Normal-appearing White Matter In Assessment Of The Pathogenesis And Prediction Of SVCI

Part I: Characterizing the penumbra of white matter hyperintensities and its association with cognitive function in patients with cerebral small vessel diseaseObjective: White matter hyperintensities(WMH)is a maior neuroimaging maker of cerebral small vessel disease(CSVD).The normal-appearing white matter(NAWM)surrounding WMH is frequently sub-damaged and at a high risk of converting into WMH,referred to as the WMH penumbra.The goal of this study was to define the extent of WMH penumbra and further explore its association with cognitive impairment in CSVD subjectsMaterials and Methods:One hundred and fourteen CSVD patients and thirty-nine healthy controls(CN)were recruited.According to cognitive status,one hundred and fourteen CSVD patients were divided into seventy-three mild cognitive impairment(MCI)patients and forty-one non-cognitive impairment(NCI)patients.All subjects underwent neuropsychological assessment and GE 3T MR scans,including diffusion tensor imaging(DTI)and arterial spin labeling(ASL).Firstly,WMH was classified into periventricular WMH(PVWMH)and deep WMH(DWMH).To determine the extents of the cerebral blood flow(CBF)and DTI penumbra,a NAWM layer mask was generated for periventricular WMH(PVWMH)and deep WMH(DWMH)separately.Mean values of CBF,fractional anisotropy(FA),mean diffusivity(MD)within the WMH and its corresponding NAWM layer masks were computed and compared using paired t-tests.Imaging paremeters of WMH and each modal WMH penumbra in all groups(CN,NCI and MCI)were compared using F test.Then,each two groups were compared using two sample t tests.The correlation between the image parameters of WMH and its penumbra with cognitive score was computed in MCI group.Results:(1)The extents of the WMH penumbra in CN group were as follows: 13 mm for the PVWMH-CBF penumbra,6 mm for the PVWMH-FA penumbra,5 mm for the PVWMH-MD penumbra;13 mm for the DWMH-CBF penumbra,3 mm for the DWMH-FA penumbra,2mm for the DWMH-MD penumbra.(2)The extents of the WMH penumbra in NCI group were as follows: 11 mm for the PVWMH-CBF penumbra,5 mm for the PVWMH-FA penumbra,4 mm for the PVWMH-MD penumbra;11 mm for the DWMH-CBF penumbra,2 mm for the DWMH-FA penumbra,1 mm for the DWMH-MD penumbra.(3)The extents of the WMH penumbra in MCI group were as follows: 10 mm for the PVWMH-CBF penumbra,6 mm for the PVWMH-FA penumbra,6 mm for the PVWMH-MD penumbra;7 mm for the DWMH-CBF penumbra,4 mm for the DWMH-FA penumbra,2 mm for the DWMH-MD penumbra.(4)Compared to NCI group,the mean CBF of WMH and CBF-penumbra in MCI group were lower.(5)In MCI group,only the mean FA of PVWMH-FA penumbra was correlated with composite z-scores of global cognition(r = 0.268,p = 0.024,uncorrected).Conclusion: Decreased FA and CBF and increased MD was found in NAWM around PVWMH and DWMH in CSVD patients,indicating that the true range of white matter damage was more extensive than WMH.The scope of CBF penumbra was wider than that of DTI penumbra,suggesting that white matter perfusion disturbances may precede the change of white matter microstructural integrity.Therefore,CBF penumbra may be a potential target for the prevention of both micro and macro structural white matter damage.Part II : The study of NAWM-DTI radiomics in SVCI classification diagnosisObjective:Accurate detection of early stage SVCI(subcortical vascular cognitive impairment)patients is important for treatment and clinical outcome.This study aimed to construct a random forest classification model to classify and predict MCI and NCI in CSVD patients.Materials and Methods:One hundred and seventy-one CSVD patients were recruited,including one hundred and four mild cognitive impairment(MCI)patients with and sixty-seven non-cognitive impairment(NCI)patients.All subjects underwent neuropsychological assessment and MR scans.The Johns Hopkins University ICBM-DTI-81 white matter atlas was applied to NAWM to extract the diffusion features of labeled white matter region in each subject.Total dataset was randomly divided into the training set(3/4 total dataset)and the validaton set(1/4 total dataset).The ratio of MCI to NCI in the training set was consistent with that in the validation set.All-relevant feature selection and model training were performed in training set.The evaluation of the random forest classifier was performed in validation set.The area under the receiver operating characteristic(ROC)curve(AUC),accuracy,sensitivity and specificity were calculated to evaluate the model performance.Then,the correlation between selected features and Mo CA scores was computed.Results:In the validation set,the random forest model achieved an AUC of 0.76,accuracy of 0.72,sensitivity of 0.84,and specificity of 0.63. The discriminative features between MCI and NCI were as follows: mean AD of right anterior limb of internal capsule(ALIC),mean MD of right ALIC,mean RD of left ALIC,mean RD of right anterior corona radiate(ACR),mean RD of right superior corona radiate(SCR),mean FA of left SCR,mean FA of left external capsule(EC),mean FA of right uncinate fasciculus(UF).These discriminative features were significantly different between MCI and NCI.Except for the mean FA of right UF,the other discriminative features were significantly correlated with Mo CA score.Conclusion:The results of this study provide preliminary evidence that the diffusion features of specific NAWM regions can allow discrimination between MCI and NCI.Most of discriminative features were located in prefrontal-subcortical loops.Part III: NAWM in the assessment of WMH evolution and c ogntive function prediction in SVCI patientsObjective:In SVCI patients,WMH evolve diversely.It remains unclear about the association between the sub-damaged normal-appe aring white matter(NAWM)surrounding white matter hyperintensiti es(WMH)and WMH progression.This study aimed to investigate whether the baseline NAWM(especially in the WMH penumbra region)can predict the subsequent evolution of WMH and cognition c hange.Materials and Methods:Thirty-eight SVCI patients were recruited,with an average follow-up interval of 1.07 years.All subjects underwent two time points MR scans with the same scanner.The follow-up images of each subject were registered to their own baseline T1 images.After WMH segmentation,NAWM regions converting into WMH during follow-up(new increased WMH)and persistent NAWM regions in were identified.In addition,shrinking WMH regions during follow-up and persistent WMH regions were identified.The baseline image variables of the following region were compared separately: new increased WMH regions versus persistent NAWM regions,shrinking WMH regions versus persistent WMH regions.Then,the image variables of WMH and NAWM were compared between baseline and follow-up,respectively.Corelations between the baseline imaging parametes of WMH and NAWM and the change of Mo CA were computed.Results:Compared with persistent NAWM regions,NAWM regions converting into WMH showed lower CBF and FA,higher MD,AD and RD at baseline.Most of the new increased WMH during follow-up located in baseline WMH penumbra regions.The baseline image variables of shrinking WMH regions and persistent WMH regions were not found significantly different.Compared with baseline,WMH and NAWM showed lower CBF and FA,higher MD,AD and RD at follow-up,respectively.Corelations between the baseline imaging parametes of WMH and NAWM and ?Mo CA were not signicantly.Conclusion : WMH progression is mostly characterized by the extension of pre-existed WMH.The perfusion and diffusion alterations of NAWM(especially in the WMH penumbra region)can predict the development of WMH.The evolution of WMH is diverse.CSVD-related white matter damage may be a dynamic process.In this study,baseline WMH and NAWM can not predict cognition change,and the underlying mechanism of SVCI remains to be further studied.