The Application Study Of Structural And Functional MRI In Cognitive Impairment Diseases

In 2009, Dr. Petersen published a review titled Mild Cognitive Impairment (MCI) Ten Years Later, in which he summarized and indictaed MCI represented the prodromal condition for cognitive disorders such as Alzheimer’s disease (AD) and other dementias. Hence, the MCI has become a focus on clinical, epidemiologic, neuroimaging, biomarker, neuropathological, disease mechanism, and clinical trials research. People with amnestic mild cognitive impairment (aMCI) are the subtype of single domain and have an increased risk of developing AD compared with their cognitively normal controls. The new MCI criteria included the clinical phenotypes of amnestic MCI (aMCI) and nonamnestic MCI (naMCI) with the subtypes of single and multiple domain classifications. The naMCI represents the prodromal condition for Frontotemporal lobar degeneration (FTLD), Dementia with Lewy Bodies (DLB) or vascular dementia (VaD). Single non-memory mild cognitive impairment (snmMCI) represents FTLD or DLB. Multiple domains mild cognitive impairment (mdMCI) may progress to AD or VaD. However, because of lack of the knowledge of DLB, FTLD and VaD, these dementias were misdiagnosed with AD. Therefore, in 2007, Dr. Dubois proposed the new criteria for AD and added the structural MRI as an important support. In 2010, Dr. Jack published on Lancet Neurol with the Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade. He put forwarded noninvasive quantitative MR would be useful for early diagnosis and treatment of dementia. He also indicated that both MRI and CSF tau were the predictive of future conversion from MCI to AD.Our aim was to identify patterns of atrophy unique to each of these diseases and to build an MRI-based differential diagnosis system. Neurodegenerative disorders can disrupt molecular pathways, local circuits in specific brain regions, as well as higher-order neural networks. Based on this theory of multiple level disruptions, we explored the value of Structural MRI (sMRI), Diffusion Weighted Imaging (DWI), Diffusion Tensor Imaging (DTI) and Proton Magnetic Resonance Spectroscopy (1MRS) in differential diagnosis of cognitive impairment diseases.Part 1 was the study of sMRI in cognitive impairment diseases. Discriminant analysis was set up by linear measurements. The atrophy pattern was analyzed by VBM8 tool based on 3DT1TFE sequences.Part 2 was the DWI and DTI study in cognitive impairment diseases. DWI and DTI map were analyzed by single ROI method. Because of the inconvenience of drawing ROIs, we designed and validated an image-processing system named Brain Search (BS) based on anatomical volumes of interest (AVOI) to quantify the alteration in diffusivity of water in aMCI and AD. Part 3 was the study of Single Voxel Spectroscopy (SVS) and Multi voxel (MV) 1HMRS (Chemical shift imaging, CSI) in cognitive impairment diseases. We improved the sequences of CSI. We investigated within-acquisition repeatability and between-acquisition reproducibility of hippocampal metabolite ratios obtained using automated proton MR spectroscopy. The metabolic pattern in hippocampus was analyzed. The abstract in detail was as following and written in six units. Unit Ⅰ:The abstract of review was omitted here. Unit Ⅱ:Structural MRI study in aMC、AD、FTLD and VCIPurpose:To investigate interobserver agreement and intraobserver agreement of linear measurements. To set up linear discriminant analysis (LDA) and MRI report model for cognitive impairment diseases. To identify patterns of atrophy by VBM8 based on 3DT1TFE sequences.Material and Methods:27 patients with AD,21 with aMCI,19 with FTLD,11 with VCIND,15 with VaD and 25 with Cognitively controls (CN) were enrolled in this study. Philips Intera 1.5T Master was employed.3DT1TFE sequence with high resolution was used to calculate the volume of gray matter and linear measurements. On the hippocampal head (HH) coronal slice, the linear measurements were set up including the height of hippocampus, the width of choroid fissure, the thickness of perirhinal cortex (PC), the width of temporal stem, the width of lateral cerebral sulcus, the width of sulci frontalig superior and the width of postcentral fissure.3DT1TFE images were normalized and segmented into grey matter map for statistical analysis by SPM8 and VBM8.Interobserver agreement and intraobserver agreement was analyzed by SPSS 18 paired t test and Intra-class Correlation Coefficients (ICC). Discriminant analysis was employed in SPSS 18. Anova test was used to analyze the statistical difference among CN, aMCI, AD and CN, VCIND, VaD groups. The independent t test was used to analyze the statistical difference between CN, FTLD and FTLD, AD group.The false discovery rate (FDR) was adopted in P value adjustment, P<0.001, Xjview was used to show the statistical map between group, and cluster size was set at 5. The full width at half maximum (FWHM) was set at 4 mm for the Smooth.Results:(1) The paired t test of interobserver agreement and intraobserver agreement has no significant difference (P>0.05). The ICC was higher than 0.8.(2) The linear measurements data:the left hippocampal atrophy was statistically significant in AD, VCIND and VaD, the right hippocampal atrophy was statistically significant in all of five cognitive diseases compared with CN. There was significant difference in the bilateral width of choroid fissure, the thickness of perirhinal cortex (PC), the width of temporal stem in all of five cognitive diseases compared with CN. The bilateral widthes of sulci frontalig superior were significantly enlarged only in FTLD. The width of postcentral fissure was significantly enlarged in AD, FTLD, VCIND and VaD.(3) The LDA was set up as belowGroup A was among CN, aMCI and AD:A1(CN):Y1=(14.178×the width of right choroid fissure)+(10.532 ×the width of left postcentral fissure)-14.061.A2(aMCI):Y2=(26.077×the width of right choroid fissure)+(l 0.660 X the width of left postcentral fissure)-17.559.A3(AD):Y3=(38.818 ×the width of right choroid fissure)+(21.689 ×the width of left postcentral fissure)-60.317.Group B was among AD, FTLD and VaD:B1(AD):Y1=(-28.164 ×the width of left choroid fissure)+(77.254 ×the width of right choroid fissure)+(13.217×the width of right sulci frontalig superior)+(9.637× the width of left postcentral fissure)-49.703.B2(FTLD):Y2=(-17.439 ×the width of left choroid fissure)+(70.680 ×the width of right choroid fissure)+(26.192×the width of right sulci frontalig superior)+(3.892 X the width of left postcentral fissure)-69.639.B3(VaD):Y3=(-2.779×the width of left choroid fissure)+(35.969×the width of right choroid fissure)+(12.833 ×the width of right sulci frontalig superior)+(9.096 × the width of left postcentral fissure)-40.393.Group C was among CN, VCIND and VaD:C1(CN):Y1=(538.226×the height of left hippocampus)-(156.939×the width of left choroid fissure)+(479.802×the width of left temporal stem)+(177.378×the width of left sulci frontalig superior)-(11.478 × the width of left postcentral fissure)-517.786.C2(VCIND):Y5=(369.724×the height of left hippocampus)-(68.461 ×the width of left choroid fissure)+(330.147×the width of left temporal stem)+(154.183×the width of left sulci frontalig superior)-(l.394 × the width of left postcentral fissure)-325.595.C3(VaD):Y6=(403.613×the height of left hippocampus)-(71.966×the width of left choroid fissure)+(352.365 X the width of left temporal stem)+(130.247×the width of left sulci frontalig superior)+(25.484 X the width of left postcentral fissure)-394.875.Group D was between aMCI and FTLD:DI(aMCI):Y7=(347.044 X the height of right hippocampus)+(218.404× the width of left choroid fissure)+(28.638 X the thickness of right perirhinal cortex)+(130247 ×the width of right sulci frontalig superior)-139.511.D2(FTLD):Y8=(435.650 × the height of right hippocampus)+(290.095 × the width of left choroid fissure)+(47.144 X the thickness of right perirhinal cortex)+(33.257×the width of right sulci frontalig superior)-226.177.(4) VBM8 analysis data:The bilateral front of hippocampal head were significantly atrophy compared with CN. The atrophy was extended to bilateral perirhinal cortex (PC), the front and exterior part of temporal lobe. The atrophy pattern in cingulate gyrus was from anterior to posterior. Compared with AD, the atrophy pattern in FTLD was found in the right uncus, right hippocampal head, bilateral end of hippocampal tail, left Thalamus_Pulvinar, anterior cingulum, bilateral anteriomedialis of frontal lobe.Conclusion:The linear measurement based on 3DT1TFE sequence is reliable tool with better inter and intra observer agreement. The LDA is helpful in the diagnosis and differential diagnosis in cognitive impairment disease. Based on these data, a MRI report system for cognitive diseases was set up. The atrophy pattern in FTLD is characterized by right uncus, right hippocampal head and bilateral frontal lobe. The significant atrophy in left Thalamus_Pulvinar can explain the character changes in FTLD patients. The atrophy in bilateral end of hippocampal tail was significantly difference compared with AD, which is useful for explain the visual-spatial disorder syndrome in FTLD.Unit III:Evaluation of single voxle 1HMRS and single ROI ADC value in cognitive impairment diseasesPurpose:To investigate alterations of advanced functional MRI marker such as proton magnetic resonance spectroscopy (1HMRS) and apparent diffusion coefficient (ADC) value of diffusion weighted imaging (DWI) in mild cognitive impairment (aMCI). Multi indicators were combined in order to improve the diagnostic value of MRS and ADC.Methods:Single voxel 1HMRS and single ROI DWI-ADC were administered to 13 patients with AD,14 patients with aMCI, and 13 CN. Alterations of NAA/Cr, ml/Cr and ADC value in left hippocampus and left temporoparietal region among groups were compared. Sensitivity and specificity of different markers were analyzed, individually and in combination. All participants were evaluated by the mini mental state examination (MMSE), and the correlation between NAA/Cr, ml/Cr, ADC and the score of MMSE were analyzed separately.Results:NAA/Cr, ml/Cr and ADC value in hippocampus among AD, aMCI patients and CN were significantly different (P< 0.05). AT a fixed specificity of 84.6%, the high sensitivity of 100% and 92.9% in differential AD and aMCI from CN were concluded by combining the three indicators. The ROC plots illustrated the area under the curve of multi markers was biggest among the all four curves and the sensitivity of multi markers was highest. Best correlation was between ADC and MMSE, not between NAA/Cr or ml/Cr and MMSE.Conclusion:Alterations of NAA/Cr, mI/Cr and ADC in the hippocampus and the temporoparietal regions were helpful to clinical diagnosis in aMCI. Furthermore, it had potential in predicting the progression of aMCI to AD if we combined above multi indicators.Unit IV:Evaluation of apparent diffusion coefficient (ADC) map in cognitive impairment diseases by using image analysis software named Brain Search based on anatomical volumes of interest (AVOI)Purpose:To investigate alterations of apparent diffusion coefficient (ADC) value in aMCI and AD. To verify the validity of Brain Search (BS) software based on anatomical volumes of interest (AVOI) for ADC maps, designed by authors of this article.Material and Methods:The images of subjects were acquired by the 1.5-T clinical MRI scanner (Philips Intera Master Medical Systems, Netherlands) with a standard quadrature head coil. 174 aged people were screened for eligibility in this study. According to exclusion criteria and quality control for ADC maps,25 patients with AD,26 with aMCI and 18 CN were recruited (Petersen and NINCDS/ADRDA criteria). EPI sequence of DWI with fluid attenuation inversion recovery (FLAIR) was adopted with TR/TI/TE=6000/1900/95 ms, b=1000 s/mm2, an in plane resolution of 2.0 mm and a slice thickness of 2.5mm. The independent ADC map was generated after imaging acquisition and analyzed by using Brain Search (BS).The ADC values in bilateral hippocampus from 18 CN were analyzed by both single ROI and BS method. The DWI sequences from 10 CN were adopted in repeatability and reproducibility test. The Pearson correlation and Intra-class Correlation Coefficients (ICC) were adopted. Statistical maps of group differences for the mean value of each of anatomical brain regions were displayed at a significance value of P< 0.05.Each image pre-processing were performed using SPM8 including conversion Dicom raw data into analyzed type (*.hdr and*.img). Each b0image as "source image" was spatially normalized with the EPI template image using the normalization function with standard defaults in SPM8. Then a set of warps were written back to the ADC map by using "image to write" function. The normalized ADC images were then imported into the Brain search (BS) software (Written by JG.Z and WH.G). Regional parcellation was adopted in BS by using the automated anatomic labeling atlas validated previously by Tzourio-Mazoyer et al. This parcellation divides each cerebral hemisphere into 45 anatomical brain regions of interest. Statistical maps of group differences for the mean value of each of anatomical brain regions were displayed at a significance value of P< 0.05.Results:ADC values in the bilateral hippocampus by single ROI were correlated with that by BS (P<0.05). The Pearson coefficient and ICC for within-acquision repeatability from 90 brain areas were higher than 0.75. And Pearson coefficient and ICC for between-acquisition reproducibility were lower than 0.7 only from 7/90 (7.8%) brain areas.Abnormalities in ADC maps in aMCI and AD compared to CN were observed by BS method as following. The ADC obviously elevated in aMCI than NC in the limbic cortex (Hippocampus_L, ParaHippocampal_L, Insula_R), Thalamus_L, Angular_R, and Frontal lobe. And between aMCI and AD, the statistical different brains areas of ADC values included the limbic cortex (Hippocampus_R), Bilateral Temporal_Pole_Sup, and Frontal lobe. Furthermore, between AD and NC, besides the bilateral hippocampus, significantly different brain areas extended to surrounding limbic system and related cortex (Cingulum_Mid, ParaHippocampal_R, Insula_R, Temporal_Pole_Sup_L, SupraMarginal_R, Angular_L and Frontal lobe). There had been a negative correlation between the water diffusivity and the scores of MMSE, MoCA, Digit-symbol coding, Digit span (in order), Digit span (backwards), Raven’s IQ, WAIS IQ. And the water diffusivity was positively correlated with the scores of CDR,ADL,ADAS-Cog.Conclusion:(1) The water diffusivity in aMCI and AD than CN had asymmetry anatomical lateralization. The pattern of water diffusivity in hippocampal and parahippocampal was initially leftward involvement, progressing to bilateral involvement and extending rightward lateralization in the later stage of AD.(2) The water diffusivity alterations in the limbic cortex and Papez circuit among groups were shown in an intuitive way by BS method. Hopefully these novel insights will better the analytic way of ADC map in diagnostic assessment of aMCI and AD or into therapeutic monitoring in the progression of aMCI to AD.(3)The advantage of AVOI is to parcellate each cerebral cortex into 45 anatomical brain ROIs with decreased type II fault which is inevitable in voxel method. Furthermore, due to the brain images with anatomically AAL label, it was convenient to use the analysis report of BS method.Unit Ⅴ:The single ROI method of DTI map in AD and DLBPurpose:To set up a single ROI method for quantifying water diffusivity changes in white matter fibre bundles by DTI method. To identify the patterns of water diflusivity changes (FA, MD,et al) in patients with Dementia with Lewy bodies (DLB) and AD.Material and Methods:This part data were acquired during the author has worked in Mayo clinic from July to October 2009. And all of this part data was reserved by the Mayo foundation. Here we courtesy Dr. Clifford R. Jack and Kejal Kantarci for permission of data presentation in my doctoral dissertation.We studied clinically diagnosed age, gender and education matched 30 DLB,30 AD, and 30 CN subjects. DTI was performed at 3T with 21 diffusion sensitive gradient directions. Color coded fractional anisotropy (FA) maps were used for measuring tract-based diffusivity. According Mori method, we selected 21 fibers and set up the ROI landmarkers. The same ROIs were located twice between two weeks by the author (B.Z.) who blinded the clinical data. The MD, FA were calculated by transferring the ROIs to the respective maps that were in the same space as the color coded FA maps.Results:The correlation coefficient for interobserver agreement with more than 0.5 between twice ROI placements included 10 ROIs as following:bilateral anterior cingulum, bilateral posterior cingulum, bilateral ilf, right cp (cerebral peduncle), Fonix, Splenium, right superior cpt (corticopontine tract).In the patients with DLB and AD had decreased FA in the inferior longitudinal fasciculus compared to CN subjects. In addition, patients with AD had decreased FA in Fonix and cingulum. Tract-based FA was reduced and MD was elevated with a trend of increase in both DR and DA in the inferior longitudinal fasciculi of patients with DLB compared to CN. Box plots showed four ILF diffusivity measurements by the presence of visual hallucinations among DLB patients. FA was lower in the inferior longitudinal fasciculi of patients with DLB who were experiencing visual hallucinations compared to the DLB patients who did not have this symptom.Conclusion:In DLB, decreased FA in the inferior longitudinal fasciculi suggests a disruption in the connections between the amygdala and the visual cortex that is associated with visual hallucinations. The radial diffusivity changes in the white matter tracts that connect to the medial temporal lobe identified in AD, point out to a loss of myelin integrity. In the current study, FA was lower in the inferior longitudinal fasciculi of patients with DLB who were experiencing visual hallucinations compared to the DLB patients who did not have this symptom, demonstrating that visual hallucinations in DLB was associated with the disruption of the temporo-occipital connectivity. The measures of diffusivity on DTI are valuable tools for characterizing the tissue abnormalities characteristic of AD and DLB.Unit VI:The study of Multi voxel’HMRS (CSI) in cognitive impairment diseases Purpose:To investigate between-acquisition repeatability and within-acquisition reproducibility of hippocampal metabolite ratios obtained using single voxel and CSI in hippocampus, to improve the sequence of chemical shift imaging (CSI), to investigate the alterations of metabolites in cognitive impairment diseases.Methods:Philips 3.0T (Achieva TX)and SENSE 8 head coil were employed in this study. Single voxel and CSI were administered to 8 CN in order to study the repeatability and reproducibility and improve the sequences of CSI. We compared the quality of bilateral and unilateral sequence with 16 or 4 rest slabs. And we enrolled 11 patients with AD,12 patients with aMCI,4 with FTLD,4 with VCI and 9 CN. VOI were placed as following, VOI 1 left hippocampus, VOI 2 right hippocampus, the reconstruction voxel size were 7mm X 7mm in hippocampus region. Alterations of NAA/Cr, mI/Cr in hippocampus among groups were compared by ANOVA SPSS 18 with p<0.05. The Pearson correlation and Intra-class Correlation Coefficients (ICC) were adopted (P< 0.05) in repeatability and reproducibility test.Results;There are less lip peak artifacts in unilateral sequence with 4 rest slabs than bilateral with 16 rest slabs (P<0.05). All the coefficients of variation from the posterior hippocampus (8-11%) were less than those from the anterior hippocampus and those from single voxel. NAA/Cr, mI/Cr in hippocampus among AD, aMCI patients and CN were significantly different (P< 0.05). NAA/Cr in the right hippocampus was lower than that in left. MI/Cr in the left hippocampus was higher than that in right.Conclusion:The NAA/Cr and mI/Cr ratio in the posterior hippocampus was the most reproducible parameter for hippocampal MR spectroscopy. The CSI sequence with unilateral acquisition and 4 rest slabs was set up for spectroscopy study. Regional variation and technical differences in metabolite ratios must be considered when interpreting proton spectra in hippocampus. NAA/Cr in the right hippocampus represented the AC2 neuron in right might be destructed more than in the left. MI/Cr elevation in hippocampus implied the compensation action in the AD pathological progress.