Deep Networks Reconstruction of 3D Neurons in a Tangled Neuronal Image PROJECT TITLE : 3D Neuron Reconstruction in Tangled Neuronal Image With Deep Networks ABSTRACT: Understanding how the brain works requires tracing or digital reconstruction of 3D neuron models. For the clean neuronal image with a single neuron, existing automatic tracing algorithms perform well. However, they are not robust for tracing the neuron surrounded by nerve fibres. Segmenting the neuron from the surrounding fibres before using trace techniques is the goal of our 3D U.Net Plus network. Clean neurons with no nerve fibre interference are not suitable for training the segmentation network in BigNeuron, the largest available neuronal picture dataset. SyntaNEI is a dataset based on BigNeuron pictures in which neurons are fused with extracted nerve fibres to train the proposed network. Dropout, štrous convolution, and štrous Spatial Pyramid Pooling (ASPP) were all used in the proposed 3D U.Net Plus network to successfully segment synthetic and real tangled neuronal pictures. Tracing algorithms using segmentation results produce neurons that are substantially closer to the ground truth than those rebuilt from original images. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest 3D APA-Net 3D Adversarial Pyramid Anisotropic Convolutional Network for Prostate Segmentation in MR Images When Group Sparsity Meets Rank Minimization, a Sparse Coding Benchmark