All data sets used in the paper, along with the labels, can be downloaded from zenodo
This data set is obtaind from [1], which can be downloaded at https://www.zenodo.org/record/215600#.XK0LBJhKjBV.
We save the 20 images used in MATLAB file Bass.mat. Inside that file, the 3D images are saved in cell array img3DLst. The mean-projected 2D images are saved in cell array imgLst. The 2D annotations are saved in annoMapLst.
This data is obtained from [2], which can be downloaded at https://neurodata.io/data/collman15/.
We save Collman 14 and Collman 15 in Collman_clean.mat. The annotations in EM channel that does not correspond to any fluorescence staining are removed. The images are saved in imgLst. The first element of it is Collman 14 and the second is Collman 15. Each is a 4D array (Height x Width by Depth by channel). The first chanenl is pre-synaptic and the second is post-synaptic. The 3D annotations from EM channel are saved in annoMapLst. The first element is for Collman14 and the second is for Collman15. Each is a 3D array (Height x Width x Depth).
This data is obtained from experiments performed in [3]. We save the data in guilai_ctrl_1_strong_high.mat.
The 16 images are save in imgLst where each element is a 2D image. The annotations are saved in annoMapLst.
We generate synthetic data with three noise levels and three ranges of puncta sizes. There are 9 (3x3) experiment conditions in all. Each condition is saved in an individual .mat file in folder synthetic. The code to generate the data is prep_synthetic.m.
| Data | Min size (pixel) | Max size (pixel) | Average SNR |
|---|---|---|---|
| synthetic_smax_16_var1_0.0004.mat | 9 | 16 | 23.8 |
| synthetic_smax_16_var1_0.0025.mat | 9 | 16 | 17.2 |
| synthetic_smax_16_var1_0.01.mat | 9 | 16 | 11.5 |
| synthetic_smax_50_var1_0.0004.mat | 9 | 50 | 23.8 |
| synthetic_smax_50_var1_0.0025.mat | 9 | 50 | 17.2 |
| synthetic_smax_50_var1_0.01.mat | 9 | 50 | 11.5 |
| synthetic_smax_150_var1_0.0004.mat | 9 | 150 | 23.8 |
| synthetic_smax_150_var1_0.0025.mat | 9 | 150 | 17.2 |
| synthetic_smax_150_var1_0.01.mat | 9 | 150 | 11.5 |
In each .mat file, imgLst contains 10 images and annoMapLst contains the ground truth labels.
[1] Bass, Cher, et al. "Detection of axonal synapses in 3d two-photon images." PloS one 12.9 (2017): e0183309.
[2] Collman, Forrest, et al. "Mapping synapses by conjugate light-electron array tomography." Journal of Neuroscience 35.14 (2015): 5792-5807.
[3] Mizuno, Grace O., et al. "Aberrant calcium signaling in astrocytes inhibits neuronal excitability in a human Down syndrome stem cell model." Cell reports 24.2 (2018): 355-365.