> [!NOTE]
> **Note:** This older single-modality model still holds a slight edge in baseline digital pathology performance. However, for a modern multimodal approach, consider [MedSigLIP](https://developers.google.com/health-ai-developer-foundations/medsiglip) instead.

Path Foundation is a machine learning (ML) model that produces embeddings based
on digital pathology images. The embeddings can be used to efficiently build AI
models for pathology analysis-related tasks, requiring less data and less
compute than having to fully train a model without the embeddings or the
pretrained model.

Trained on large scale datasets, Path Foundation helps businesses and
institutions in healthcare and life sciences do more with less pathology data,
accelerating their ability to build AI models for pathology image analysis.

For details about how to use the model and how it was trained, see the
[Path Foundation model card](https://developers.google.com/health-ai-developer-foundations/path-foundation/model-card).

## Common Use Cases

The following sections present some common use cases for the model. You're free
to pursue any use case, as long as it adheres to the
[Health AI Developer Foundations terms of use](https://developers.google.com/health-ai-developer-foundations/terms).

### Data-efficient classification

Path Foundation can be used for data-efficient classification tasks, including:

- Identifying tumor tissue and other distinct tissue classes
- Grading tumors
- Identifying the presence of known biomarkers to predict treatment response
- Exploration of pathology features for novel biomarker discovery
- Any variety of other feature detection and classification tasks performed on Whole Slide Images (WSIs)
- Detecting clinical features within tissues
- Determining the type of tissue or type of stain
- Determining pathology image quality

With a small amount of labelled data, you can train a classifier model on top of
Path Foundation embeddings. Furthermore, the embedding from each tissue patch
only needs to be generated once and can be used as an input for a variety of
different classifiers, with very little additional compute.

For an example of how to use the model to train classifiers using the
[CAMELYON16 Dataset](https://camelyon16.grand-challenge.org), see the following
Colab notebooks:

- [Path Foundation linear classifier notebook in Colab - for Cloud Storage](https://colab.research.google.com/github/google-health/path-foundation/blob/master/notebooks/train_data_efficient_classifier_gcs.ipynb)
- [Path Foundation linear classifier notebook in Colab - for Google Cloud
  DICOM](https://colab.research.google.com/github/google-health/path-foundation/blob/master/notebooks/train_data_efficient_classifier_dicom.ipynb)

### Similar-image search

Path Foundation can also be used to find similar images within or even between
WSIs. By selecting reference patches of interest, you can use the embeddings
from Path Foundation to quantify similarity between any other set of patches
based on distances in the Path Foundation embedding space. This lets you
identify and retrieve patches most similar to your reference regions.

## Next Steps

- [Get started using the model](https://developers.google.com/health-ai-developer-foundations/path-foundation/get-started)