Never-Ending Learning for Explainable Brain Computing.

Exploring the nature of human intelligence and behavior is a longstanding pursuit in cognitive neuroscience, driven by the accumulation of knowledge, information, and data across various studies. However, achieving a unified and transparent interpretation of findings presents formidable challenges. In response, an explainable brain computing framework is proposed that employs the never-ending learning paradigm, integrating evidence combination and fusion computing within a Knowledge-Information-Data (KID) architecture. The framework supports continuous brain cognition investigation, utilizing joint knowledge-driven forward inference and data-driven reverse inference, bolstered by the pre-trained language modeling techniques and the human-in-the-loop mechanisms. In particular, it incorporates internal evidence learning through multi-task functional neuroimaging analyses and external evidence learning via topic modeling of published neuroimaging studies, all of which involve human interactions at different stages. Based on two case studies, the intricate uncertainty surrounding brain localization in human reasoning is revealed. The present study also highlights the potential of systematization to advance explainable brain computing, offering a finer-grained understanding of brain activity patterns related to human intelligence.

Impulsivity-related right superior frontal gyrus as a biomarker of internet gaming disorder.

Background: Internet gaming disorder (IGD) is a mental health issue that affects individuals worldwide. However, the lack of knowledge about the biomarkers associated with the development of IGD has restricted the diagnosis and treatment of this disorder. Aims: We aimed to reveal the biomarkers associated with the development of IGD through resting-state brain network analysis and provide clues for the diagnosis and treatment of IGD. Methods: Twenty-six patients with IGD, 23 excessive internet game users (EIUs) who recurrently played internet games but were not diagnosed with IGD and 29 healthy controls (HCs) performed delay discounting task (DDT) and Iowa gambling task (IGT). Resting-state functional magnetic resonance imaging (fMRI) data were also collected. Results: Patients with IGD exhibited significantly lower hubness in the right medial orbital part of the superior frontal gyrus (ORBsupmed) than both the EIU and the HC groups. Additionally, the hubness of the right ORBsupmed was found to be positively correlated with the highest excessive internet gaming degree during the past year in the EIU group but not the IGD group; this might be the protective mechanism that prevents EIUs from becoming addicted to internet games. Moreover, the hubness of the right ORBsupmed was found to be related to the treatment outcome of patients with IGD, with higher hubness of this region indicating better recovery when undergoing forced abstinence. Further modelling analysis of the DDT and IGT showed that patients with IGD displayed higher impulsivity during the decision-making process, and impulsivity-related parameters were negatively correlated with the hubness of right ORBsupmed. Conclusions: Our findings revealed that the impulsivity-related right ORBsupmed hubness could serve as a potential biomarker of IGD and provide clues for the diagnosis and treatment of IGD.

BNLoop-GAN: a multi-loop generative adversarial model on brain network learning to classify Alzheimer's disease.

Recent advancements in AI, big data analytics, and magnetic resonance imaging (MRI) have revolutionized the study of brain diseases such as Alzheimer's Disease (AD). However, most AI models used for neuroimaging classification tasks have limitations in their learning strategies, that is batch training without the incremental learning capability. To address such limitations, the systematic Brain Informatics methodology is reconsidered to realize evidence combination and fusion computing with multi-modal neuroimaging data through continuous learning. Specifically, we introduce the BNLoop-GAN (Loop-based Generative Adversarial Network for Brain Network) model, utilizing multiple techniques such as conditional generation, patch-based discrimination, and Wasserstein gradient penalty to learn the implicit distribution of brain networks. Moreover, a multiple-loop-learning algorithm is developed to combine evidence with better sample contribution ranking during training processes. The effectiveness of our approach is demonstrated through a case study on the classification of individuals with AD and healthy control groups using various experimental design strategies and multi-modal brain networks. The BNLoop-GAN model with multi-modal brain networks and multiple-loop-learning can improve classification performance.

The influence of perceptual and semantic chunking on the neural mechanism of remote association.

Studies on remote association tests (RATs) have mainly focused on cognitive processes involved in searching for remote associations. However, factors affecting these search processes and remote associations remain unclear. In order to address this issue, this study compared non-chunking condition (e.g., "//") with perceptual chunking (two red color characters in the three-character item "//") and semantic chunking (high-frequency word-pair in the item; e.g., "//", "," literally "philosophy") conditions in the Chinese Remote Association Test (CRAT). The behavioral results on the CRAT found that the semantic ones resulted in significantly lower successful solving rates and longer response times than the other two conditions. The event-related potential (ERP) results showed that in contrast to the perceptual-chunking and the non-chunking condition, the semantic-chunking elicited enhanced P200, which might be related to the intuitive awareness of the mental fixation. However, relative to the non-chunking condition, the two chunking conditions evoked increased N400 and late positive component (LPC), indexing the late reflection and implementation of cognitive control. Our results suggest that it is the early awareness of the semantic chunk, rather than the general cognitive control process involved in representing and solving the semantically and perceptually chunk-induced interferences, that critically determines the final solving of RATs.

Shorter TR combined with finer atlas positively modulate topological organization of brain network: A resting state fMRI study.

BACKGROUND: The use of shorter TR and finer atlases in rs-fMRI can provide greater detail on brain function and anatomy. However, there is limited understanding of the effect of this combination on brain network properties. METHODS: A study was conducted with 20 healthy young volunteers who underwent rs-fMRI scans with both shorter (0.5s) and long (2s) TR. Two atlases with different degrees of granularity (90 vs 200 regions) were used to extract rs-fMRI signals. Several network metrics, including small-worldness, Cp, Lp, Eloc, and Eg, were calculated. Two-factor ANOVA and two-sample t-tests were conducted for both the single spectrum and five sub-frequency bands. RESULTS: The network constructed using the combination of shorter TR and finer atlas showed significant enhancements in Cp, Eloc, and Eg, as well as reductions in Lp and γ in both the single spectrum and subspectrum (p < 0.05, Bonferroni correction). Network properties in the 0.082-0.1 Hz frequency range were weaker than those in the 0.01-0.082 Hz range. CONCLUSION: Our findings suggest that the use of shorter TR and finer atlas can positively affect the topological characteristics of brain networks. These insights can inform the development of brain network construction methods.

Reduced loss aversion in value-based decision-making and edge-centric functional connectivity in patients with internet gaming disorder.

Background and aims: Impaired value-based decision-making is a feature of substance and behavioral addictions. Loss aversion is a core of value-based decision-making and its alteration plays an important role in addiction. However, few studies explored it in internet gaming disorder patients (IGD). Methods: In this study, IGD patients (PIGD) and healthy controls (Con-PIGD) performed the Iowa gambling task (IGT), under functional magnetic resonance imaging (fMRI). We investigated group differences in loss aversion, brain functional networks of node-centric functional connectivity (nFC) and the overlapping community features of edge-centric functional connectivity (eFC) in IGT. Results: PIGD performed worse with lower average net score in IGT. The computational model results showed that PIGD significantly reduced loss aversion. There was no group difference in nFC. However, there were significant group differences in the overlapping community features of eFC1. Furthermore, in Con-PIGD, loss aversion was positively correlated with the edge community profile similarity of the edge2 between left IFG and right hippocampus at right caudate. This relationship was suppressed by response consistency3 in PIGD. In addition, reduced loss aversion was negatively correlated with the promoted bottom-to-up neuromodulation from the right hippocampus to the left IFG in PIGD. Discussion and conclusions: The reduced loss aversion in value-based decision making and their related edge-centric functional connectivity support that the IGD showed the same value-based decision-making deficit as the substance use and other behavioral addictive disorders. These findings may have important significance for understanding the definition and mechanism of IGD in the future.

Internet gaming disorder and tobacco use disorder share neural connectivity patterns between the subcortical and the motor network.

Internet gaming disorder (IGD) and tobacco use disorder (TUD) are globally common, non-substance-related disorders and substance-related disorders worldwide, respectively. Recognizing the commonalities between IGD and TUD will deepen understanding of the underlying mechanisms of addictive behavior and excessive online gaming. Using node strength, 141 resting-state data were collected in this study to compute network homogeneity. The participants included participants with IGD (PIGD: n = 34, male = 29, age: 15-25 years), participants with TUD (PTUD: n = 33, male = 33, age: 19-42 years), and matched healthy controls (control-for-IGD: n = 41, male = 38, age: 17-32 years; control-for-TUD: n = 33, age: 21-27 years). PIGD and PTUD exhibited common enhanced node strength between the subcortical and motor networks. Additionally, a common enhanced resting-state functional connectivity (RSFC) was found between the right thalamus and right postcentral gyrus in PIGD and PTUD. Node strength and RSFC were used to distinguish PIGD and PTUD from their respective healthy controls. Interestingly, models trained on PIGD versus controls could classify PTUD versus controls and vice versa, suggesting that these disorders share common neurological patterns. Enhanced connectivity may indicate a greater association between rewards and behaviors, inducing addiction behaviors without flexible and complex regulation. This study discovered that the connectivity between the subcortical and motor networks is a potential biological target for developing addiction treatment in the future.

Perceived danger associated with a property modulates cross category generalization.

Identifying projectable predicates is a key issue in understanding inductive inference. It is proposed that looking into the evolutionary psychology literature for adaptive properties may be one useful approach. One hypothesis that emerges from this literature is that properties that signal danger or harm should be more salient than properties that do not. Two studies are carried out to test this hypothesis. In study 1 participants were presented with a scenario involving the discovery of novel animals, for which there was incomplete information. Three types of properties (a harmful property, a neutral property, a beneficial property) were associated with animals in one (base) category and participants were asked to indicate strength of generalization of the property to a target within the category, and to a target across to another category. In the second experiment, the procedure was repeated, but in addition, subjects were also explicitly asked to indicate whether the base and target belonged to the same or different categories. Study 1 showed that the harmful property was more projectable compared to the beneficial and neutral properties. Study 2 reconfirmed this and further showed that it also promoted excessive generalization across categories. The results suggest that examination of adaptations identified by evolutionary psychologists may be a useful source of insight in the study of inductive inference. Supplementary Information: The online version contains supplementary material available at 10.1007/s11571-022-09793-3.

Systematic Fusion of Multi-Source Cognitive Networks With Graph Learning - A Study on Fronto-Parietal Network.

Cognitive tasks induce fluctuations in the functional connectivity between brain regions which constitute cognitive networks in the human brain. Although several cognitive networks have been identified, consensus still cannot be achieved on the precise borders and distribution of involved brain regions for each network, due to the multifarious use of diverse brain atlases in different studies. To address the problem, the current study proposed a novel approach to generate a fused cognitive network with the optimal performance in discriminating cognitive states by using graph learning, following the synthesization of one cognitive network defined by different brain atlases, and the construction of a hierarchical framework comprised of one main version and other supplementary versions of the specific cognitive network. As a result, the proposed method demonstrated better results compared with other machine learning methods for recognizing cognitive states, which was revealed by analyzing an fMRI dataset related to the mental arithmetic task. Our findings suggest that the fused cognitive network provides the potential to develop new mind decoding approaches.

Improving segmentation reliability of multi-scanner brain images using a generative adversarial network.

Background: Magnetic resonance (MR) images generated by different scanners generally have inconsistent contrast properties, making it difficult to perform a combined quantitative analysis of images from a range of scanners. In this study, we aimed to develop an automatic brain image segmentation model to provide a more reliable analysis of MR images taken with different scanners. Methods: The spatially localized atlas network tiles-27 (SLANT-27) deep learning model was used to train the automatic segmentation module, based on a multi-center dataset of 1,917 three-dimensional (3D) T1-weighted MR images. Subsequently, a framework called Qbrain, consisting of a new generative adversarial network (GAN) image transfer module and the SLANT-27 segmentation module, was developed. Another 3D T1-weighted MRI interscan dataset of 48 participants who were scanned in 3 MRI scanners (1.5T Siemens Avanto, 3T Siemens Trio Tim, and 3T Philips Ingenia) on the same day was used to train and test the Qbrain model. Volumetric T1-weighted images were processed with Qbrain, SLANT-27, and FreeSurfer (FS). The automatic segmentation reliability across the scanners was assessed using test-retest variability (TRV). Results: The reproducibility of different segmentation methods across scanners showed a consistent trend in the greater reliability and robustness of QBrain compared to SLANT-27 which, in turn, showed greater reliability and robustness compared to FS. Furthermore, when the GAN image transfer module was added, the mean segmentation error of the TRV of the 3T Siemens vs. 1.5T Siemens, the 3T Philips vs. 1.5T Siemens, and the 3T Siemens vs. 3T Philips scanners was reduced by 1.57%, 2.01%, and 0.56%, respectively. In addition, the segmentation model improved intra-scanner variability (0.9-1.67%) compared with that of FS (2.47-4.32%). Conclusions: The newly developed QBrain method combined with GAN image transfer module and a SLANT-27 segmentation module was shown to improve the reliability of whole-brain automatic structural segmentation results across multiple scanners, thus representing a suitable alternative quantitative method of comparative brain tissue analysis for individual patients.

Involvement of the dorsal and ventral attention networks in visual attention span.

Visual attention span (VAS), which refers to the window size of multielement parallel processing in a short time, plays an important role in higher-level cognition (e.g., reading) as required by encoding large amounts of information input. However, it is still a matter of debate about the underlying neural mechanism of VAS. In the present study, a modified visual 1-back task was designed by using nonverbal stimuli and nonverbal responses, in which possible influences of target presence and position were considered to identify more pure VAS processing. A task-driven functional magnetic resonance imaging (fMRI) experiment was then performed, and 30 healthy adults participated in this study. Results of confirmatory and exploratory analyses consistently revealed that both dorsal attention network (DAN) and ventral attention network (VAN) were significantly activated during this visual simultaneous processing. In particular, more significant activation in the left superior parietal lobule (LSPL), as compared to that in the bilateral inferior frontal gyrus (IFGs), suggested a greater involvement of DAN in VAS-related processing in contrast to VAN. In addition, it was also found that the activation in temporoparietal junctions (TPJs) were suppressed during multielement processing only in the target-absent condition. The current results suggested the recruitment of LSPL in covert attentional shifts and top-down control of VAS resources distribution during the rapid visual simultaneous processing, as well as the involvement of bilateral IFGs (especially RIFG) in both VAS processing and inhibitory control. The present findings might bring some enlightenments for diagnosis of the atypicality of attentional disorders and reading difficulties.

The Effect of Light Sedation with Midazolam on Functional Connectivity of the Dorsal Attention Network.

Altered connectivity within and between the resting-state networks (RSNs) brought about by anesthetics that induce altered consciousness remains incompletely understood. It is known that the dorsal attention network (DAN) and its anticorrelations with other RSNs have been implicated in consciousness. However, the role of DAN-related functional patterns in drug-induced sedative effects is less clear. In the current study, we investigated altered functional connectivity of the DAN during midazolam-induced light sedation. In a placebo-controlled and within-subjects experimental study, fourteen healthy volunteers received midazolam or saline with a 1-week interval. Resting-state fMRI data were acquired before and after intravenous drug administration. A multiple region of interest-driven analysis was employed to investigate connectivity within and between RSNs. It was found that functional connectivity was significantly decreased by midazolam injection in two regions located in the left inferior parietal lobule and the left middle temporal area within the DAN as compared with the saline condition. We also identified three clusters in anticorrelation between the DAN and other RSNs for the interaction effect, which included the left medial prefrontal cortex, the right superior temporal gyrus, and the right superior frontal gyrus. Connectivity between all regions and DAN was significantly decreased by midazolam injection. The sensorimotor network was minimally affected. Midazolam decreased functional connectivity of the dorsal attention network. These findings advance the understanding of the neural mechanism of sedation, and such functional patterns might have clinical implications in other medical conditions related to patients with cognitive impairment.

ERP Characteristics of Inducing Rule Validity in Number Series Under Time Pressure.

A great deal of research has been devoted to examining the neural mechanisms of inductive reasoning. However, the influences of rule validity and time pressure on numerical inductive reasoning remain unclear. In the current study, we aimed to examine the effects of these variables on the time course of rule identification in numerical inductive reasoning. We designed a 3 (task type: valid, invalid, and anomalous) × 2 (time pressure: with time pressure and without time pressure) within-subject experiment based on electroencephalographic event-related potentials (ERP). Behaviorally, we found significant effects of rule validity and time pressure on rule identification. Neurologically, we considered the elicited N200 ERP to reflect conflict detection and found it to be modulated by rule validity but not time pressure. We considered the induced P300 ERP to be primarily related to updating working memory, affected by both rule validity and time pressure. These findings have new implications for better understanding dynamic information processing within numerical inductive reasoning.

Exploring the Relationship between Gray and White Matter in Healthy Adults: A Hybrid Research of Cortical Reconstruction and Tractography.

The gray matter (GM) and white matter (WM) are structurally and functionally related in the human brain. Among the numerous neuroimaging studies, yet only a few have investigated these two structures in the same sample. So, there is limited and inconsistent information about how they are correlated in the brain of healthy adults. In this study, we combined cortical reconstruction with diffusion spectrum imaging (DSI) tractography to investigate the relationship between cortical morphology and microstructural properties of major WM tracts in 163 healthy young adults. The results showed that cortical thickness (CTh) was positively correlated with the coherent tract-wise fractional anisotropy (FA) value, and the correlation was stronger in the dorsal areas than in the ventral areas. For other diffusion parameters, CTh was positively correlated with axial diffusivity (AD) of coherent fibers in the frontal areas and negatively correlated with radial diffusivity (RD) of coherent fibers in the dorsal areas. These findings suggest that the correlation between GM and WM is inhomogeneity and could be interpreted with different mechanisms in different brain regions. We hope our research could provide new insights into the studies of diseases in which the GM and WM are both affected.

Neural correlates of improved inductive reasoning ability in abacus-trained children: A resting state fMRI study.

Abacus-based mental calculation (AMC) training may improve mathematics-related abilities and transfer to other cognitive domains. Thus, it was hypothesized that inductive reasoning abilities can be improved by AMC training given the overlapping cognitive processes and neural correlates between AMC and inductive reasoning. The aim of the current study was to examine the underlying neurobiological mechanisms of this possible adaption by resting-state functional magnetic resonance imaging (rs-fMRI). Sixty-three children were randomly assigned to either the AMC-trained or the nontrained group. The AMC-trained group was required to perform abacus training for 2 hours per week for 5 years whereas the nontrained group was not required to perform any abacus training. Each participant's rs-fMRI data were collected after abacus training, and regional homogeneity (ReHo) analysis was performed to determine the neural activity differences between groups. The participants' posttraining mathematical ability, intelligence quotients, and inductive reasoning ability were recorded and evaluated. The results revealed that AMC-trained children exhibited a significantly higher mathematical ability and inductive reasoning performance and higher ReHo in the rostrolateral prefrontal cortex (RLPFC) compared to the nontrained group. In particular, the increased ReHo in the RLPFC was found to be positively correlated with improved inductive reasoning performance. Our findings suggest that rs-fMRI may reflect the modulation of training in task-related networks.

Evaluating the association between brain atrophy, hypometabolism, and cognitive decline in Alzheimer's disease: a PET/MRI study.

Glucose metabolism reduction and brain volume losses are widely reported in Alzheimer's disease (AD). Considering that neuroimaging changes in the hippocampus and default mode network (DMN) are promising important candidate biomarkers and have been included in the research criteria for the diagnosis of AD, it is hypothesized that atrophy and metabolic changes of the abovementioned regions could be evaluated concurrently to fully explore the neural mechanisms underlying cognitive impairment in AD. Twenty-three AD patients and Twenty-four age-, sex- and education level-matched normal controls underwent a clinical interview, a detailed neuropsychological assessment and a simultaneous 18F-fluoro-2-deoxy-D-glucose positron emission tomography (18F-FDG PET)/high-resolution T1-weighted magnetic resonance imaging (MRI) scan on a hybrid GE SIGNA PET/MR scanner. Brain volume and glucose metabolism were examined in patients and controls to reveal group differences. Multiple linear regression models were employed to explore the relationship between multiple imaging features and cognitive performance in AD. The AD group had significantly reduced volume in the hippocampus and DMN regions (P < 0.001) relative to that of normal controls determined by using ROI analysis. Compared to normal controls, significantly decreased metabolism in the DMN (P < 0.001) was also found in AD patients, which still survived after controlling for gray matter atrophy (P < 0.001). These findings from ROI analysis were further confirmed by whole-brain confirmatory analysis (P < 0.001, FWE-corrected). Finally, multiple linear regression results showed that impairment of multiple cognitive tasks was significantly correlated with the combination of DMN hypometabolism and atrophy in the hippocampus and DMN regions. This study demonstrated that combining functional and structural features can better explain the cognitive decline of AD patients than unimodal FDG or brain volume changes alone. These findings may have important implications for understanding the neural mechanisms of cognitive decline in AD.

Constructing Connectome Atlas by Graph Laplacian Learning.

The recent development of neuroimaging technology and network theory allows us to visualize and characterize the whole-brain functional connectivity in vivo. The importance of conventional structural image atlas widely used in population-based neuroimaging studies has been well verified. Similarly, a "common" brain connectivity map (also called connectome atlas) across individuals can open a new pathway to interpreting disorder-related brain cognition and behaviors. However, the main obstacle of applying the classic image atlas construction approaches to the connectome data is that a regular data structure (such as a grid) in such methods breaks down the intrinsic geometry of the network connectivity derived from the irregular data domain (in the setting of a graph). To tackle this hurdle, we first embed the brain network into a set of graph signals in the Euclidean space via the diffusion mapping technique. Furthermore, we cast the problem of connectome atlas construction into a novel learning-based graph inference model. It can be constructed by iterating the following processes: (1) align all individual brain networks to a common space spanned by the graph spectrum bases of the latent common network, and (2) learn graph Laplacian of the common network that is in consensus with all aligned brain networks. We have evaluated our novel method for connectome atlas construction in comparison with non-learning-based counterparts. Based on experiments using network connectivity data from populations with neurodegenerative and neuropediatric disorders, our approach has demonstrated statistically meaningful improvement over existing methods.

Reproducibility of volume and asymmetry measurements of hippocampus, amygdala, and entorhinal cortex on traveling volunteers: a multisite MP2RAGE prospective study.

BACKGROUND: Multisite studies can considerably increase the pool of normally aging individuals with neurodegenerative disorders and thereby expedite the associated research. Understanding the reproducibility of the parameters of related brain structures-including the hippocampus, amygdala, and entorhinal cortex-in multisite studies is crucial in determining the impact of healthy aging or neurodegenerative diseases. PURPOSE: To estimate the reproducibility of the fascinating structures by automatic (FreeSurfer) and manual segmentation methods in a well-controlled multisite dataset. MATERIAL AND METHODS: Three traveling individuals were scanned at 10 sites, which were equipped with the same equipment (3T Prisma Siemens). They used the same scan protocol (two inversion-contrast magnetization-prepared rapid gradient echo sequences) and operators. Validity coefficients (intraclass correlations coefficient [ICC]) and spatial overlap measures (Dice Similarity Coefficient [DSC]) were used to estimate the reproducibility of multisite data. RESULTS: ICC and DSC values varied substantially among structures and segmentation methods, and values of manual tracing were relatively higher than the automated method. ICC and DSC values of structural parameters were greater than 0.80 and 0.60 across sites, as determined by manual tracing. Low reproducibility was observed in the amygdala parameters by automatic segmentation method (ICC = 0.349-0.529, DSC = 0.380-0.873). However, ICC and DSC scores of the hippocampus were higher than 0.60 and 0.65 by two segmentation methods. CONCLUSION: This study suggests that a well-controlled multisite study could provide a reliable MRI dataset. Manual tracing of volume assessments is recommended for low reproducibility structures that require high levels of precision in multisite studies.

Neuropsychopharmacological effects of midazolam on the human brain.

As a commonly used anesthetic agent, midazolam has the properties of water-soluble, rapid onset, and short duration of action. With the rapid development in the field of neuroimaging, numerous studies have investigated how midazolam acts on the human brain to induce the alteration of consciousness. However, the neural bases of midazolam-induced sedation or anesthesia remain beginning to be understood in detail. In this review, we summarize findings from neuroimaging studies that have used midazolam to study altered consciousness at different levels and content. We also compare the results to those of neuroimaging studies using diverse anesthetic agents and describe the common neural correlates of anesthetic-induced alteration of consciousness.

Altered multimodal magnetic resonance parameters of basal nucleus of Meynert in Alzheimer's disease.

OBJECTIVES: We aimed to examine how gray matter volume (GMV), regional blood flow (rCBF), and resting-state functional connectivity (FC) of the basal nucleus of Meynert (BNM) are altered in 40 patients with AD, relative to 30 healthy controls (HCs). METHODS: We defined the BNM on the basis of a mask histochemically reconstructed from postmortem human brains. We examined GMV with voxel-based morphometry of high-resolution structural images, rCBF with arterial spin labeling imaging, and whole-brain FC with published routines. We performed partial correlations to explore how the imaging metrics related to cognitive and living status in patients with AD. Further, we employed receiver operating characteristic analysis to compute the "diagnostic" accuracy of these imaging markers. RESULTS: AD relative to HC showed lower GMV and higher rCBF of the BNM as well as lower BNM connectivity with the right insula and cerebellum. In addition, the GMVs of BNM were correlated with cognitive and daily living status in AD. Finally, these imaging markers predicted AD (vs. HC) with an accuracy (area under the curve) of 0.70 to 0.86. Combination of BNM metrics provided the best prediction accuracy. CONCLUSIONS: By combining multimode MR imaging, we demonstrated volumetric atrophy, hyperperfusion, and disconnection of the BNM in AD. These findings support cholinergic dysfunction as an etiological marker of AD and related dementia.