A multi-rater benchmark for perineuronal nets detection and counting in fluorescence microscopy images

Dataset of fluorescence microscopy images of mice brain slices stained against perineuronal nets (PNNs). The dataset is composed of two subsets: a large single-rater subset (PNN-SR) and a smaller multi-rater subset (PNN-MR). i) PNN-SR consists of 25 images having different sizes ranging from 8184×6163 to 15120×9477 pixels. Among all the images, there are roughly 34k annotated PNNs, varying from a few dozens to some thousand per image, dot-annotated by a single human rater. ii) PNN-MR comprises 12 microscopic images of 2000×2000 pixels representing different portions of a mouse brain, with a total of 2,532 dot-annotated PNNs. The annotation procedure has been performed by seven different raters.