Three-dimensional Medical Image Segmentation with SE-VNet Neural Networks

In computer-aided clinical diagnosis, automatic segmentation of medical images collected from computed tomography and magnetic resonance imaging is a key and challenging task. To improve the accuracy of medical image segmentation, this paper proposes a three-dimensional medical image segmentation neural network based on a channel attention mechanism-SE-VNet. This method can realize end-to-end automatic segmentation of three-dimensional medical images and has a high accuracy rate. The experimental results on the prostate MRI data set show that the accuracy of the Dice coefficient reaches 88.7% and the Hausdorff distance reaches 5.491 mm; the experimental results on the femoral CT data set reach 93.2% and 4.175 mm. Compared with the current mainstream three-dimensional medical image segmentation methods, the method proposed in this paper greatly improves the segmentation accuracy.