1_3D_cell_nuclear_morphology:_microscopy_imaging_dataset

SOCR 3D Cell Morphometry Project

1. 3D cell nuclear morphology: microscopy imaging dataset

Source code, models, and documentation for 3D image pre-processing, per-channel segmentation, and curation

Inputs: 3D TIFF images (volumes) in each of 3 channels (available for downloading on the project data webpage).

Nuclei 3D pre-processing, segmentation, and curation (DAPI, c0):

1. Convert each volume to 8-bit greyscale and apply despeckling using ImageJ
2. Segment each volume in 3D using Farsight toolkit's Nuclear Segmentation algorithm
3. Fill holes in derived nuclear masks using MATLAB
4. Separate nuclear masks into individual volumes (implemented using LibTIFF)
5. Filter out nuclear masks that do not pass min and/or max thresholds for voxel count (mask volume) and compactness

Nucleoli 3D pre-processing, segmentation, and curation (both fibrillarin, c1, and EtBr, c2):

1. Mask out background in each volume, convert it to 16-bit greyscale and apply despeckling using ImageJ
2. Segment each volume in 3D using Trainable Weka Segmentation (Fiji)
3. Separate binary blobs by the watershed algorithm and find connected components in ImageJ
4. Separate nuclear masks into individual volumes (implemented using LibTIFF) – same as for nuclei
5. Run co-localization method to confirm c2 masks with c1 masks using ImageJ
6. Filter out c2 masks that do not pass min and/or max thresholds for voxel count (mask volume) and compactness– same as for nuclei, with own config

Outputs: individual c0 and c2 binary masks in TIFF format (also available for downloading on the project data webpage).