Related Experiment Video
Updated: Jul 6, 2025
07:32
Author Spotlight: Investigating Immune Cell Dynamics in the Tumor Microenvironment — Challenges and Innovations in Cancer Prognosis
Published on: April 12, 2024
1.3K
Open and reusable deep learning for pathology with WSInfer and QuPath.
Jakub R Kaczmarzyk1, Alan O'Callaghan2, Fiona Inglis2
1Department of Biomedical Informatics, Stony Brook University, Stony Brook, NY, USA. jakub.kaczmarzyk@stonybrookmedicine.edu.
NPJ Precision Oncology
|January 10, 2024
Summary
This study introduces WSInfer, an open-source software ecosystem for sharing and reusing deep learning models in digital pathology. This promotes wider research into pathology
Area of Science:
- Computational pathology
- Artificial intelligence in medicine
- Biomedical informatics
Background:
- Deep learning models are increasingly used in digital pathology.
- A significant challenge is the lack of reusability for these complex models.
- This hinders the broader application and validation of AI tools in pathology.
Purpose of the Study:
- To develop an open-source software ecosystem for digital pathology.
- To facilitate the sharing and reuse of deep learning models.
- To accelerate research in diagnostic, prognostic, and predictive pathology.
Main Methods:
- Development of WSInfer, an open-source software ecosystem.
- Focus on streamlining the process of model sharing and reuse.
- Implementation of tools for managing and deploying deep learning models.
Main Results:
- WSInfer provides a centralized platform for deep learning models in digital pathology.
- The ecosystem simplifies model accessibility and integration into research workflows.
- Facilitates collaboration and reduces redundant development efforts.
Conclusions:
- WSInfer enhances the accessibility and reusability of deep learning models.
- This empowers researchers to advance diagnostic, prognostic, and predictive capabilities in digital pathology.
- Promotes standardization and reproducibility in AI-driven pathology research.
