There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Marthe Mieling |
| Institution: | University of Lübeck |
| Department: | Department of Psychology |
| Country: | |
| Proposed Analysis: | We would like to request data, collected by the Alzheimer’s Disease Neuroimaging Initiative (ADNI), to further investigate the pathological stage model of AD. Our main goal is to analyze resting state functional magnetic resonance imaging (rsfMRI) data in combination with markers of amyloid- and p-Tau as derived by cerebrospinal fluid (CSF). This approach is mainly based on a previous analysis of anatomical ADNI data and the corresponding publication by Fernández-Cabello et al. (2020). The pathological stage model of AD describes the process of neurodegeneration as a trans-synaptic spreading of pTau and amyloid- originating in the nucleus basalis of Meynert (NbM), projecting to the entorhinal cortex (EC) and then further to cortical targets. Since previous work (including Fernández-Cabello et al., 2020) mainly focused on structural analyses, it remains unclear how these structural changes relate to functional properties within the NbM and EC. Moreover, effective connectivity analyses, such as Dynamic Causal Modelling (DCM), allow for a more direct investigation of the predictions made by the pathological stage model of AD (Liu et al., 2008). Therefore, we intend to investigate rsfMRI properties of the NbM, EC and their connectivity changes on the basis of pTau and amyloid- properties. As such, our results are expected to give new insights into the underlying mechanisms and progression of AD. Requested Data: In order to achieve our goals, we would like to use data from ADNI-Go, ADNI-2 and ADNI-3. We are aware that ADNI-3 data collection is still in progress; however, we hope to begin our analyses using the baseline data and, at a later stage, proceed with the follow-up data. This procedure would provide the larges possible data-set and therefore robust and reliable results. Within each cohort, we kindly ask for: • rsfMRI data (including baseline and follow-up) • CSF biomarker of pTau and amyloid- • APOE4 genotye information (to be included in our models as covariate, see below) • neuropsychological assessment (including scores on memory performance, executive functions, Boston Naming test, the Clock Drawing tests, the six subcategories of the Clinical Dementia Rating and the demographic information). Planned analyses: Following Fernández-Cabello et al. (2020), participants will be assigned into one of two groups on the basis of their CSF ratio. While a CSF ratio < 0.028 represent a normal CSF (nCSF), a CSF ratio > 0.028 represents an abnormal CSF (aCSF) (Schindler et al., 2018). The rsfMRI data will be analysed at baseline, the two year follow-up, and in relation to one another (Fernández-Cabello et al., 2020). Here, the focus will be on the NbM and EC, which will be defined as regions of interest (ROI) using probabilistic maps (Amunts et al., 2005; Zaborszky et al., 2008). While pre-processing of the fMRI data is planned using SPM12, most subsequent analyses will be performed using in house scripts (see below). We plan to investigate (1) regionally specific and (2) interregional rsfMRI properties. (1) regionally specific properties will be based on: • Regional Homogeniety (ReHo), and • Low-frequency fluctuation (fALFF) analyses While ReHo describes the temporal coherence of BOLD time series within a cluster of adjacent voxels and represents the node’s importance of local functional connectivity (Jiang & Zuo, 2016), fALFF describes the ratio of local low-frequency range amplitudes (0.01-0.08 Hz) within a brain region relative to the entire frequency range (Zou et al., 2008). For both measures (ReHo and fALFF), we expect changes in NbM and EC depending on CSF markers and timepoint (Fernández-Cabello et al., 2020). Specifically, based on a 2x2x2 analyses of variance (ANOVA) with the factors group (aCSF, nCSF), region (NbM, EC) and time point (Baseline, follow-up), we expect reduced ReHo and fALFF for aCSF vs nCSF (main effect of group), and this effect should be more pronounced in the NbM as compared to the EC (interaction between group and regions). Moreover, we assume a decrease of ReHo and fALFF in NbM and EC over time (main effect of time), which should be more pronounced in aCSF compared to nCSF (interaction between time and regions). Finally, this time and group dependent reduction should be more pronounced in the NbM as compared to the EC (three-way interaction between group, time and region). To further pinpoint the directionality of the effects (NbMEC), we also plan to employ regression-based modelling to investigate if the pathological changes in the NbM and EC underlie interdependent processes (Fernández-Cabello et al., 2020). In both models (ANOVA and regression), we plan to include the factors CSF, genotyping (APOE4), neuropsychological scores and demographic information as covariates (Fernández-Cabello et al., 2020). Since ReHo and fALFF are different measures of regional connectivity, we will also explore which of them might serve as a better marker. This analysis will be exploratory in nature. In a subsequent step, (2) interregional properties will be investigated based on • Functional connectivity, and • Effective connectivity (Granger Causality and Dynamic Causal Modelling, DCM) For both types of connectivity analyses, we will use the NbM and EC as ROIs. For the functional connectivity analysis (Smitha et al., 2017), we expect the connectivity between both regions to be reduced in the aCSF vs. nCSF group at both time points (main effect group), and this effect should be more pronounced at the second time point for the aCSF group (interaction group x time point). Again, this analysis will be based on ANOVAs including both regions (NbM, EC) and groups (aCSF, nCSF), as well as the covariates demographic data, genotyping (APOE4), clinical diagnosis and neuropsychological scores. Moreover, we will run a moderation analysis to understand the impact of CSF on the functional connectivity between NbM and EC. Since functional connectivity does not allow conclusions regarding the direction of the effect, effective connectivity analyses will be performed. Here, we will apply Granger’s causality (Gates et al., 2010) and, as a second, complementary measure, DCM (Friston et al., 2011). Our main hypotheses are that NbM and EC are reciprocally connected, and this connectivity should change as a function of time and group. Specifically, connectivity strength NbMEC should be reduced in the aCSF group already at baseline and a reduction in NbMEC connectivity should be especially pronounced in the aCSF group at the second time point (follow up). If such a pattern would be observed, an additional analysis could include those parts of the temporo-parietal cortex, that have been reported in the structural analysis by Fernández-Cabello et al. (2020). Data are planned to be analyzed by the PI of the group, Prof. Dr. Nico Bunzeck, a PhD student Ms. Marthe Mieling, and Dr. Martin Göttlich (physicist). To summarize, we would like to receive rsfMRI data, CSF biomarkers, APOE4 information, and neuropsychological data from ADNI-Go, ADNI-2 and ADNI-3. Our main goal is to further investigate the pathological stage model of AD with a focus on functional properties, including effective connectivity, of the NbM and EC. From a more general point of view, our planned analyses have the potential to better understand the disease progression of AD which might help to further develop treatment strategies. |
| Additional Investigators | |
| Investigator's Name: | Nico Bunzeck |
| Proposed Analysis: | We would like to request data, collected by the Alzheimer’s Disease Neuroimaging Initiative (ADNI), to further investigate the pathological stage model of AD. Our main goal is to analyze resting state functional magnetic resonance imaging (rsfMRI) data in combination with markers of amyloid- and p-Tau as derived by cerebrospinal fluid (CSF). This approach is mainly based on a previous analysis of anatomical ADNI data and the corresponding publication by Fernández-Cabello et al. (2020). The pathological stage model of AD describes the process of neurodegeneration as a trans-synaptic spreading of pTau and amyloid- originating in the nucleus basalis of Meynert (NbM), projecting to the entorhinal cortex (EC) and then further to cortical targets. Since previous work (including Fernández-Cabello et al., 2020) mainly focused on structural analyses, it remains unclear how these structural changes relate to functional properties within the NbM and EC. Moreover, effective connectivity analyses, such as Dynamic Causal Modelling (DCM), allow for a more direct investigation of the predictions made by the pathological stage model of AD (Liu et al., 2008). Therefore, we intend to investigate rsfMRI properties of the NbM, EC and their connectivity changes on the basis of pTau and amyloid- properties. As such, our results are expected to give new insights into the underlying mechanisms and progression of AD. Requested Data: In order to achieve our goals, we would like to use data from ADNI-Go, ADNI-2 and ADNI-3. We are aware that ADNI-3 data collection is still in progress; however, we hope to begin our analyses using the baseline data and, at a later stage, proceed with the follow-up data. This procedure would provide the larges possible data-set and therefore robust and reliable results. Within each cohort, we kindly ask for: • rsfMRI data (including baseline and follow-up) • CSF biomarker of pTau and amyloid- • APOE4 genotye information (to be included in our models as covariate, see below) • neuropsychological assessment (including scores on memory performance, executive functions, Boston Naming test, the Clock Drawing tests, the six subcategories of the Clinical Dementia Rating and the demographic information). Planned analyses: Following Fernández-Cabello et al. (2020), participants will be assigned into one of two groups on the basis of their CSF ratio. While a CSF ratio < 0.028 represent a normal CSF (nCSF), a CSF ratio > 0.028 represents an abnormal CSF (aCSF) (Schindler et al., 2018). The rsfMRI data will be analysed at baseline, the two year follow-up, and in relation to one another (Fernández-Cabello et al., 2020). Here, the focus will be on the NbM and EC, which will be defined as regions of interest (ROI) using probabilistic maps (Amunts et al., 2005; Zaborszky et al., 2008). While pre-processing of the fMRI data is planned using SPM12, most subsequent analyses will be performed using in house scripts (see below). We plan to investigate (1) regionally specific and (2) interregional rsfMRI properties. (1) regionally specific properties will be based on: • Regional Homogeniety (ReHo), and • Low-frequency fluctuation (fALFF) analyses While ReHo describes the temporal coherence of BOLD time series within a cluster of adjacent voxels and represents the node’s importance of local functional connectivity (Jiang & Zuo, 2016), fALFF describes the ratio of local low-frequency range amplitudes (0.01-0.08 Hz) within a brain region relative to the entire frequency range (Zou et al., 2008). For both measures (ReHo and fALFF), we expect changes in NbM and EC depending on CSF markers and timepoint (Fernández-Cabello et al., 2020). Specifically, based on a 2x2x2 analyses of variance (ANOVA) with the factors group (aCSF, nCSF), region (NbM, EC) and time point (Baseline, follow-up), we expect reduced ReHo and fALFF for aCSF vs nCSF (main effect of group), and this effect should be more pronounced in the NbM as compared to the EC (interaction between group and regions). Moreover, we assume a decrease of ReHo and fALFF in NbM and EC over time (main effect of time), which should be more pronounced in aCSF compared to nCSF (interaction between time and regions). Finally, this time and group dependent reduction should be more pronounced in the NbM as compared to the EC (three-way interaction between group, time and region). To further pinpoint the directionality of the effects (NbMEC), we also plan to employ regression-based modelling to investigate if the pathological changes in the NbM and EC underlie interdependent processes (Fernández-Cabello et al., 2020). In both models (ANOVA and regression), we plan to include the factors CSF, genotyping (APOE4), neuropsychological scores and demographic information as covariates (Fernández-Cabello et al., 2020). Since ReHo and fALFF are different measures of regional connectivity, we will also explore which of them might serve as a better marker. This analysis will be exploratory in nature. In a subsequent step, (2) interregional properties will be investigated based on • Functional connectivity, and • Effective connectivity (Granger Causality and Dynamic Causal Modelling, DCM) For both types of connectivity analyses, we will use the NbM and EC as ROIs. For the functional connectivity analysis (Smitha et al., 2017), we expect the connectivity between both regions to be reduced in the aCSF vs. nCSF group at both time points (main effect group), and this effect should be more pronounced at the second time point for the aCSF group (interaction group x time point). Again, this analysis will be based on ANOVAs including both regions (NbM, EC) and groups (aCSF, nCSF), as well as the covariates demographic data, genotyping (APOE4), clinical diagnosis and neuropsychological scores. Moreover, we will run a moderation analysis to understand the impact of CSF on the functional connectivity between NbM and EC. Since functional connectivity does not allow conclusions regarding the direction of the effect, effective connectivity analyses will be performed. Here, we will apply Granger’s causality (Gates et al., 2010) and, as a second, complementary measure, DCM (Friston et al., 2011). Our main hypotheses are that NbM and EC are reciprocally connected, and this connectivity should change as a function of time and group. Specifically, connectivity strength NbMEC should be reduced in the aCSF group already at baseline and a reduction in NbMEC connectivity should be especially pronounced in the aCSF group at the second time point (follow up). If such a pattern would be observed, an additional analysis could include those parts of the temporo-parietal cortex, that have been reported in the structural analysis by Fernández-Cabello et al. (2020). Data are planned to be analyzed by the PI of the group, Prof. Dr. Nico Bunzeck, a PhD student Ms. Marthe Mieling, and Dr. Martin Göttlich (physicist). To summarize, we would like to receive rsfMRI data, CSF biomarkers, APOE4 information, and neuropsychological data from ADNI-Go, ADNI-2 and ADNI-3. Our main goal is to further investigate the pathological stage model of AD with a focus on functional properties, including effective connectivity, of the NbM and EC. From a more general point of view, our planned analyses have the potential to better understand the disease progression of AD which might help to further develop treatment strategies. |
| Investigator's Name: | Martin Göttlich |
| Proposed Analysis: | We would like to request data, collected by the Alzheimer’s Disease Neuroimaging Initiative (ADNI), to further investigate the pathological stage model of AD. Our main goal is to analyze resting state functional magnetic resonance imaging (rsfMRI) data in combination with markers of amyloid- and p-Tau as derived by cerebrospinal fluid (CSF). This approach is mainly based on a previous analysis of anatomical ADNI data and the corresponding publication by Fernández-Cabello et al. (2020). The pathological stage model of AD describes the process of neurodegeneration as a trans-synaptic spreading of pTau and amyloid- originating in the nucleus basalis of Meynert (NbM), projecting to the entorhinal cortex (EC) and then further to cortical targets. Since previous work (including Fernández-Cabello et al., 2020) mainly focused on structural analyses, it remains unclear how these structural changes relate to functional properties within the NbM and EC. Moreover, effective connectivity analyses, such as Dynamic Causal Modelling (DCM), allow for a more direct investigation of the predictions made by the pathological stage model of AD (Liu et al., 2008). Therefore, we intend to investigate rsfMRI properties of the NbM, EC and their connectivity changes on the basis of pTau and amyloid- properties. As such, our results are expected to give new insights into the underlying mechanisms and progression of AD. Requested Data: In order to achieve our goals, we would like to use data from ADNI-Go, ADNI-2 and ADNI-3. We are aware that ADNI-3 data collection is still in progress; however, we hope to begin our analyses using the baseline data and, at a later stage, proceed with the follow-up data. This procedure would provide the larges possible data-set and therefore robust and reliable results. Within each cohort, we kindly ask for: • rsfMRI data (including baseline and follow-up) • CSF biomarker of pTau and amyloid- • APOE4 genotye information (to be included in our models as covariate, see below) • neuropsychological assessment (including scores on memory performance, executive functions, Boston Naming test, the Clock Drawing tests, the six subcategories of the Clinical Dementia Rating and the demographic information). Planned analyses: Following Fernández-Cabello et al. (2020), participants will be assigned into one of two groups on the basis of their CSF ratio. While a CSF ratio < 0.028 represent a normal CSF (nCSF), a CSF ratio > 0.028 represents an abnormal CSF (aCSF) (Schindler et al., 2018). The rsfMRI data will be analysed at baseline, the two year follow-up, and in relation to one another (Fernández-Cabello et al., 2020). Here, the focus will be on the NbM and EC, which will be defined as regions of interest (ROI) using probabilistic maps (Amunts et al., 2005; Zaborszky et al., 2008). While pre-processing of the fMRI data is planned using SPM12, most subsequent analyses will be performed using in house scripts (see below). We plan to investigate (1) regionally specific and (2) interregional rsfMRI properties. (1) regionally specific properties will be based on: • Regional Homogeniety (ReHo), and • Low-frequency fluctuation (fALFF) analyses While ReHo describes the temporal coherence of BOLD time series within a cluster of adjacent voxels and represents the node’s importance of local functional connectivity (Jiang & Zuo, 2016), fALFF describes the ratio of local low-frequency range amplitudes (0.01-0.08 Hz) within a brain region relative to the entire frequency range (Zou et al., 2008). For both measures (ReHo and fALFF), we expect changes in NbM and EC depending on CSF markers and timepoint (Fernández-Cabello et al., 2020). Specifically, based on a 2x2x2 analyses of variance (ANOVA) with the factors group (aCSF, nCSF), region (NbM, EC) and time point (Baseline, follow-up), we expect reduced ReHo and fALFF for aCSF vs nCSF (main effect of group), and this effect should be more pronounced in the NbM as compared to the EC (interaction between group and regions). Moreover, we assume a decrease of ReHo and fALFF in NbM and EC over time (main effect of time), which should be more pronounced in aCSF compared to nCSF (interaction between time and regions). Finally, this time and group dependent reduction should be more pronounced in the NbM as compared to the EC (three-way interaction between group, time and region). To further pinpoint the directionality of the effects (NbMEC), we also plan to employ regression-based modelling to investigate if the pathological changes in the NbM and EC underlie interdependent processes (Fernández-Cabello et al., 2020). In both models (ANOVA and regression), we plan to include the factors CSF, genotyping (APOE4), neuropsychological scores and demographic information as covariates (Fernández-Cabello et al., 2020). Since ReHo and fALFF are different measures of regional connectivity, we will also explore which of them might serve as a better marker. This analysis will be exploratory in nature. In a subsequent step, (2) interregional properties will be investigated based on • Functional connectivity, and • Effective connectivity (Granger Causality and Dynamic Causal Modelling, DCM) For both types of connectivity analyses, we will use the NbM and EC as ROIs. For the functional connectivity analysis (Smitha et al., 2017), we expect the connectivity between both regions to be reduced in the aCSF vs. nCSF group at both time points (main effect group), and this effect should be more pronounced at the second time point for the aCSF group (interaction group x time point). Again, this analysis will be based on ANOVAs including both regions (NbM, EC) and groups (aCSF, nCSF), as well as the covariates demographic data, genotyping (APOE4), clinical diagnosis and neuropsychological scores. Moreover, we will run a moderation analysis to understand the impact of CSF on the functional connectivity between NbM and EC. Since functional connectivity does not allow conclusions regarding the direction of the effect, effective connectivity analyses will be performed. Here, we will apply Granger’s causality (Gates et al., 2010) and, as a second, complementary measure, DCM (Friston et al., 2011). Our main hypotheses are that NbM and EC are reciprocally connected, and this connectivity should change as a function of time and group. Specifically, connectivity strength NbMEC should be reduced in the aCSF group already at baseline and a reduction in NbMEC connectivity should be especially pronounced in the aCSF group at the second time point (follow up). If such a pattern would be observed, an additional analysis could include those parts of the temporo-parietal cortex, that have been reported in the structural analysis by Fernández-Cabello et al. (2020). Data are planned to be analyzed by the PI of the group, Prof. Dr. Nico Bunzeck, a PhD student Ms. Marthe Mieling, and Dr. Martin Göttlich (physicist). To summarize, we would like to receive rsfMRI data, CSF biomarkers, APOE4 information, and neuropsychological data from ADNI-Go, ADNI-2 and ADNI-3. Our main goal is to further investigate the pathological stage model of AD with a focus on functional properties, including effective connectivity, of the NbM and EC. From a more general point of view, our planned analyses have the potential to better understand the disease progression of AD which might help to further develop treatment strategies. |
| Investigator's Name: | Mushfa Yousuf |
| Proposed Analysis: | We would like to request data, collected by the Alzheimer’s Disease Neuroimaging Initiative (ADNI), to further investigate the pathological stage model of AD. Our main goal is to analyze resting state functional magnetic resonance imaging (rsfMRI) data in combination with markers of amyloid- and p-Tau as derived by cerebrospinal fluid (CSF). This approach is mainly based on a previous analysis of anatomical ADNI data and the corresponding publication by Fernández-Cabello et al. (2020). The pathological stage model of AD describes the process of neurodegeneration as a trans-synaptic spreading of pTau and amyloid- originating in the nucleus basalis of Meynert (NbM), projecting to the entorhinal cortex (EC) and then further to cortical targets. Since previous work (including Fernández-Cabello et al., 2020) mainly focused on structural analyses, it remains unclear how these structural changes relate to functional properties within the NbM and EC. Moreover, effective connectivity analyses, such as Dynamic Causal Modelling (DCM), allow for a more direct investigation of the predictions made by the pathological stage model of AD (Liu et al., 2008). Therefore, we intend to investigate rsfMRI properties of the NbM, EC and their connectivity changes on the basis of pTau and amyloid- properties. As such, our results are expected to give new insights into the underlying mechanisms and progression of AD. Requested Data: In order to achieve our goals, we would like to use data from ADNI-Go, ADNI-2 and ADNI-3. We are aware that ADNI-3 data collection is still in progress; however, we hope to begin our analyses using the baseline data and, at a later stage, proceed with the follow-up data. This procedure would provide the larges possible data-set and therefore robust and reliable results. Within each cohort, we kindly ask for: • rsfMRI data (including baseline and follow-up) • CSF biomarker of pTau and amyloid- • APOE4 genotye information (to be included in our models as covariate, see below) • neuropsychological assessment (including scores on memory performance, executive functions, Boston Naming test, the Clock Drawing tests, the six subcategories of the Clinical Dementia Rating and the demographic information). Planned analyses: Following Fernández-Cabello et al. (2020), participants will be assigned into one of two groups on the basis of their CSF ratio. While a CSF ratio < 0.028 represent a normal CSF (nCSF), a CSF ratio > 0.028 represents an abnormal CSF (aCSF) (Schindler et al., 2018). The rsfMRI data will be analysed at baseline, the two year follow-up, and in relation to one another (Fernández-Cabello et al., 2020). Here, the focus will be on the NbM and EC, which will be defined as regions of interest (ROI) using probabilistic maps (Amunts et al., 2005; Zaborszky et al., 2008). While pre-processing of the fMRI data is planned using SPM12, most subsequent analyses will be performed using in house scripts (see below). We plan to investigate (1) regionally specific and (2) interregional rsfMRI properties. (1) regionally specific properties will be based on: • Regional Homogeniety (ReHo), and • Low-frequency fluctuation (fALFF) analyses While ReHo describes the temporal coherence of BOLD time series within a cluster of adjacent voxels and represents the node’s importance of local functional connectivity (Jiang & Zuo, 2016), fALFF describes the ratio of local low-frequency range amplitudes (0.01-0.08 Hz) within a brain region relative to the entire frequency range (Zou et al., 2008). For both measures (ReHo and fALFF), we expect changes in NbM and EC depending on CSF markers and timepoint (Fernández-Cabello et al., 2020). Specifically, based on a 2x2x2 analyses of variance (ANOVA) with the factors group (aCSF, nCSF), region (NbM, EC) and time point (Baseline, follow-up), we expect reduced ReHo and fALFF for aCSF vs nCSF (main effect of group), and this effect should be more pronounced in the NbM as compared to the EC (interaction between group and regions). Moreover, we assume a decrease of ReHo and fALFF in NbM and EC over time (main effect of time), which should be more pronounced in aCSF compared to nCSF (interaction between time and regions). Finally, this time and group dependent reduction should be more pronounced in the NbM as compared to the EC (three-way interaction between group, time and region). To further pinpoint the directionality of the effects (NbMEC), we also plan to employ regression-based modelling to investigate if the pathological changes in the NbM and EC underlie interdependent processes (Fernández-Cabello et al., 2020). In both models (ANOVA and regression), we plan to include the factors CSF, genotyping (APOE4), neuropsychological scores and demographic information as covariates (Fernández-Cabello et al., 2020). Since ReHo and fALFF are different measures of regional connectivity, we will also explore which of them might serve as a better marker. This analysis will be exploratory in nature. In a subsequent step, (2) interregional properties will be investigated based on • Functional connectivity, and • Effective connectivity (Granger Causality and Dynamic Causal Modelling, DCM) For both types of connectivity analyses, we will use the NbM and EC as ROIs. For the functional connectivity analysis (Smitha et al., 2017), we expect the connectivity between both regions to be reduced in the aCSF vs. nCSF group at both time points (main effect group), and this effect should be more pronounced at the second time point for the aCSF group (interaction group x time point). Again, this analysis will be based on ANOVAs including both regions (NbM, EC) and groups (aCSF, nCSF), as well as the covariates demographic data, genotyping (APOE4), clinical diagnosis and neuropsychological scores. Moreover, we will run a moderation analysis to understand the impact of CSF on the functional connectivity between NbM and EC. Since functional connectivity does not allow conclusions regarding the direction of the effect, effective connectivity analyses will be performed. Here, we will apply Granger’s causality (Gates et al., 2010) and, as a second, complementary measure, DCM (Friston et al., 2011). Our main hypotheses are that NbM and EC are reciprocally connected, and this connectivity should change as a function of time and group. Specifically, connectivity strength NbMEC should be reduced in the aCSF group already at baseline and a reduction in NbMEC connectivity should be especially pronounced in the aCSF group at the second time point (follow up). If such a pattern would be observed, an additional analysis could include those parts of the temporo-parietal cortex, that have been reported in the structural analysis by Fernández-Cabello et al. (2020). Data are planned to be analyzed by the PI of the group, Prof. Dr. Nico Bunzeck, a PhD student Ms. Marthe Mieling, and Dr. Martin Göttlich (physicist). To summarize, we would like to receive rsfMRI data, CSF biomarkers, APOE4 information, and neuropsychological data from ADNI-Go, ADNI-2 and ADNI-3. Our main goal is to further investigate the pathological stage model of AD with a focus on functional properties, including effective connectivity, of the NbM and EC. From a more general point of view, our planned analyses have the potential to better understand the disease progression of AD which might help to further develop treatment strategies. |
| Investigator's Name: | Alexandra Korda |
| Proposed Analysis: | • Analysis 1: Abnormalities in brain gray matter (GM) topology comprising cortical surface areas and cortical thickness have been consistently observed in Alzheimer’s disease (AD). The GM topology1 is inherently complex and nonlinear dynamic analysis of these spatial data can support the research on structural biomarkers in AD. Converting images into sequences with time-series analysis tools has been used to solve various image data mining problems. The aim of the study is to employ the chaos analysis approach for the identification of specific brain topology changes related to the onset of AD beyond classical approaches based on structural MRI. We hypothesized that the structural complexity expressed by the chaos analysis a) differs in AD compared to HC and MCI and b) it is independent of the brain volume changes in AD patients. • Analysis 2: Texture analysis (TA) enables the quantification of the gray levels and brain patterns on MRI, via voxel inter-relations and spectral properties of the images. Common features of TA are second-order statistical features, such as entropy, contrast, and variance that express the heterogeneity of the brain by measuring the inter-relations between voxels. The method will be applied to non-segmented images reducing the computation complexity and the error associated with the segmentation process. We hypothesized that the TA feature can identify AD compared to HC and MCI. |
| Investigator's Name: | Seba Abdullatif |
| Proposed Analysis: | • Analysis 1: Abnormalities in brain gray matter (GM) topology comprising cortical surface areas and cortical thickness have been consistently observed in Alzheimer’s disease (AD). The GM topology1 is inherently complex and nonlinear dynamic analysis of these spatial data can support the research on structural biomarkers in AD. Converting images into sequences with time-series analysis tools has been used to solve various image data mining problems. The aim of the study is to employ the chaos analysis approach for the identification of specific brain topology changes related to the onset of AD beyond classical approaches based on structural MRI. We hypothesized that the structural complexity expressed by the chaos analysis a) differs in AD compared to HC and MCI and b) it is independent of the brain volume changes in AD patients. • Analysis 2: Texture analysis (TA) enables the quantification of the gray levels and brain patterns on MRI, via voxel inter-relations and spectral properties of the images. Common features of TA are second-order statistical features, such as entropy, contrast, and variance that express the heterogeneity of the brain by measuring the inter-relations between voxels. The method will be applied to non-segmented images reducing the computation complexity and the error associated with the segmentation process. We hypothesized that the TA feature can identify AD compared to HC and MCI. |
