Neuronal Image Reconstruction

Series on Language Processing, Pattern Recognition, and Intelligent SystemsFrontiers in Bioimage Informatics Methodology, pp. 89-125 (2024) No AccessChapter 3: Neuronal Image ReconstructionYufeng Liu, Kaifeng Chen, and Hanchuan PengYufeng LiuSEU-ALLEN Joint Center, Institute for Brain and Intelligence, Southeast University, Nanjing, Jiangsu, China, Kaifeng ChenSEU-ALLEN Joint Center, Institute for Brain and Intelligence, Southeast University, Nanjing, Jiangsu, China, and Hanchuan PengSEU-ALLEN Joint Center, Institute for Brain and Intelligence, Southeast University, Nanjing, Jiangsu, Chinahttps://doi.org/10.1142/9789811286131_0003Cited by:0 (Source: Crossref) PreviousNext AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack CitationsRecommend to Library ShareShare onFacebookTwitterLinked InRedditEmail Abstract: Neurons are the fundamental units responsible for the structure and function of the brain, making their morphologies crucial for characterizing the brain. In this chapter, we provide an introduction to neuron morphology and tracing algorithms for light microscopic images. We begin with a comprehensive overview of the structure and function of neurons and the acquisition and analysis of neuronal images. We then discuss manual, semi-automatic, and automatic approaches to neuron tracing, with a particular focus on recent advancements in automatic tracing using graph-based and deep-learning techniques. Pre-processing techniques, such as standardization, denoising, illumination correction, and signal enhancement, are commonly applied to improve the quality of microscopy images before tracing. Automatic tracing algorithms are commonly classified into local, global, and meta methods, which are categorized according to their utilization of context and whether they are built upon other methods. Key algorithms, including Dijkstra's algorithm and the Minimum Spanning Tree (MST) algorithm, are highlighted for their relevance in automatic tracing. Additionally, we mention various platforms and tools developed to facilitate data production and algorithm verification. The chapter summarizes the major applications of deep learning-enhanced tracing, such as critical point detection and segmentation. We also discuss the availability of publicly available datasets and evaluation metrics for benchmarking. Finally, we acknowledge the progress made in automatic neuron tracing while highlighting the remaining challenges, such as addressing low signal-to-noise ratios in neuronal images and effectively handling the complexity of ultra-scale images. Overall, this chapter provides a comprehensive introduction to neuron tracing from light microscopic images. FiguresReferencesRelatedDetails Recommended Frontiers in Bioimage Informatics MethodologyMetrics History PDF download