Abstract
Registration-based methods are commonly used in the automatic segmentation of magnetic resonance (MR) brain images. However, these methods are not robust to the presence of gross pathologies that can alter the brain anatomy and affect the alignment of the atlas image with the target image. In this work, we develop a robust algorithm, MU-Net-R, for automatic segmentation of the normal and injured rat hippocampus based on an ensemble of U-net-like Convolutional Neural Networks (CNNs). MU-Net-R was trained on manually segmented MR images of sham-operated rats and rats with traumatic brain injury (TBI) by lateral fluid percussion. The performance of MU-Net-R was quantitatively compared with methods based on single and multi-atlas registration using MR images from two large preclinical cohorts. Automatic segmentations using MU-Net-R and multi-atlas registration were of excellent quality, achieving cross-validated Dice scores above 0.90 despite the presence of brain lesions, atrophy, and ventricular enlargement. In contrast, the performance of single-atlas segmentation was unsatisfactory (cross-validated Dice scores below 0.85). Interestingly, the registration-based methods were better at segmenting the contralateral than the ipsilateral hippocampus, whereas MU-Net-R segmented the contralateral and ipsilateral hippocampus equally well. We assessed the progression of hippocampal damage after TBI by using our automatic segmentation tool. Our data show that the presence of TBI, time after TBI, and whether the hippocampus was ipsilateral or contralateral to the injury were the parameters that explained hippocampal volume.
Highlights
In-vivo magnetic resonance imaging (MRI) is a key technology for tracking neuroanatomical changes in brain diseases in both animal models and humans
While few methods have been proposed for the anatomical segmentation of brain MRIs with lesions on human data [10,11,12,13], the literature comparing the region segmentation accuracy of different methods in lesioned brains is limited and the potential biases arising from difference of the segmentation accuracy between healthy and lesioned brains has been rarely analyzed
We have analyzed the data from the Abbreviations: BET, brain extraction tool; CNN, convolutional neural network; CS, compactness score; HD95, 95th percentile of the Hausdorff distance; MR, magnetic resonance; MRI, magnetic resonance imaging; ROI, region of interest; Symmetric image Normalization (SyN), symmetric image normalization; STEPS, similarity and truth estimation for propagated segmentation; TBI, traumatic brain injury; VS, volume similarity
Summary
In-vivo magnetic resonance imaging (MRI) is a key technology for tracking neuroanatomical changes in brain diseases in both animal models and humans. Magnetic resonance imaging is noninvasive and allows longitudinal studies to be performed in living animals, with brains imaged in three dimensions (3D) at multiple time points This enables monitoring of disease progression and Robust Automatic Hippocampus Segmentation image registration. While few methods have been proposed for the anatomical segmentation of brain MRIs with lesions on human data [10,11,12,13], the literature comparing the region segmentation accuracy of different methods in lesioned brains is limited and the potential biases arising from difference of the segmentation accuracy between healthy and lesioned brains has been rarely analyzed In this regard [14] documented a significant drop in the performance of registration-based segmentation of the hippocampus due to atrophy, and [15] detected a drop in segmentation quality as a consequence of Huntington’s disease across different segmentation methods. The information needed to segment a new image is encoded in the parameters of the neural network, eliminating the need for
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have