brain segmentation using orthogonal moments

•This paper proposes orthogonal moment filters as the initial filter to improve convolutional neural network output for brain segmentation in MRI.•The comparison of three different invariant moments method was performed.•The comparison of U-Net, U-Net++, and U-Net3+ was performed.•The use of an initial orthogonal moment filter for convolutional neural network in brain segmentation in MRI achieved an improvement over conventional method.•The proposed method improved the conventional approach.