Deck 13 and 14: Squidpy Integration and Deep Learning on Image Tiles

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/5

encourage image

There's no tags or description

Looks like no tags are added yet.

Last updated 4:45 PM on 9/8/26
Name
Mastery
Learn
Test
Matching
Spaced
Call with Kai
Chat

No analytics yet

Send a link to your students to track their progress

6 Terms

1
New cards

Why does sq.gr.spatial_neighbors(sdata["table"]) work without any adaptation?

Because sdata["table"] is a real, unmodified AnnData object.

2
New cards

What functions convert between SpatialData and the legacy spatial AnnData format?

to_legacy_anndata() and from_legacy_anndata() (from spatialdata_io.experimental).

3
New cards

For a single coordinate system, how does .obsm['spatial'] normally get populated for legacy compatibility?

Automatically, by the spatialdata-io reader.

4
New cards

What real PyTorch-compatible class does SpatialData provide for tile-based datasets?

ImageTilesDataset (from spatialdata.dataloader.datasets).

5
New cards

Why is tile_dim_in_units specified in physical/spatial units rather than pixels?

So tile size stays consistent regardless of image resolution → in the tutorial, set to 3× the mean Xenium cell diameter.

6
New cards

What deep learning model and image source are combined in this tutorial?

A Monai DenseNet121 predicting Xenium-derived cell types from a Visium H&E image.