TY - JOUR
T1 - Difference in distribution functions
T2 - A new diffusion weighted imaging metric for estimating white matter integrity
AU - Du, Jing
AU - Koch, Forrest C.
AU - Xia, Aihua
AU - Jiang, Jiyang
AU - Crawford, John D.
AU - Lam, Ben C.P.
AU - Thalamuthu, Anbupalam
AU - Lee, Teresa
AU - Kochan, Nicole
AU - Fawns-Ritchie, Chloe
AU - Brodaty, Henry
AU - Xu, Qun
AU - Sachdev, Perminder S.
AU - Wen, Wei
N1 - Funding Information:
MAS cohort was supported by National Health and Medical Research Council (NHMRC) Australia Project Program Grants ID350833 , ID568969 and ID1093083 . RCCS was supported by grants from the National Key Research and Development Program of China ( 2016YFC1300600 ), Shanghai Science and Technology Committee Project ( Natural Science Funding ; No. 19ZR1430500 ), the Project of Collaborative Innovation Centre of Translational Medicine ( TM201808 ) and SJTU-UNSW Collaborative Research Fund . The work was also supported by NHMRC Project Grant ID630593 and John Holden Family Foundation.
PY - 2021/10/15
Y1 - 2021/10/15
N2 - Diffusion weighted imaging (DWI) is a widely recognized neuroimaging technique to evaluate the microstructure of brain white matter. The objective of this study is to establish an improved automated DWI marker for estimating white matter integrity and investigating ageing related cognitive decline. The concept of Wasserstein distance was introduced to help establish a new measure: difference in distribution functions (DDF), which captures the difference of reshaping one's mean diffusivity (MD) distribution to a reference MD distribution. This new DWI measure was developed using a population-based cohort (n=19,369) from the UK Biobank. Validation was conducted using the data drawn from two independent cohorts: the Sydney Memory and Ageing Study, a community-dwelling sample (n=402), and the Renji Cerebral Small Vessel Disease Cohort Study (RCCS), which consisted of cerebral small vessel disease (CSVD) patients (n=171) and cognitively normal controls (NC) (n=43). DDF was associated with age across all three samples and better explained the variance of changes than other established DWI measures, such as fractional anisotropy, mean diffusivity and peak width of skeletonized mean diffusivity (PSMD). Significant correlations between DDF and cognition were found in the UK Biobank cohort and the MAS cohort. Binary logistic analysis and receiver operator characteristic curve analysis of RCCS demonstrated that DDF had higher sensitivity in distinguishing CSVD patients from NC than the other DWI measures. To demonstrate the flexibility of DDF, we calculated regional DDF which also showed significant correlation with age and cognition. DDF can be used as a marker for monitoring the white matter microstructural changes and ageing related cognitive decline in the elderly.
AB - Diffusion weighted imaging (DWI) is a widely recognized neuroimaging technique to evaluate the microstructure of brain white matter. The objective of this study is to establish an improved automated DWI marker for estimating white matter integrity and investigating ageing related cognitive decline. The concept of Wasserstein distance was introduced to help establish a new measure: difference in distribution functions (DDF), which captures the difference of reshaping one's mean diffusivity (MD) distribution to a reference MD distribution. This new DWI measure was developed using a population-based cohort (n=19,369) from the UK Biobank. Validation was conducted using the data drawn from two independent cohorts: the Sydney Memory and Ageing Study, a community-dwelling sample (n=402), and the Renji Cerebral Small Vessel Disease Cohort Study (RCCS), which consisted of cerebral small vessel disease (CSVD) patients (n=171) and cognitively normal controls (NC) (n=43). DDF was associated with age across all three samples and better explained the variance of changes than other established DWI measures, such as fractional anisotropy, mean diffusivity and peak width of skeletonized mean diffusivity (PSMD). Significant correlations between DDF and cognition were found in the UK Biobank cohort and the MAS cohort. Binary logistic analysis and receiver operator characteristic curve analysis of RCCS demonstrated that DDF had higher sensitivity in distinguishing CSVD patients from NC than the other DWI measures. To demonstrate the flexibility of DDF, we calculated regional DDF which also showed significant correlation with age and cognition. DDF can be used as a marker for monitoring the white matter microstructural changes and ageing related cognitive decline in the elderly.
KW - diffusion weighted imaging
KW - white matter
KW - ageing
KW - cognition
U2 - 10.1016/j.neuroimage.2021.118381
DO - 10.1016/j.neuroimage.2021.118381
M3 - Article
C2 - 34252528
AN - SCOPUS:85109939184
SN - 1053-8119
VL - 240
JO - NeuroImage
JF - NeuroImage
M1 - 118381
ER -