Generalization of voxelwise classifiers is hampered by differences between MRI-scanners, e.g. different acquisition protocols and field strengths. To address this limitation, we propose a Siamese neural network (MRAI-NET) that extracts acquisition-invariant feature vectors. These can consequently be used by task-specific methods, such as voxelwise classifiers for tissue segmentation. MRAI-NET is tested on both simulated and real patient data. Experiments show that MRAI-NET outperforms voxelwise classifiers trained on the source or target scanner data when a small number of labeled samples is available.
@article{arxiv.1810.07430,
title = {Learning an MR acquisition-invariant representation using Siamese neural networks},
author = {Wouter M. Kouw and Marco Loog and Wilbert Bartels and Adriënne M. Mendrik},
journal= {arXiv preprint arXiv:1810.07430},
year = {2019}
}
Comments
3 figures, submitted to International Symposium on Biomedical Imaging 2019