论文标题
DDU NET：3D MRI脑肿瘤分割的分布式密集模型
DDU-Nets: Distributed Dense Model for 3D MRI Brain Tumor Segmentation
论文作者
论文摘要
由于其特征较弱和形状可变形，脑肿瘤及其子区域的分割仍然是一项艰巨的任务。在本文中，提出了分布式密集连接（DDC）的三种模式（交叉SKIP，SKIP-1和SKIP-2），以通过在网络的钥匙层之间构造隧道来增强CNN的特征再利用和传播。为了更好地检测和分割从多模式3D MR图像中的脑肿瘤，嵌入DDCS（DDU-NET）的基于CNN的模型是从像素到像素的有效训练的，参数数量有限。然后，通过减少假阳性样本来进行后处理以完善分割结果。在Brats 2019数据集上评估了所提出的方法，结果证明了DDU-NET的有效性，同时需要更少的计算成本。
Segmentation of brain tumors and their subregions remains a challenging task due to their weak features and deformable shapes. In this paper, three patterns (cross-skip, skip-1 and skip-2) of distributed dense connections (DDCs) are proposed to enhance feature reuse and propagation of CNNs by constructing tunnels between key layers of the network. For better detecting and segmenting brain tumors from multi-modal 3D MR images, CNN-based models embedded with DDCs (DDU-Nets) are trained efficiently from pixel to pixel with a limited number of parameters. Postprocessing is then applied to refine the segmentation results by reducing the false-positive samples. The proposed method is evaluated on the BraTS 2019 dataset with results demonstrating the effectiveness of the DDU-Nets while requiring less computational cost.
