The invention discloses a self-adaptive threshold segmentation method and device for an
astrocyte dynamic
calcium event, and relates to the technical field of
cell calcium signal recognition, and the method comprises the steps: carrying out the
calcium signal enhancement and denoising of an original video, constructing a
time sequence mask to dynamically track a calcium active event, and carrying out the recognition of the
time sequence mask; the method comprises the following steps: fitting gray scale distribution into a plurality of
Gaussian component superposition based on a
Gaussian mixture model, carrying out segmented
estimation to extract an active calcium
signal, calculating a highlight pixel proportion and an inter-frame similarity index, adaptively and dynamically updating a global threshold value and a local threshold value, and generating a calcium signal image through
window function transformation. The method can effectively inhibit random imaging
noise and retain weak signals in an
astrocyte calcium imaging video with high
noise and low signal-to-
noise ratio, significantly improves the transient calcium event detection capability of astrocytes, adaptively identifies the brightest active calcium signal peak, dynamically adapts to the brightness distribution change of each frame of image, and improves the detection accuracy. The dynamic calcium event can be stably and accurately segmented, the robustness is strong, and the
automation degree is high.