Skull stripping in brain magnetic resonance imaging (MRI) is an essential step to analyze\nimages of the brain. Although manual segmentation has the highest accuracy, it is a time-consuming\ntask. Therefore, various automatic segmentation algorithms of the brain in MRI have been devised\nand proposed previously. However, there is still no method that solves the entire brain extraction\nproblem satisfactorily for diverse datasets in a generic and robust way. To address these shortcomings\nof existing methods, we propose the use of a 3D-UNet for skull stripping in brain MRI. The 3D-UNet\nwas recently proposed and has been widely used for volumetric segmentation in medical images due\nto its outstanding performance. It is an extended version of the previously proposed 2D-UNet, which\nis based on a deep learning network, specifically, the convolutional neural network. We evaluated\n3D-UNet skull-stripping using a publicly available brain MRI dataset and compared the results with\nthree existing methods (BSE, ROBEX, and Kleesiekâ??s method; BSE and ROBEX are two conventional\nmethods, and Kleesiekâ??s method is based on deep learning). The 3D-UNet outperforms two typical\nmethods and shows comparable results with the specific deep learning-based algorithm, exhibiting a\nmean Dice coefficient of 0.9903, a sensitivity of 0.9853, and a specificity of 0.9953.
Loading....