20 April 2017 BrainSegNet: a convolutional neural network architecture for automated segmentation of human brain structures
Author Affiliations +
Abstract
Automated segmentation of cortical and noncortical human brain structures has been hitherto approached using nonrigid registration followed by label fusion. We propose an alternative approach for this using a convolutional neural network (CNN) which classifies a voxel into one of many structures. Four different kinds of two-dimensional and three-dimensional intensity patches are extracted for each voxel, providing local and global (context) information to the CNN. The proposed approach is evaluated on five different publicly available datasets which differ in the number of labels per volume. The obtained mean Dice coefficient varied according to the number of labels, for example, it is 0.844 ± 0.031 and 0.743 ± 0.019 for datasets with the least (32) and the most (134) number of labels, respectively. These figures are marginally better or on par with those obtained with the current state-of-the-art methods on nearly all datasets, at a reduced computational time. The consistently good performance of the proposed method across datasets and no requirement for registration make it attractive for many applications where reduced computational time is necessary.
© 2017 Society of Photo-Optical Instrumentation Engineers (SPIE)
Raghav Mehta, Aabhas Majumdar, Jayanthi Sivaswamy, "BrainSegNet: a convolutional neural network architecture for automated segmentation of human brain structures," Journal of Medical Imaging 4(2), 024003 (20 April 2017). https://doi.org/10.1117/1.JMI.4.2.024003 . Submission: Received: 24 September 2016; Accepted: 28 March 2017
Received: 24 September 2016; Accepted: 28 March 2017; Published: 20 April 2017
JOURNAL ARTICLE
11 PAGES


SHARE
Back to Top