TY - JOUR
T1 - Automatic Skull Stripping of Rat and Mouse Brain MRI Data Using U-Net
AU - Hsu, Li Ming
AU - Wang, Shuai
AU - Ranadive, Paridhi
AU - Ban, Woomi
AU - Chao, Tzu Hao Harry
AU - Song, Sheng
AU - Cerri, Domenic Hayden
AU - Walton, Lindsay R.
AU - Broadwater, Margaret A.
AU - Lee, Sung Ho
AU - Shen, Dinggang
AU - Shih, Yen Yu Ian
N1 - Funding Information:
We thank Alicia Stevans at CAMRI for insightful discussion on this manuscript. Funding. This work was supported by the National Institute of Neurological Disorders and Stroke (R01NS091236), National Institute of Mental Health (RF1MH117053, R01MH111429, R41MH113252, and F32MH115439), National Institute on Alcohol Abuse and Alcoholism (P60AA011605, K01AA025383, and T32AA007573), and National Institute of Child Health and Human Development (P50HD103573).
PY - 2020/10/7
Y1 - 2020/10/7
N2 - Accurate removal of magnetic resonance imaging (MRI) signal outside the brain, a.k.a., skull stripping, is a key step in the brain image pre-processing pipelines. In rodents, this is mostly achieved by manually editing a brain mask, which is time-consuming and operator dependent. Automating this step is particularly challenging in rodents as compared to humans, because of differences in brain/scalp tissue geometry, image resolution with respect to brain-scalp distance, and tissue contrast around the skull. In this study, we proposed a deep-learning-based framework, U-Net, to automatically identify the rodent brain boundaries in MR images. The U-Net method is robust against inter-subject variability and eliminates operator dependence. To benchmark the efficiency of this method, we trained and validated our model using both in-house collected and publicly available datasets. In comparison to current state-of-the-art methods, our approach achieved superior averaged Dice similarity coefficient to ground truth T2-weighted rapid acquisition with relaxation enhancement and T2∗-weighted echo planar imaging data in both rats and mice (all p < 0.05), demonstrating robust performance of our approach across various MRI protocols.
AB - Accurate removal of magnetic resonance imaging (MRI) signal outside the brain, a.k.a., skull stripping, is a key step in the brain image pre-processing pipelines. In rodents, this is mostly achieved by manually editing a brain mask, which is time-consuming and operator dependent. Automating this step is particularly challenging in rodents as compared to humans, because of differences in brain/scalp tissue geometry, image resolution with respect to brain-scalp distance, and tissue contrast around the skull. In this study, we proposed a deep-learning-based framework, U-Net, to automatically identify the rodent brain boundaries in MR images. The U-Net method is robust against inter-subject variability and eliminates operator dependence. To benchmark the efficiency of this method, we trained and validated our model using both in-house collected and publicly available datasets. In comparison to current state-of-the-art methods, our approach achieved superior averaged Dice similarity coefficient to ground truth T2-weighted rapid acquisition with relaxation enhancement and T2∗-weighted echo planar imaging data in both rats and mice (all p < 0.05), demonstrating robust performance of our approach across various MRI protocols.
KW - MRI
KW - U-net
KW - brain mask
KW - mouse brain
KW - rat brain
KW - segmentation
KW - skull stripping
UR - http://www.scopus.com/inward/record.url?scp=85093864210&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85093864210&partnerID=8YFLogxK
U2 - 10.3389/fnins.2020.568614
DO - 10.3389/fnins.2020.568614
M3 - Article
AN - SCOPUS:85093864210
VL - 14
JO - Frontiers in Neuroscience
JF - Frontiers in Neuroscience
SN - 1662-4548
M1 - 568614
ER -