Data Augmentation via Image Registration

Data augmentation helps improve generalization of deep neural networks, and can be perceived as implicit regularization. It is pivotal in scenarios in which the amount of ground-truth data is limited, and acquiring new examples is costly and time-consuming. This is a common problem in medical image...

Full description

Saved in:
Bibliographic Details
Main Authors: Nalepa, Jakub, Cwiek, Marcin, Dudzik, Wojciech, Kawulok, Michal, Hayball, Michael P., Mrukwa, Grzegorz, Piechaczek, Szymon, Lorenzo, Pablo Ribalta, Marcinkiewicz, Michal, Bobek-Billewicz, Barbara, Wawrzyniak, Pawel, Ulrych, Pawel, Szymanek, Janusz
Format: Conference Proceeding
Language:English
Subjects:
Biomedical imaging
brain-tumor segmentation
data augmentation
Deep learning
Gallium nitride
Image registration
Image segmentation
Magnetic resonance imaging
Training
Tumors
Citations: Items that cite this one
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Data augmentation helps improve generalization of deep neural networks, and can be perceived as implicit regularization. It is pivotal in scenarios in which the amount of ground-truth data is limited, and acquiring new examples is costly and time-consuming. This is a common problem in medical image analysis, especially tumor delineation-in this paper, we focus on brain-tumor segmentation from magnetic resonance imaging (MRI), and propose a novel augmentation technique which exploits image registration to benefit from subtle spatial and/or tissue characteristics captured within the training set. We used a set of MRI scans of 44 low-grade glioma patients, augmented it using the proposed technique, and exploited it to train U-Net-based deep networks. The results show that our augmentation delivers statistically important boost of performance without sacrificing inference speed.
ISSN:2381-8549
DOI:10.1109/ICIP.2019.8803423