Data Directory Structure

For batch processing (i.e., whole brain / multiple channels processing)

YW001_reg/
    ├── raw/ (optional) -- (1)
    │
    ├── zproj/ -- (2)
    │    └── YW001_g*_s*_{channel}.tif
    │
    ├── roi/ -- (3)
    │    └── YW001_g*_s*_{channel}.roi
    │
    ├── roi_cpose/ -- (3')
    │    └── YW001_g*_s*_{channel}.roi
    │
    ├── resize/ (src for the allenccf)
    │    ├── YW001_g*_s*_resize.tif -- (4)
    │    │
    │    └── processed/
    │           ├── YW001_g*_s*_resize_processed.tif -- (5)
    │           └── transformations/
    │                 ├── YW001_g*_s*_resize_processed_transformed.tif -- (6)
    │                 │
    │                 ├── YW001_g*_s*_resize_processed_transform_data.mat -- (7)
    │                 │
    │                 └── labelled_regions/
    │                       ├── {*channel}_roitable.csv -- (8)
    │                       └── parsed_data/
    │                             └── parsed_csv_merge.csv -- (9)
    │
    ├── resize_overlap/* (optional) -- (10)
    └── output_files/ (for generated figures or analysis outputs)

Placeholders

• g* : Index of glass slide.

• s* : Index of Slice .

• {channel} : Channel name (e.g., r, g, b, o(overlap)).

Reference (1)–(10)

1. raw/ — Raw confocal data (e.g., .lsm, .czi, or .tiff). (optional)

2. zproj/ — Z-projection stacks. Must be RGB and saved per channel (r, g, b, o).

3. roi/ — ROI files after ImageJ selection (.roi). 3’. roi_cpose/ — ROI files from the Cellpose pipeline (dev). 4. resize/ → *_resize.tif — Merged ROI, scaled RGB image for registration. Typically use blue (DAPI) as reference. 5. processed/ → *_resize_processed.tif — Contrast-adjusted and/or rotated image. 6. transformations/ → *_transformed.tif — Image after applying the transformation. 7. transformations/ → *_transform_data.mat — Transformation matrix/data for each slice. 8. labelled_regions/ → {*channel}_roitable.csv — AllenCCF output (per ROI). 9. labelled_regions/parsed_data/ → parsed_csv_merge.csv — Parsed & classified CSV for visualization. 10. resize_overlap/ — If overlap channels exist, mirror the same structure as r/g channels. Use the same transformation matrix. (optional)

Note

• For (4): If channel count is limited, save overlap channels to a separate pseudo-color file, then apply the same transformation matrix during registration.

• Ensure the DAPI (blue) channel is used as the registration reference unless otherwise specified.

• Maintain consistent filename patterns to keep downstream parsing robust.