Ca2+ transient detection and segmentation with the Astronomically motivated algorithm for Background Estimation And Transient Segmentation (Astro-BEATS)

Dataset For Quantitative Analysis of Miniature Synaptic Calcium Transients This dataset is suppliment to the testing benchmark dataset for miniature Synaptic Calcium transient detection described in https://doi.org/10.5281/zenodo.18561818. This dataset provides detection centers for later reverified transients missing from the 'manual-expert' annotation sets. Additionally, transient types as described in 'Ca2+ transient detection and segmentation with the Astronomically motivated algorithm for Background Estimation And Transient Segmentation (Astro-BEATS)' are manually labeled for each transient. Dataset Structure 'updated-detections-with-types' contrains revised ground truth manual annotations for testing videos in https://doi.org/10.5281/zenodo.18561818. Each .csv file contains the X-coordinate, Y-coordinate, and Slice (Time) position for every synaptic event, as identified by an expert annotator. A qualitatively defined transient class is also listed. Transient classes include 'den' = dendritic transient 'syn' = synaptic transient 'oof' = out of focus transient Transient classes are described in detail in 'Ca2+ transient detection and segmentation with the Astronomically motivated algorithm for Background Estimation And Transient Segmentation (Astro-BEATS)'. Additional events detected after the initial manual expert labels are also included. 're-verification' contains machine annotations made by each of the following algorithms; AQuA, ITD, Astro-BEATs, 3D-Unet Trained on Astro-BEATs, and 3D-UNet trained on ITD. Annotations are given in the centroids folder, while false possitive and false negative detections are assessed by each expert in the reverifications folder. When re-verifying: (status = 1) denotes a false detection that is confirmed by an expert to be a real detection, that was misclassified due to center missalignment (status = 0) denotes a genuine false detection Raw videos data avaliability This Zenodo is meant to suppliment data presented in the testing dataset presented in 'Quantitative Analysis of Miniature Synaptic Calcium Transients using Positive Unlabeled Deep Learning.' Training data, Validation data, complete testing data videos, and transient segmentations as described in 'Ca2+ transient detection and segmentation with the Astronomically motivated algorithm for Background Estimation And Transient Segmentation (Astro-BEATS)' are hosted via our laboratory's dedicated repository. Full Training Set & Code: https://github.com/FLClab/Calcium-Analysis Methodology: Please refer to associated manuscript for detailed description of transient classes and expert annotation. Technical Specifications File Formats: CSV (classification and manual reannotation), npy arrays (machine annotations) Domain: Neuroscience, Calcium Imaging, Transient microscopy Annotation Method: Manual expert curation Abstract/Paper Outline: Significance: Fluorescence-based Ca2+-imaging is a powerful tool for studying localized neuronal activity, including miniature Synaptic Calcium transients (mSCTs), which provide real-time insights into synaptic activity. However, these transients induce only subtle changes in the fluorescence signal, often barely above baseline, which poses a significant challenge for automated mSCT detection and segmentation. Aim: Detecting astronomical transients similarly requires efficient algorithms that will remain robust over a large field of view with varying noise properties. We leverage techniques used in astronomical transient detection for mSCT detection in fluorescence microscopy. Approach: We present Astro-BEATS, an automatic segmentation algorithm designed to detect mSCTs in Ca2+-imaging videos that incorporates image estimation and source-finding techniques used in astronomy. Astro-BEATS uses the Rolling Hough Transform (RHT) filament detector to construct an estimate of the expected (transient free) fluorescence signal of both the dendritic foreground and the background. Subtracting this baseline signal yields difference images displaying transient signals. We use Density-Based Spatial Clustering of Applications with Noise (DBSCAN) to find sources clustered in spatial and temporal space. Results: We compare the classification and segmentation of \astrobeats\ to other approaches. Astro-BEATS outperforms current threshold-based approaches and is competitive with methods where supervised deep learning (DL) algorithms are trained on manual detections and segmentations of mSCTs in Ca2+ -imaging data. The speed of Astro-BEATS and its applicability to previously unseen datasets without re-optimization makes it particularly useful for generating training datasets for DL-based detection of mSCTs. Conclusion Astro-BEATS greatly reduces the time needed for the annotation of mSCTs and removes the significant overhead of human expert annotation, enabling consistent analysis of new Ca2+ -imaging datasets.