ESTRO 2024 - Abstract Book

S4481

Physics - Machine learning models and clinical applications

ESTRO 2024

Conclusion:

Although the tasks presented involve predicting highly non-rigid deformations with large displacements, deep learning based image registration allowed to streamline the process of annotating the images required for a modern radiotherapy treatment planning while maintaining robust results. Close to perfect dice scores are obtained with a near real-time latency. The complete pipeline is currently being evaluated on an in-house dataset comprised of 137 4DCT/3DMRI pairs from patients over the last 4 years.

Keywords: Registration, 4DCT, Deep Learning

References:

[ref1.] National Lung Screening Trial Research Team. (2013). Data from the National Lung Screening Trial (NLST) [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/TCIA.HMQ8-J677

[ref 2.] Xu, Zhoubing et al. Evaluation of six registration methods for the human abdomen on clinically acquired CT, IEEE Transactions on Biomedical Engineering, 63 (8), pages=1563-1572, 2016

[ref

3.] VoxelMorph:

A

Learning

Framework

for

Deformable

Medical

Image

Registration

Guha

Balakrishnan, Amy

Zhao, Mert

R.

Sabuncu, John

Guttag, Adrian

V.

Dalca

IEEE TMI: Transactions on Medical Imaging. 2019. eprint arXiv:1809.05231

[ref 4.] U-Net vs Transformer: Is U-Net Outdated in Medical Image Registration, Jia, Xi and Bartlett, Joseph and Zhang, Tianyang and Lu, Wenqi and Qiu, Zhaowen and Duan, Jinming, 2022, arXiv preprint arXiv:2208.04939

1505

Digital Poster