Automatic Cell Segmentation by Adaptive Thresholding (ACSAT) for Large-Scale Calcium Imaging Datasets.

Area of Science:

• Neuroscience
• Computational Biology
• Bioimaging

Background:

• Calcium imaging enables simultaneous recording from thousands of neurons.
• Processing large-scale imaging datasets presents a significant computational challenge.
• Manual neuron identification methods are not scalable for big data analysis.

Purpose of the Study:

• To develop an automated method for neuron identification in large calcium imaging datasets.
• To introduce Automated Cell Segmentation by Adaptive Thresholding (ACSAT) for scalable neurodata processing.

Main Methods:

• ACSAT utilizes a time-collapsed image and an iterative thresholding procedure.
• The algorithm adaptively calculates global and local thresholds based on pixel intensity distributions.
• It is designed to accommodate variations in neuron morphology and fluorescence intensity.

Main Results:

• ACSAT demonstrated high performance on simulated and real-world calcium imaging datasets.
• For simulated data with known ground truth, ACSAT achieved over 80% recall and precision at a signal-to-noise ratio of ~24 dB.
• The method proved effective across diverse datasets, including wide-field and two-photon imaging of hippocampus, striatum, and cell cultures.

Conclusions:

• ACSAT provides an effective and scalable solution for automated neuron segmentation in large calcium imaging datasets.
• The adaptive thresholding approach enhances robustness to variations in imaging conditions.
• This automated method facilitates efficient analysis of complex neural activity patterns.