Abstract
One of the most accurate and non-invasive prostate imaging methods is magnetic resonance imaging (MRI). Segmentation is needed to find the boundary of the prostate, either automatically or semi-automatically. Recently, fully convolutional neural networks (FCNN) are being used for this purpose. In this paper, to improve the FCNN performance for prostate MRI segmentation, we analyze various structures of shortcut connections together with the size of a deep network and suggest eight different FCNNs-based deep 2D network structures for automatic MRI prostate segmentation. Our evaluations on the PROMISE12 dataset with ten-fold cross-validation indicate improved and competitive results. We analyze the results in detail, considering MRI slices, MRI volumes, test folds, and also the impact on prostate segmentation of using an EndoRectal Coil to capture the prostate MRI. Our best 2D network outperforms the state-of-the-art 3D FCNN-based methods for prostate MRI segmentation on publicly available data, without any further post-processing.
Original language | English |
---|---|
Article number | 8666973 |
Pages (from-to) | 36748-36760 |
Number of pages | 13 |
Journal | IEEE Access |
Volume | 7 |
DOIs | |
Publication status | Published - 13 Mar 2019 |
Bibliographical note
Copyright the Author(s) 2019. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.Keywords
- Image segmentation
- Magnetic resonance imaging
- Feature extraction
- Prostate cancer
- Convolutional Neural Network
- Automatic MRI segmentation
- Fully Convolutional Neural Network
- Prostate MRI segmentation