The HNCcorr algorithm developed for cell identification in calcium-imaging datasets is a top performer for the Neurofinder benchmark. The Neuro finder community benchmark is an initiative of the CodeNeuro collective of neuroscientists that encourages software tool development for neuroscience research.
The Neurofi nder benchmark consists of 28 calcium imaging movies, 19 for training algorithms and the remaining nine for unbiased evaluation of algorithmic performance. The datasets have been contributed by various labs and have been recorded under different experimental conditions. They differ in sample frequency, length of the movie, magni cation, signal-to-noise ratio, and region of the brain that was recorded.
Update (January 20th, 2018)
HNCcorr recently lost its first place after having lead the benchmark for most of 2017.