Algorithm Bake-Off: Comparing AI Algorithms for CT Segmentation with Flywheel Data Exchange and Jupyter Notebooks
We use one platform to conduct a comparison of two deep learning liver and kidney segmentation models, revealing the importance of complete DICOM header curation to reduce segmentation errors.
Fill form to unlock content
Error - something went wrong!
Access the Article
Thank you for your submission!
In this example, we use one platform to conduct a comprehensive comparison of two deep learning-based CT segmentation algorithms for liver and kidney segmentation—Total-Segmentator and Swin-UNETR. We evaluate their performance on an independent, open-source CT dataset, to address the challenges of segmentation performance across different datasets. The comparison reveals the importance of complete DICOM header curation to identify conditions that contribute to segmentation errors.
