Automated 3D Soma Segmentation with Morphological Surface Evolution for Neuron Reconstruction

3D Soma Detection in Large-Scale Whole Brain Images via a Two-Stage Neural Network

3D soma detection in whole brain images is a critical step for neuron reconstruction. However, existing soma detection methods are not suitable for whole mouse brain images with large amounts of data and complex structure. In this paper, we propose a two-stage deep neural network to achieve fast and accurate soma detection in large-scale and high-resolution whole mouse brain images (more than 1TB). For the first stage, a lightweight Multi-level Cross Classification Network (MCC-Net) is proposed to filter out images without somas and generate coarse candidate images by combining the advantages of the multi convolution layer's feature extraction ability. It can speed up the detection of somas and reduce the computational complexity. For the second stage, to further obtain the accurate locations of somas in the whole mouse brain images, the Scale Fusion Segmentation Network (SFS-Net) is developed to segment soma regions from candidate images. Specifically, the SFS-Net captures multi-scale context information and establishes a complementary relationship between encoder and decoder by combining the encoder-decoder structure and a 3D Scale-Aware Pyramid Fusion (SAPF) module for better segmentation performance. The experimental results on three whole mouse brain images verify that the proposed method can achieve excellent performance and provide the reconstruction of neurons with beneficial information. Additionally, we have established a public dataset named WBMSD, including 798 high-resolution and representative images ( 256 ×256 ×256 voxels) from three whole mouse brain images, dedicated to the research of soma detection, which will be released along with this paper.

Deep-Learning-Based Automated Neuron Reconstruction From 3D Microscopy Images Using Synthetic Training Images

Digital reconstruction of neuronal structures from 3D microscopy images is critical for the quantitative investigation of brain circuits and functions. It is a challenging task that would greatly benefit from automatic neuron reconstruction methods. In this paper, we propose a novel method called SPE-DNR that combines spherical-patches extraction (SPE) and deep-learning for neuron reconstruction (DNR). Based on 2D Convolutional Neural Networks (CNNs) and the intensity distribution features extracted by SPE, it determines the tracing directions and classifies voxels into foreground or background. This way, starting from a set of seed points, it automatically traces the neurite centerlines and determines when to stop tracing. To avoid errors caused by imperfect manual reconstructions, we develop an image synthesizing scheme to generate synthetic training images with exact reconstructions. This scheme simulates 3D microscopy imaging conditions as well as structural defects, such as gaps and abrupt radii changes, to improve the visual realism of the synthetic images. To demonstrate the applicability and generalizability of SPE-DNR, we test it on 67 real 3D neuron microscopy images from three datasets. The experimental results show that the proposed SPE-DNR method is robust and competitive compared with other state-of-the-art neuron reconstruction methods.

DeepRayburst for Automatic Shape Analysis of Tree-Like Structures in Biomedical Images

Precise quantification of tree-like structures from biomedical images, such as neuronal shape reconstruction and retinal blood vessel caliber estimation, is increasingly important in understanding normal function and pathologic processes in biology. Some handcrafted methods have been proposed for this purpose in recent years. However, they are designed only for a specific application. In this paper, we propose a shape analysis algorithm, DeepRayburst, that can be applied to many different applications based on a Multi-Feature Rayburst Sampling (MFRS) and a Dual Channel Temporal Convolutional Network (DC-TCN). Specifically, we first generate a Rayburst Sampling (RS) core containing a set of multidirectional rays. Then the MFRS is designed by extending each ray of the RS to multiple parallel rays which extract a set of feature sequences. A Gaussian kernel is then used to fuse these feature sequences and outputs one feature sequence. Furthermore, we design a DC-TCN to make the rays terminate on the surface of tree-like structures according to the fused feature sequence. Finally, by analyzing the distribution patterns of the terminated rays, the algorithm can serve multiple shape analysis applications of tree-like structures. Experiments on three different applications, including soma shape reconstruction, neuronal shape reconstruction, and vessel caliber estimation, confirm that the proposed method outperforms other state-of-the-art shape analysis methods, which demonstrate its flexibility and robustness.

Accurate Neuronal Soma Segmentation Using 3D Multi-Task Learning U-Shaped Fully Convolutional Neural Networks

Neuronal soma segmentation is a crucial step for the quantitative analysis of neuronal morphology. Automated neuronal soma segmentation methods have opened up the opportunity to improve the time-consuming manual labeling required during the neuronal soma morphology reconstruction for large-scale images. However, the presence of touching neuronal somata and variable soma shapes in images brings challenges for automated algorithms. This study proposes a neuronal soma segmentation method combining 3D U-shaped fully convolutional neural networks with multi-task learning. Compared to existing methods, this technique applies multi-task learning to predict the soma boundary to split touching somata, and adopts U-shaped architecture convolutional neural network which is effective for a limited dataset. The contour-aware multi-task learning framework is applied to the proposed method to predict the masks of neuronal somata and boundaries simultaneously. In addition, a spatial attention module is embedded into the multi-task model to improve neuronal soma segmentation results. The Nissl-stained dataset captured by the micro-optical sectioning tomography system is used to validate the proposed method. Following comparison to four existing segmentation models, the proposed method outperforms the others notably in both localization and segmentation. The novel method has potential for high-throughput neuronal soma segmentation in large-scale optical imaging data for neuron morphology quantitative analysis.

Cell numbers, distribution, shape, and regional variation throughout the murine hippocampal formation from the adult brain Allen Reference Atlas

Quantifying the distribution of cells in every brain region is fundamental to attaining a comprehensive census of distinct neuronal and glial types. Until recently, estimating neuron numbers involved time-consuming procedures that were practically limited to stereological sampling. Progress in open-source image recognition software, growth in computing power, and unprecedented neuroinformatics developments now offer the potentially paradigm-shifting alternative of comprehensive cell-by-cell analysis in an entire brain region. The Allen Brain Atlas provides free digital access to complete series of raw Nissl-stained histological section images along with regional delineations. Automated cell segmentation of these data enables reliable and reproducible high-throughput quantification of regional variations in cell count, density, size, and shape at whole-system scale. While this strategy is directly applicable to any regions of the mouse brain, we first deploy it here on the closed-loop circuit of the hippocampal formation: the medial and lateral entorhinal cortices; dentate gyrus (DG); areas Cornu Ammonis 3 (CA3), CA2, and CA1; and dorsal and ventral subiculum. Using two independent image processing pipelines and the adult mouse reference atlas, we report the first cellular-level soma segmentation in every sub-region and non-principal layer of the left hippocampal formation through the full rostral-caudal extent. It is important to note that our techniques excluded the layers with the largest number of cells, DG granular and CA pyramidal, due to dense packing. The numerical estimates for the remaining layers are corroborated by traditional stereological sampling on a data subset and well match sparse published reports.

PAT-Probabilistic Axon Tracking for Densely Labeled Neurons in Large 3-D Micrographs

A major goal of contemporary neuroscience research is to map the structural connectivity of mammalian brain using microscopy imaging data. In this context, the reconstruction of densely labeled axons from two-photon microscopy images is a challenging and important task. The visually overlapping, crossing, and often strongly distorted images of the axons allow many ambiguous interpretations to be made. We address the problem of tracking axons in densely labeled samples of neurons in large image data sets acquired from marmoset brains. Our high-resolution images were acquired using two-photon microscopy and they provided whole brain coverage, occupying terabytes of memory. Both the image distortions and the large data set size frequently make it impractical to apply present-day neuron tracing algorithms to such data due to the optimization of such algorithms to the precise tracing of either single or sparse sets of neurons. Thus, new tracking techniques are needed. We propose a probabilistic axon tracking algorithm (PAT). PAT tackles the tracking of axons in two steps: locally (L-PAT) and globally (G-PAT). L-PAT is a probabilistic tracking algorithm that can tackle distorted, cluttered images of densely labeled axons. L-PAT divides a large micrograph into smaller image stacks. It then processes each image stack independently before mapping the axons in each image to a sparse model of axon trajectories. G-PAT merges the sparse L-PAT models into a single global model of axon trajectories by minimizing a global objective function using a probabilistic optimization method. We demonstrate the superior performance of PAT over standard approaches on synthetic data. Furthermore, we successfully apply PAT to densely labeled axons in large images acquired from marmoset brains.

Cocaine-Induced Preference Conditioning: a Machine Vision Perspective

Existing work on drug-induced synaptic changes has shown that the expression of perineuronal nets (PNNs) at the cerebellar cortex can be regulated by cocaine-related memory. However, these studies on animals have mostly relied on limited manually-driven procedures, and lack some more rigorous statistical approaches and more automated techniques. In this work, established methods from computer vision and machine learning are considered to build stronger evidence of those previous findings. To that end, an image descriptor is designed to characterize PNNs images; unsupervised learning (clustering) is used to automatically find distinctive patterns of PNNs; and supervised learning (classification) is adopted for predicting the experiment group of the mice from their PNN images. Experts in neurobiology, who were not aware of the underlying computational procedures, were asked to describe the patterns emerging from the automatically found clusters, and their descriptions were found to align surprisingly well with the two types of PNN images revealed from previous studies, namely strong and weak PNNs. Furthermore, when the set of PNN images corresponding to every mice in the saline (control) group and the conditioned (experimental) group were characterized using a bag-of-words representation, and subject to supervised learning (saline vs conditioned mice), the high classification results suggest the ability of the proposed representation and procedures in recognizing these groups. Therefore, despite the limited size of the dataset (1,032 PNN images of 6 saline and 6 conditioned mice), the results support existing evidence on the drug-related brain plasticity, while providing higher objectivity.