Pipeline Database Integration

Pipeline Steps (current)

Preprocess Steps

1. User enters in relevant metadata (slice thickness, resolution, stain)
2. CZI -> TIFF extraction
3. Rotate extracted tiffs in 90 degree increments such that rostral goes left, caudal goes right using GUI
4. Global Intensity Normalization
5. Section to section alignment script
6. User corrects erroneous alignments
7. User creates initial-masks
8. Masks generation script is run
9. User corrects erroneous masks
10. Local Adaptive Intensity Normalization
11. While-Slice Brain Crop
12. Brainstem Crop, requires user to specify bounding box

Pipeline Steps

Preprocess Steps

1. Slides are scanned and czi files are created and placed in czi dir.
2. Czi files are captured by pipeline and metadata info is placed in database.
3. Tiff file metadata is captured from czi file and placed in database. Actual width and height of each tif file is captured from the metadata and entered into database.
4. CZI file to TIF - done on EC2.
5. TIF dir is scanned and file size from dir is compared to width and height in slide_czi_to_tif table.
6. Set depth to 8 for all images.
7. Flip, rotate all images.
8. Linear normalize for counter stain channel, normalize other channels according to new algorithm.
9. Adaptive normalization.
10. Channels combined into one image for sections.
11. Scaled down images set available for viewing.

Alignment/Segmenting/Masking process

1. Done in existing GUI that has been upgraded to Python 3.
2. All necessary GUI files and supporting files imported in the new pipeline project.
3. All files cleaned up of any unnecessary code and ported to Python 3.
4. All files run through pylint

Post processing

1. Files renamed to proper numbering.
2. Precomputing step for Neuroglancer
3. Precomputed files placed in appropriate dir for access by web server.