AbstractWe present an algorithm for detecting the location of cells from two-photon calcium imaging data. In our framework, multiple coupled active contours evolve, guided by a model-based cost function, to identify cell boundaries. An active contour seeks to partition a local region into two subregions, a cell interior and ex-terior, in which all pixels have maximally ‘similar’ time courses. This simple, local model allows contours to be evolved predominantly independently. When contours are sufficiently close, their evolution is coupled, in a manner that permits overlap. We illustrate the ability of the proposed method to demix overlapping cells on real data. The proposed framework is flexible, incorporating no prior information regarding a cell’s morphology or stereotypical temporal activity, which enables the detection of cells with diverse properties. We demonstrate algorithm performance on a challenging mouse in vitro dataset, containing synchronously spiking cells, and a manually labelled mouse in vivo dataset, on which ABLE achieves a 67.5% success rate.Significance statementTwo-photon calcium imaging enables the study of brain activity during learning and behaviour at single-cell resolution. To decode neuronal spiking activity from the data, algorithms are first required to detect the location of cells in the video. It is still common for scientists to perform this task manually, as the heterogeneity in cell shape and frequency of cellular overlap impede automatic segmentation algorithms. We developed a versatile algorithm based on a popular image segmentation approach (the Level Set Method) and demonstrated its capability to overcome these challenges. We include no assumptions on cell shape or stereotypical temporal activity. This lends our framework the flexibility to be applied to new datasets with minimal adjustment.
Efficient implementation of convolutional neural networks in the data processing of two-photon in vivo imaging
