Zhou, P. et al. eLife 7, e28728 (2018).

Several approaches enable the extraction of single-neuron activity from calcium imaging data. However, this task is particularly difficult in microendoscopic data sets because of the high background and signal overlap. Zhou et al. have optimized a constrained matrix factorization (CNMF) approach for enhanced performance on these difficult data sets that models the background more realistically than has been done in previous implementations. The researchers found that their CNMF-E approach accurately extracted neuronal activity, with better performance than that of, for example, a PCA/ICA approach. The team demonstrated the CNMF-E approach on data sets acquired in the mouse dorsal striatum, the prefrontal cortex, the ventral hippocampus and the bed nucleus of the stria terminalis.