From slides to AI-ready maps: Standardized multi-layer tissue maps as metadata for artificial intelligence in digital pathology

Artificial Intelligence in Medicine, vol. 174, pp. 103368

Abstract

A Whole Slide Image (WSI) is a high-resolution digital image created by scanning an entire glass slide containing a biological specimen, such as tissue sections or cell samples, at multiple magnifications. These images are digitally viewable, analyzable, and shareable, and are widely used for Artificial Intelligence (AI) algorithm development. WSIs play an important role in pathology for disease diagnosis and oncology for cancer research, but are also applied in neurology, veterinary medicine, hematology, microbiology, dermatology, pharmacology, toxicology, immunology, and forensic science. When assembling cohorts for AI training or validation, it is essential to know the content of a WSI. However, no standard currently exists for this metadata, and such a selection has largely relied on manual inspection, which is not suitable for large collections with millions of objects. We propose a general framework to generate 2D index maps (tissue maps) that describe the morphological content of WSIs using common syntax and semantics to achieve interoperability between catalogs. The tissue maps are structured in three layers: source, tissue type, and pathological alterations. Each layer assigns WSI segments to specific classes, providing AI-ready metadata. We demonstrate the advantages of this standard by applying AI-based metadata extraction from WSIs to generate tissue maps and integrating them into a WSI archive. This integration enhances search capabilities within WSI archives, thereby facilitating the accelerated assembly of high-quality, balanced, and more targeted datasets for AI training, validation, and cancer research. • Metadata model enables faster assembling of balanced, high-quality datasets for AI development. • The model includes coarse-grained multi-layer tissue maps providing AI-ready metadata. • Metadata model contributes to the community standards of BBMRI-ERIC. • Tool WSIDOM to visualize the three layers (tissue map) of the metadata model. • AI-based proof-of-concept implementation for generating multi-layer tissue maps.

Authors 17

1. Gernot Fiala

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

2. Markus Plass

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

3. Robert Harb

   Medical University of Graz · Graz University of Technology

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria; Graz University of Technology, Institute of Computer Graphics and Vision, Graz, Austria

4. Peter Regitnig

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

5. Kristijan Skok

   Medical University of Graz · University of Maribor

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria; University of Maribor, Institute of Biomedical Sciences, Maribor, Slovenia

6. Wael Al Zoughbi

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

7. Carmen Zerner

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

8. Paul R. Torke

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

9. Michaela Kargl

   Medical University of Graz

   Affiliation as printed

   Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria

10. Heimo Müller corresponding

    Medical University of Graz · Biobanking and Biomolecular Resources Research Infrastructure Consortium

    Affiliation as printed

    Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria; BBMRI-ERIC, Graz, Austria. Electronic address: heimo.mueller@medunigraz.at

11. Tomǎš Brázdil

    Masaryk University

    Affiliation as printed

    Masaryk University, Faculty of Informatics, Brno, Czech Republic

12. Matej Gallo

    Masaryk University

    Affiliation as printed

    Masaryk University, Faculty of Informatics, Brno, Czech Republic

13. Jaroslav Kubín

    Masaryk University

    Affiliation as printed

    Masaryk University, Faculty of Informatics, Brno, Czech Republic

14. Roman Stoklasa

    Masaryk University

    Affiliation as printed

    Masaryk University, Faculty of Informatics, Brno, Czech Republic

15. Rudolf Nenutil

    Masaryk Memorial Cancer Institute

    Affiliation as printed

    Masaryk Memorial Cancer Institute, Department of Pathology, Brno, Czech Republic

16. Norman Zerbe Aachen Institute of Pathology

    Humboldt-Universität zu Berlin · Universitätsklinikum Aachen · Freie Universität Berlin · Charité - Universitätsmedizin Berlin · RWTH Aachen University

    Affiliation as printed

    Institute of Pathology, University Hospital RWTH Aachen, Aachen, Germany; Charité - Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Institute of Medical Informatics, Berlin, Germany; EMPAIA International e.V., Berlin, Germany; European Society of Digital and Integrative Pathology, Lisbon, Portugal

17. Andreas Holzinger

    Medical University of Graz · BOKU University

    Affiliation as printed

    Medical University of Graz, Diagnostic and Research Institute of Pathology, Graz, Austria; University of Natural Resources and Life Sciences Vienna, Human-Centered AI Lab, Institute of Forest Engineering, Department for Ecosystem Management, Climate and Biodiversity, Vienna, Austria