S2320
Physics - Machine learning and AI algorithms
ESTRO 2026
segmentation model and a user-defined app for segmenting stem-cell-rich regions which have been shown to influence the risk of xerostomia. Parotid, Mandible and masseter segmentation generated by the H&N-OAR-segmentation app were passed as inputs to stem-cell-segmentation app. Chaining the two apps together created a seamless analysis workflow that can be reproducibly used. Results can be downloaded via the API allowing for visualization in a client-based software. We provide a notebook to visualize the results using pyCERR and to batch- capture snapshots for review.
Conclusion: The CGC serves as a resource for reproducible deployment of radiological and radiotherapy image analysis to validate outcome models by simplifying access to multi-modal data, compute-hardware, and re-runnable software. References: Lau, Jessica W., et al. "The Cancer Genomics Cloud: collaborative, reproducible, and democratized—a new paradigm in large-scale computational research." Cancer research 77.21 (2017): e3-e6.Iyer, Aditi, et al. "Prospectively-validated deep learning model for segmenting swallowing and chewing structures in CT." Physics in Medicine & Biology 67.2 (2022): 024001.Jiang, Jue, et al. "Self-supervised 3D anatomy segmentation using self-distilled masked image transformer (SMIT)." International conference on medical image computing and computer-assisted intervention. Cham: Springer Nature Switzerland, 2022.Apte, Aditya P., et al. https://doi.org/10.1002/mp.13046van Rijn-Dekker, et al.. https://doi.org/10.3390/cancers16244283 Keywords: Cloud, Reproducibility, Validation
Made with FlippingBook - Share PDF online