Lightsheet microscopy whole brain 3D nuclei instance segmentation

 

2D Unet-based

Motivations:

• None has developed a nuclei instance segmentation for whole brain.
• Lightsheet microscopy 3D image has anisotropic resolution.
• Resolution is isotropic in X-Y plane, where human annotating nuclei, then tracking Z stack.

Performance and time cost:

• Recall > 90%
• Precision > 90%
• Time cost ~= 12hr/brain

Data:

Multiple whole brains of mouse in different grown stage. Each brain has ~1500x9000x9000 voxels, and about 30,000,000 to 50,000,000 cells.

Example to test a whole-brain:

1 Obtain NIS results of one brain

P_tag=Name the test (e.g., P4)
pair_tag=Name the brain group
brain_tag=Name the brain
dataroot=/path/to/directory/2D_slice_image
saveroot=/path/to/directory/saving/results/${P_tag}/${pair_tag}/${brain_tag}/
mkdir -p ${saveroot}
nohup cpp/build/test ${pair_tag} ${brain_tag} ${device} ${dataroot} ${saveroot} > cpp_logs/${brain_tag}_${P_tag}.log

2 Statistic NIS results and collect all NIS center, coordinates, intensity

Change paths in run_whole_brain/statistic_cpp.py

device='cuda:X'
seg_root = 'XXX'
save_root = 'XXX'
P_tag = 'XXX'
brain_tag = 'XXX'
pair_tag = 'XXX'
data_root = 'XXX'

Adjust img_tags to get intensity of different image channels.

Then python run_whole_brain/statistic_cpp.py. All NIS will be saved to save_root.

2.5 (Optional) Cell type labeling using intensity of different channels

See nis_coloc.py

3 Visualize downsampled whole-brain NIS result

Similarly, change paths in brain_render.py

downsample_res = X.X
seg_res = X.X
seg_root = 'XXX'
stat_root = 'XXX'
save_root = 'XXX'

Then python brain_render.py

4 (Optional) Statistic downsampled NIS result in region-level with registered atlas

See stats/statistic_nii.py, stats/statistic_csv.py.

TODO: Interactive visualization of whole brain nuclei segmentation results

• Github page.

Codes availability

• Train 2D Unet
• Source code of the executible to test a whole brain

Data availability

• Train-val data
• Test whole brain

Usage

• Follow README under directories.