Skip to content

Cardio-AI/fl-multi-modal-dataset-creation

Repository files navigation

Multi-Modal Dataset Creation

This repository contains code for additional SR-template structures extending the highdicom Python library. The following templates have been added to describe annotations for ECG reports (TID 3700):

Additionally, a Kaapana extension has been developed to enable querying of annotation metadata(segmentation or structure reports) within their respective report modalities, such as images (CT, MR, CR...) or waveforms (ECG). This allows cohorts to be selected multimodally and with greater specificity to additional reports. To install the extension, refer to the instructions here.

For more detailed description we refer to our associated paper.

Abstract

The unification of electronic health records promises interoperability of medical data. Divergent data storage options, inconsistent naming schemes, varied annotation procedures, and disparities in label quality, among other factors, pose significant challenges to the integration of expansive datasets especially across instiutions. This is particularly evident in the emerging multi-modal learning paradigms where dataset harmonization is of paramount importance. Leveraging the DICOM standard, we designed a data integration and filter tool that streamlines the creation of multi-modal datasets. This ensures that datasets from various locations consistently maintain a uniform structure. We enable the concurrent filtering of DICOM data (i.e. images and waveforms) and corresponding annotations (i.e. segmentations and structured reports) in a graphical user interface. The graphical interface as well as example structured report templates is openly available at https://github.com/Cardio-AI/fl-multi-modal-dataset-creation.

License

high-dicom SR templates

The high dicom-SR templates are MIT licensed.

annotation-collect-metadata

The annotation-collect-metadata can redistribute and/or modify under the terms of the GNU Affero General Public License. For more information see annotation-collect-metadata

Citation

@InProceedings{10.1007/978-3-658-44037-4_39,
author="T{\"o}lle, Malte and Burger, Lukas and Kelm, Halvar and Engelhardt, Sandy",
editor="Maier, Andreas and Deserno, Thomas M. and Handels, Heinz and Maier-Hein, Klaus and Palm, Christoph and Tolxdorff, Thomas",
title="Towards Unified Multi-modal Dataset Creation for Deep Learning Utilizing Structured Reports",
booktitle="Bildverarbeitung f{\"u}r die Medizin 2024",
year="2024",
publisher="Springer Fachmedien Wiesbaden",
address="Wiesbaden",
pages="130--135",
isbn="978-3-658-44037-4"
}

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •