Automatic segmentation of brain MRI using a novel patch-wise U-net deep architecture

Abstract

Accurate segmentation of brain magnetic resonance imaging (MRI) is an essential step in quantifying the changes in brain structure. Deep learning in recent years has been extensively used for brain image segmentation with highly promising performance. In particular, the U-net architecture has been widely used for segmentation in various biomedical related fields. In this paper, we propose a patch-wise U-net architecture for the automatic segmentation of brain structures in structural MRI. In the proposed brain segmentation method, the non-overlapping patch-wise U-net is used to overcome the drawbacks of conventional U-net with more retention of local information. In our proposed method, the slices from an MRI scan are divided into non-overlapping patches that are fed into the U-net model along with their corresponding patches of ground truth so as to train the network. The experimental results show that the proposed patch-wise U-net model achieves a Dice similarity coefficient (DSC) score of 0.93 in average and outperforms the conventional U-net and the SegNetbased methods by 3% and 10%, respectively, for on Open Access Series of Imaging Studies (OASIS) and Internet Brain Segmentation Repository (IBSR) dataset.

Convolutional Neural Networks and Geometric Moments to Identify the Bilateral Symmetric Midplane in Facial Skeletons from CT Scans

“…The framework chosen in this paper for biomedical image segmentation was the U-Net [ 11 ]. U-Net has been used in a number of biomedical image segmentation applications such as kidney segmentation [ 12 ], prostate and prostate zones segmentation [ 13 ], brain tumor segmentation [ 14 ], brain segmentation [ 15 ], and so forth. Its name emerged from the idea of a U-shape architecture where in the first step, downsampling path, the spatial information is reduced while feature information is increased.…”

Brain Image Segmentation in Recent Years: A Narrative Review

“…Deep learning using 2D images requires brain image slices or extracted 2D patches from 3D images as an input for the 2D convolutional kernel. Several studies [21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] have been published on the deep learning-based method using 2D images. Sergio Pereira et al [21] introduced cascade layers using small 3*3 convolutions kernels to reduce overfitting.…”

“…Recently, an innovative brain tissue segmentation method from MR images was proposed by Lee et al [33], where a patch-wise U-net architecture was used to divide the MR image slices into non-overlapping patches. Corresponding patches of ground truth were incorporated into the U-net model, and input patches were predicted individually.…”

Segmentation of brain magnetic resonance images using a novel fuzzy clustering based method

“…However, their acceptance in the medical community typically depends on the ease of use and robustness in the segmentation. These methods can be broadly classified into four types, namely threshold based methods [17,18], active contour based methods [7,8,19,20], classification based methods [21,22], and clustering based methods [5,14,15,[23][24][25][26][27]. Threshold based segmentation methods [17,18] partition the image into different segments based on few threshold values.…”