Skill Tracer
Skills › Scientific Agent Skills › histolab › core_capabilities.md

core_capabilities.md

25 lines · skills/histolab/references/core_capabilities.md at 5f2bc03

Open on GitHub
1# Histolab core API map
2
3The skill targets published Histolab 0.7.0. Read the per-topic reference for
4executable examples and scientific limitations; placeholder WSI paths require
5user data. Upstream [API index](https://histolab.readthedocs.io/en/latest/py-modindex.html)
6currently labels itself 0.6.0, so use the linked 0.7.0 source for release-specific
7behavior.
8
9| Capability | Public API | Key contract |
10| --- | --- | --- |
11| Slide inspection | `histolab.slide.Slide` | `levels` is a list; `level_dimensions(level)` is a method |
12| Tissue detection | `histolab.masks.TissueMask`, `BiggestTissueBoxMask`, `BinaryMask` | Boolean mask; custom filters are positional |
13| Tiling | `histolab.tiler.RandomTiler`, `GridTiler`, `ScoreTiler` | Mask passed to methods; preview returns Pillow image |
14| Ranking | `histolab.scorer.NucleiScorer`, `CellularityScorer` | Stain-derived area heuristics, not cell counts or diagnosis |
15| Filters | `histolab.filters.image_filters.Compose` and filter classes | Preserve Pillow/NumPy type transitions |
16| Normalization | `histolab.stain_normalizer.MacenkoStainNormalizer`, `ReinhardStainNormalizer` | `fit(target)` then `transform(source)` |
17| Spatial provenance | `histolab.types.CoordinatePair`, `Tile.coords` | Four level-0 pixel coordinates |
19The slide-management reference covers native OpenSlide, exact MPP backend
20requirements and small versus remote sample data. The tissue-mask reference
21provides ROI conversion and pen-mask examples. The tile-extraction reference
22covers strategy choice, CSV schema and concentric multi-resolution patches.
23The filters reference covers type-correct compositions and stain normalization.
24The visualization reference provides aligned mask overlays, mosaics and score
25plots. The typical-workflows reference combines these APIs into extraction runs.

Histolab core API map

The skill targets published Histolab 0.7.0. Read the per-topic reference for executable examples and scientific limitations; placeholder WSI paths require user data. Upstream API index currently labels itself 0.6.0, so use the linked 0.7.0 source for release-specific behavior.

CapabilityPublic APIKey contract
Slide inspectionhistolab.slide.Slidelevels is a list; level_dimensions(level) is a method
Tissue detectionhistolab.masks.TissueMask, BiggestTissueBoxMask, BinaryMaskBoolean mask; custom filters are positional
Tilinghistolab.tiler.RandomTiler, GridTiler, ScoreTilerMask passed to methods; preview returns Pillow image
Rankinghistolab.scorer.NucleiScorer, CellularityScorerStain-derived area heuristics, not cell counts or diagnosis
Filtershistolab.filters.image_filters.Compose and filter classesPreserve Pillow/NumPy type transitions
Normalizationhistolab.stain_normalizer.MacenkoStainNormalizer, ReinhardStainNormalizerfit(target) then transform(source)
Spatial provenancehistolab.types.CoordinatePair, Tile.coordsFour level-0 pixel coordinates

The slide-management reference covers native OpenSlide, exact MPP backend requirements and small versus remote sample data. The tissue-mask reference provides ROI conversion and pen-mask examples. The tile-extraction reference covers strategy choice, CSV schema and concentric multi-resolution patches. The filters reference covers type-correct compositions and stain normalization. The visualization reference provides aligned mask overlays, mosaics and score plots. The typical-workflows reference combines these APIs into extraction runs.

Raw lines 1 to 25