Eggl Maximilian F, Wagle Surbhit, Filling Jean P, Chater Thomas E, Goda Yukiko, Tchumatchenko Tatjana
Institute of Experimental Epileptology and Cognition Research, University of Bonn Medical Centre, Bonn 53127, Germany.
Institute of Neuroscience, CSIC-UMH, Alicante 03550, Spain.
Bioinformatics. 2025 Jul 1;41(7). doi: 10.1093/bioinformatics/btaf339.
Investigating the molecular composition of different neural compartments such as axons, dendrites, or synapses is critical for understanding learning and memory. State-of-the-art microscopy techniques now resolve individual molecules and pinpoint their position with a micrometer or nanometre resolution across hundreds of micrometres, allowing the labelling of multiple structures of interest simultaneously. Algorithmically, tracking individual molecules across hundreds of micrometres and determining whether they are inside a particular cellular compartment can be challenging. Historically, microscopy images are annotated manually, often using multiple software packages to detect fluorescence puncta and quantify cellular compartments of interest. Advanced ANN-based automated tools, while powerful, often can only help with selected parts of the data analysis, may be optimized for specific spatial resolutions, cell preparations, and may not be fully open source and open access to be sufficiently customizable.
Thus, we developed SpyDen, a Python package based upon three principles: (i) ease of use for multi-task scenarios, (ii) open-source accessibility and data export to a standard, open data format, (iii) the ability to edit any software-generated annotation and generalize across spatial resolutions. SpyDen operates on 2D microscopy time-series data, offering robust temporal tracking and spatial analysis capabilities. Equipped with a graphical user interface and accompanied by video tutorials, SpyDen provides a collection of powerful algorithms that can be used for neurite and synapse detection, fluorescent puncta, and intensity analysis. We validated SpyDen using expert annotation across numerous use cases to prove a powerful, integrated platform for efficient and reproducible molecular imaging analysis.
SpyDen is available on https://github.com/meggl23/SpyDen while the compiled executables can be found at https://gin.g-node.org/CompNeuroNetworks/SpyDenTrainedNetwork.
研究不同神经部分（如轴突、树突或突触）的分子组成对于理解学习和记忆至关重要。如今，先进的显微镜技术能够分辨单个分子，并以微米或纳米分辨率在数百微米范围内精确确定其位置，从而实现对多个感兴趣结构的同时标记。从算法角度来看，在数百微米范围内追踪单个分子并确定它们是否位于特定细胞区域内可能具有挑战性。历史上，显微镜图像是手动标注的，通常使用多个软件包来检测荧光点并量化感兴趣的细胞区域。基于先进人工神经网络的自动化工具虽然功能强大，但往往只能帮助进行数据分析的某些部分，可能针对特定的空间分辨率、细胞制备进行了优化，而且可能并非完全开源且开放获取，无法充分定制。
因此，我们开发了SpyDen，这是一个基于三个原则的Python软件包：（i）易于用于多任务场景；（ii）开源可访问性以及数据导出为标准的开放数据格式；（iii）能够编辑任何软件生成的注释并在不同空间分辨率下通用。SpyDen可处理二维显微镜时间序列数据，具有强大的时间追踪和空间分析能力。它配备了图形用户界面并附有视频教程，提供了一系列强大的算法，可用于神经突和突触检测、荧光点以及强度分析。我们通过在众多用例中使用专家注释对SpyDen进行了验证，以证明它是一个用于高效且可重复的分子成像分析的强大集成平台。
Zotero 插件