Fig. S2
CaImAn batch outperforms the Suite2p algorithm in all datasets when benchmarked against the consensus annotation.(a) Contour plots of selected components against consensus annotation (CA) for (left) and Suite2p with the use of a classifier (middle) and direct comparison between the algorithms (right) for the test dataset N.00.00. identifies better components with a weak footprint in the summary correlation image. (b) Performance metrics F1 score (top), precision (middle) and recall (bottom), for Suite2p (with and the without the use of the classifier) and for the eight test datasets. consistently outperforms Suite2p, which can have significant variations between precision and recall. See Materials and methods (Comparison with Suite2p) for more details on the comparison.
|