Multi-task Collaborative Pre-training and Adaptive Token Selection: A Unified Framework for Brain Representation Learning.

Structural magnetic resonance imaging (sMRI) reveals the structural organization of the brain. Learning general brain representations from sMRI is an enduring topic in neuroscience. Previous deep learning models neglect that the brain, as the core of cognition, is distinct from other organs whose primary attribute is anatomy. Capturing the high-level representation associated with inter-individual cognitive variability is key to appropriately represent the brain. Given that this cognition-related information is subtle, mixed, and distributed in the brain structure, sMRI-based models need to both capture fine-grained details and understand how they relate to the overall global structure. Additionally, it is also necessary to explicitly express the cognitive information that implicitly embedded in local-global image features. Therefore, we propose MCPATS, a brain representation learning framework that combines Multi-task Collaborative Pre-training (MCP) and Adaptive Token Selection (ATS). First, we develop MCP, including mask-reconstruction to understand global context, distort-restoration to capture fine-grained local details, adversarial learning to integrate features at different granularities, and age-prediction, using age as a surrogate for cognition to explicitly encode cognition-related information from local-global image features. This co-training allows progressive learning of implicit and explicit cognition-related representations. Then, we develop ATS based on mutual attention for downstream use of the learned representation. During fine-tuning, the ATS highlights discriminative features and reduces the impact of irrelevant information. MCPATS was validated on three different public datasets for brain disease diagnosis, outperforming competing methods and achieving accurate diagnosis. Further, we performed detailed analysis to confirm that the MCPATS-learned representation captures cognition-related information.

Brain fingerprinting and cognitive behavior predicting using functional connectome of high inter-subject variability.

The functional connectivity (FC) graph of the brain has been widely recognized as a ``fingerprint'' that can be used to identify individuals from a group of subjects. Research has indicated that individual identification accuracy can be improved by eliminating the impact of shared information among individuals. However, current research extracts not only shared information of inter-subject but also individual-specific information from FC graphs, resulting in incomplete separation of shared information and fingerprint information among individuals, leading to lower individual identification accuracy across all functional magnetic resonance imaging (fMRI) states session pairs and poor cognitive behavior prediction performance. In this paper, we propose a method to enhance inter-subject variability combining conditional variational autoencoder (CVAE) network and sparse dictionary learning (SDL) module. By embedding fMRI state information in the encoding and decoding processes, the CVAE network can better capture and represent the common features among individuals and enhance inter-subject variability by residual. Our experimental results on Human Connectome Project (HCP) data show that the refined connectomes obtained by using CVAE with SDL can accurately distinguish an individual from the remaining participants. The success accuracies reached 99.7 % and 99.6 % in the session pair rest1-rest2 and reverse rest2-rest1, respectively. In the identification experiment involving task-task combinations carried out on the same day, the identification accuracies ranged from 94.2 % to 98.8 %. Furthermore, we showed the Frontoparietal and Default networks make the most significant contributions to individual identification and the edges that significantly contribute to individual identification are found within and between the Frontoparietal and Default networks. Additionally, high-level cognitive behaviors can also be better predicted with the obtained refined connectomes, suggesting that higher fingerprinting can be useful for resulting in higher behavioral associations. In summary, our proposed framework provides a promising approach to use functional connectivity networks for studying cognition and behavior, promoting a deeper understanding of brain functions.

Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip.

By mimicking the neurons and synapses of the human brain and employing spiking neural networks on neuromorphic chips, neuromorphic computing offers a promising energy-efficient machine intelligence. How to borrow high-level brain dynamic mechanisms to help neuromorphic computing achieve energy advantages is a fundamental issue. This work presents an application-oriented algorithm-software-hardware co-designed neuromorphic system for this issue. First, we design and fabricate an asynchronous chip called "Speck", a sensing-computing neuromorphic system on chip. With the low processor resting power of 0.42mW, Speck can satisfy the hardware requirements of dynamic computing: no-input consumes no energy. Second, we uncover the "dynamic imbalance" in spiking neural networks and develop an attention-based framework for achieving the algorithmic requirements of dynamic computing: varied inputs consume energy with large variance. Together, we demonstrate a neuromorphic system with real-time power as low as 0.70mW. This work exhibits the promising potentials of neuromorphic computing with its asynchronous event-driven, sparse, and dynamic nature.

Brain-Controlled Hand Exoskeleton Based on Augmented Reality-Fused Stimulus Paradigm.

Advancements in brain-machine interfaces (BMIs) have led to the development of novel rehabilitation training methods for people with impaired hand function. However, contemporary hand exoskeleton systems predominantly adopt passive control methods, leading to low system performance. In this work, an active brain-controlled hand exoskeleton system is proposed that uses a novel augmented reality-fused stimulus (AR-FS) paradigm as a human-machine interface, which enables users to actively control their fingers to move. Considering that the proposed AR-FS paradigm generates movement artifacts during hand movements, an enhanced decoding algorithm is designed to improve the decoding accuracy and robustness of the system. In online experiments, participants performed online control tasks using the proposed system, with an average task time cost of 16.27 s, an average output latency of 1.54 s, and an average correlation instantaneous rate (CIR) of 0.0321. The proposed system shows 35.37% better efficiency, 8.03% reduced system delay, and 35.28% better stability than the traditional system. This study not only provides an efficient rehabilitation solution for people with impaired hand function but also expands the application prospects of brain-control technology in areas such as human augmentation, patient monitoring, and remote robotic interaction. The video in Graphical Abstract Video demonstrates the user's process of operating the proposed brain-controlled hand exoskeleton system.

Connectional-style-guided contextual representation learning for brain disease diagnosis.

Structural magnetic resonance imaging (sMRI) has shown great clinical value and has been widely used in deep learning (DL) based computer-aided brain disease diagnosis. Previous DL-based approaches focused on local shapes and textures in brain sMRI that may be significant only within a particular domain. The learned representations are likely to contain spurious information and have poor generalization ability in other diseases and datasets. To facilitate capturing meaningful and robust features, it is necessary to first comprehensively understand the intrinsic pattern of the brain that is not restricted within a single data/task domain. Considering that the brain is a complex connectome of interlinked neurons, the connectional properties in the brain have strong biological significance, which is shared across multiple domains and covers most pathological information. In this work, we propose a connectional style contextual representation learning model (CS-CRL) to capture the intrinsic pattern of the brain, used for multiple brain disease diagnosis. Specifically, it has a vision transformer (ViT) encoder and leverages mask reconstruction as the proxy task and Gram matrices to guide the representation of connectional information. It facilitates the capture of global context and the aggregation of features with biological plausibility. The results indicate that CS-CRL achieves superior accuracy in multiple brain disease diagnosis tasks across six datasets and three diseases and outperforms state-of-the-art models. Furthermore, we demonstrate that CS-CRL captures more brain-network-like properties, and better aggregates features, is easier to optimize, and is more robust to noise, which explains its superiority in theory.

Using a deep generation network reveals neuroanatomical specificity in hemispheres.

Asymmetry is an important property of brain organization, but its nature is still poorly understood. Capturing the neuroanatomical components specific to each hemisphere facilitates the understanding of the establishment of brain asymmetry. Since deep generative networks (DGNs) have powerful inference and recovery capabilities, we use one hemisphere to predict the opposite hemisphere by training the DGNs, which automatically fit the built-in dependencies between the left and right hemispheres. After training, the reconstructed images approximate the homologous components in the hemisphere. We use the difference between the actual and reconstructed hemispheres to measure hemisphere-specific components due to asymmetric expression of environmental and genetic factors. The results show that our model is biologically plausible and that our proposed metric of hemispheric specialization is reliable, representing a wide range of individual variation. Together, this work provides promising tools for exploring brain asymmetry and new insights into self-supervised DGNs for representing the brain.

Using unsupervised capsule neural network reveal spatial representations in the human brain.

Humans can extract high-level spatial features from visual signals, but spatial representations in the brain are complex and remain unclear. The unsupervised capsule neural network (U-CapsNet) is sensitive to the spatial location and relationship of the object, contains a special recurrent mechanism and uses a self-supervised generation strategy to represent images, which is similar to the computational principle in the human brain. Therefore, we hypothesized that U-CapsNet can help us understand how the human brain processes spatial information. First, brain activities were studied using functional magnetic resonance imaging during spatial working memory in which participants had to remember the locations of circles for a short time. Then, U-CapsNet served as a computational model of the brain to perform tasks that are identical to those performed by humans. Finally, the representational models were used to compare the U-CapsNet with the brain. The results showed that some human-defined spatial features naturally emerged in the latent space of U-CapsNet. Moreover, representations in U-CapsNet captured the response structure of two types of brain regions during different activity patterns, as well as important factors associated with human behavior. Together, our study not only provides a computationally feasible framework for modeling how the human brain encodes spatial features but also provides insights into the representational format and goals of the human brain.

Beta-informativeness-diffusion multilayer graph embedding for brain network analysis.

Brain network analysis provides essential insights into the diagnosis of brain disease. Integrating multiple neuroimaging modalities has been demonstrated to be more effective than using a single modality for brain network analysis. However, a majority of existing brain network analysis methods based on multiple modalities often overlook both complementary information and unique characteristics from various modalities. To tackle this issue, we propose the Beta-Informativeness-Diffusion Multilayer Graph Embedding (BID-MGE) method. The proposed method seamlessly integrates structural connectivity (SC) and functional connectivity (FC) to learn more comprehensive information for diagnosing neuropsychiatric disorders. Specifically, a novel beta distribution mapping function (beta mapping) is utilized to increase vital information and weaken insignificant connections. The refined information helps the diffusion process concentrate on crucial brain regions to capture more discriminative features. To maximize the preservation of the unique characteristics of each modality, we design an optimal scale multilayer brain network, the inter-layer connections of which depend on node informativeness. Then, a multilayer informativeness diffusion is proposed to capture complementary information and unique characteristics from various modalities and generate node representations by incorporating the features of each node with those of their connected nodes. Finally, the node representations are reconfigured using principal component analysis (PCA), and cosine distances are calculated with reference to multiple templates for statistical analysis and classification. We implement the proposed method for brain network analysis of neuropsychiatric disorders. The results indicate that our method effectively identifies crucial brain regions associated with diseases, providing valuable insights into the pathology of the disease, and surpasses other advanced methods in classification performance.

Functional and structural gradients reveal atypical hierarchical organization of Parkinson's disease.

Parkinson's disease (PD) patients exhibit deficits in primary sensorimotor and higher-order executive functions. The gradient reflects the functional spectrum in sensorimotor-associated areas of the brain. We aimed to determine whether the gradient is disrupted in PD patients and how this disruption is associated with treatment outcome. Seventy-six patients (mean age, 59.2 ± 12.4 years [standard deviation], 44 women) and 34 controls participants (mean age, 58.1 ± 10.0 years [standard deviation], 19 women) were evaluated. We explored functional and structural gradients in PD patients and control participants. Patients were followed during 2 weeks of multidisciplinary intensive rehabilitation therapy (MIRT). The Unified Parkinson's Disease Rating Scale Part III (UPDRS-III) was administered to patients before and after treatment. We investigated PD-related alterations in the principal functional and structural gradients. We further used a support vector machine (SVM) and correlation analysis to assess the classification ability and treatment outcomes related to PD gradient alterations, respectively. The gradients showed significant differences between patients and control participants, mainly in somatosensory and visual networks involved in primary function, and higher-level association networks (dorsal attentional network (DAN) and default mode network (DMN)) related to motor control and execution. On the basis of the combined functional and structural gradient features of these networks, the SVM achieved an accuracy of 91.2% in discriminating patients from control participants. Treatment reduced the gradient difference. The altered gradient exhibited a significant correlation with motor improvement and was mainly distributed across the visual network, DAN and DMN. This study revealed damage to gradients in the brain characterized by sensorimotor and executive control deficits in PD patients. The application of gradient features to neurological disorders could lead to the development of potential diagnostic and treatment markers for PD.

Metal-non-tolerant ecotypes of ectomycorrhizal fungi can protect plants from cadmium pollution.

The application of mycorrhizal fungi as a bioaugmentation technology for phytoremediation of heavy metal (HM) contaminated soil has attracted widespread attention. In order to explore whether the adaptation of Pinus massoniana (P. massoniana) to metal polluted soil depends on the metal adaptation potential of their associated ectomycorrhizal fungi (ECMF), we evaluated the cadmium (Cd) tolerance of 10 ecotypes of Cenococcum geophilum (C. geophilum) through a membership function method, and P. massoniana seedlings were not (NM) or inoculated by Cd non-tolerant type (JaCg144), low-tolerant (JaCg32, JaCg151) and high-tolerant (JaCg205) isolates of C. geophilum were exposed to 0 and 100 mg·kg-1 for 3 months. The result showed that, each ecotype of C. geophilum significantly promoted the growth, photosynthesis and chlorophyll content, proline (Pro) content and the activity of peroxidase (POD) of P. massoniana seedlings, and decreased malonaldehyde (MDA) content and catalase (CAT) and superoxide dismutase (SOD) activity. The comprehensive evaluation D value of the tolerance to Cd stress showed that the order of the displaced Cd resistance of the four ecotypic mycorrhizal P. massoniana was: JaCg144 > JaCg151 > JaCg32 > JaCg205. Pearson correlation analysis showed that the Sig. value of the comprehensive evaluation (D) values of the strains and mycorrhizal seedlings was 0.077 > 0.05, indicating that the Cd tolerance of the the C. geophilum isolates did not affect its regulatory effect on the Cd tolerance of the host plant. JaCg144 and JaCg151 which are non-tolerant and low-tolerant ecotype significantly increased the Cd content in the shoots and roots by about 136.64-181.75% and 153.75-162.35%, indicating that JaCg144 and JaCg151 were able to effectively increase the enrichment of Cd from the soil to the root. Transcriptome results confirmed that C. geophilum increased the P. massoniana tolerance to Cd stress through promoting antioxidant enzyme activity, photosynthesis, and lipid and carbohydrate synthesis metabolism. The present study suggests that mental-non-tolerant ecotypes of ECMF can protect plants from Cd pollution, providing more feasible strategies for ectomycorrhizal-assisted phytoremediation.

Neurovascular coupling dysfunction of visual network organization in Parkinson's disease.

Parkinson's disease (PD) has been showed perfusion and neural activity alterations in specific regions, such as the motor and visual networks; however, the clinical significance of coupling changes is still unknown. To identify how neurovascular coupling changes during the pathophysiology of PD, patients and healthy controls underwent multiparametric magnetic resonance imaging to measure neural activity organization of segregation and integration using amplitude of low-frequency fluctuation (ALFF) and functional connectivity strength (FCS), and measure vascular responses using cerebral blood flow (CBF). Neurovascular coupling was calculated as the global CBF-ALFF and CBF-FCS coupling and the regional CBF/ALFF and CBF/FCS ratio. Correlations and dynamic causal modeling was then used to evaluate relationships with disease-alterations to clinical variables and information flow. Neurovascular coupling was impaired in PD with decreased global CBF-ALFF and CBF-FCS coupling, as well as decreased CBF/ALFF in the parieto-occipital cortex (dorsal visual stream) and CBF/FCS in the temporo-occipital cortex (ventral visual stream); these decouplings were associated with motor and non-motor impairments. The distinctive patterns of neurovascular coupling alterations within the dorsal and ventral visual streams of the visual system could potentially provide additional understanding into the pathophysiological mechanisms of PD.

Functional connectivity in people at clinical and familial high risk for schizophrenia.

Patients diagnosed with schizophrenia (SZ) exhibit compromised functional connectivity within extensive brain networks. However, the precise development of this impairment during disease progression in the clinical high-risk (CHR) population and their relatives remains unclear. Our study leveraged data from 128 resting electroencephalography (EEG) channels acquired from 30 SZ patients, 21 CHR individuals, 17 unaffected healthy relatives (RSs; those at heightened SZ risk due to family history), and 31 healthy controls (HCs). These data were harnessed to establish functional connectivity patterns. By calculating the geometric distance between EEG sequences, we unveiled local and global nonlinear relationships within the entire brain. The process of dimensionality reduction led to low-dimensional representations, providing insights into high-dimensional EEG data. Our findings indicated that CHR participants exhibited aberrant functional connectivity across hemispheres, whereas RS individuals showcased anomalies primarily concentrated within hemispheres. In the realm of low-dimensional analysis, RS participants' third-dimensional occipital lobe values lay between those of the CHR individuals and HCs, significantly correlating with scale scores. This low-dimensional approach facilitated the visualization of brain states, potentially offering enhanced comprehension of brain structure, function, and early-stage functional impairment, such as occipital visual deficits, in RS individuals before cognitive decline onset.

Investigating the heterogeneity within the somatosensory-motor network and its relationship with the attention and default systems.

The somatosensory-motor network (SMN) not only plays an important role in primary somatosensory and motor processing but is also central to many disorders. However, the SMN heterogeneity related to higher-order systems still remains unclear. Here, we investigated SMN heterogeneity from multiple perspectives. To characterize the SMN substructures in more detail, we used ultra-high-field functional MRI to delineate a finer-grained cortical parcellation containing 430 parcels that is more homogenous than the state-of-the-art parcellation. We personalized the new parcellation to account for individual differences and identified multiscale individual-specific brain structures. We found that the SMN subnetworks showed distinct resting-state functional connectivity (RSFC) patterns. The Hand subnetwork was central within the SMN and exhibited stronger RSFC with the attention systems than the other subnetworks, whereas the Tongue subnetwork exhibited stronger RSFC with the default systems. This two-fold differentiation was observed in the temporal ordering patterns within the SMN. Furthermore, we characterized how the distinct attention and default streams were carried forward into the functions of the SMN using dynamic causal modeling and identified two behavioral domains associated with this SMN fractionation using meta-analytic tools. Overall, our findings provided important insights into the heterogeneous SMN organization at the system level and suggested that the Hand subnetwork may be preferentially involved in exogenous processes, whereas the Tongue subnetwork may be more important in endogenous processes.

Individual prefrontal neurons contribute to sensory-to-motor information transformation by rotating reference frames during spatial working memory performance.

Performing working memory tasks correctly requires not only the temporary maintenance of information but also the visual-to-motor transformation of information. Although sustained delay-period activity is known to be a mechanism for temporarily maintaining information, the mechanism for information transformation is not well known. An analysis using a population of delay-period activities recorded from prefrontal neurons visualized a gradual change of maintained information from sensory to motor as the delay period progressed. However, the contributions of individual prefrontal neurons to this process are not known. In the present study, we used a version of the delayed-response task, in which monkeys needed to make a saccade 90o clockwise from a visual cue after a 3-s delay, and examined the temporal change in the preferred directions of delay-period activity during the delay period for individual neurons. One group of prefrontal neurons encoded the cue direction by a retinotopic reference frame and either maintained it throughout the delay period or rotated it 90o counterclockwise to adjust visual information to saccade information, whereas other groups of neurons encoded the cue direction by a saccade-based reference frame and rotated it 90o clockwise. The results indicate that visual-to-motor information transformation is achieved by manipulating the reference frame to adjust visual coordinates to motor coordinates.

Evidence of a large current of transcranial alternating current stimulation directly to deep brain regions.

Deep brain regions such as hippocampus, insula, and amygdala are involved in neuropsychiatric disorders, including chronic insomnia and depression. Our recent reports showed that transcranial alternating current stimulation (tACS) with a current of 15 mA and a frequency of 77.5 Hz, delivered through a montage of the forehead and both mastoids was safe and effective in intervening chronic insomnia and depression over 8 weeks. However, there is no physical evidence to support whether a large alternating current of 15 mA in tACS can send electrical currents to deep brain tissue in awake humans. Here, we directly recorded local field potentials (LFPs) in the hippocampus, insula and amygdala at different current strengths (1 to 15 mA) in 11 adult patients with drug-resistant epilepsy implanted with stereoelectroencephalography (SEEG) electrodes who received tACS at 77.5 Hz from 1 mA to 15 mA at 77.5 Hz for five minutes at each current for a total of 40 min. For the current of 15 mA at 77.5 Hz, additional 55 min were applied to add up a total of 60 min. Linear regression analysis revealed that the average LFPs for the remaining contacts on both sides of the hippocampus, insula, and amygdala of each patient were statistically associated with the given currents in each patient (p < 0.05-0.01), except for the left insula of one subject (p = 0.053). Alternating currents greater than 7 mA were required to produce significant differences in LFPs in the three brain regions compared to LFPs at 0 mA (p < 0.05). The differences remained significant after adjusting for multiple comparisons (p < 0.05). Our study provides direct evidence that the specific tACS procedures are capable of delivering electrical currents to deep brain tissues, opening a realistic avenue for modulating or treating neuropsychiatric disorders associated with hippocampus, insula, and amygdala.

Effects of Microstate Dynamic Brain Network Disruption in Different Stages of Schizophrenia.

Schizophrenia is a heterogeneous mental disorder with unknown etiology or pathological characteristics. Microstate analysis of the electroencephalogram (EEG) signal has shown significant potential value for clinical research. Importantly, significant changes in microstate-specific parameters have been extensively reported; however, these studies have ignored the information interactions within the microstate network in different stages of schizophrenia. Based on recent findings, since rich information about the functional organization of the brain can be revealed by functional connectivity dynamics, we use the first-order autoregressive model to construct the functional connectivity of intra- and intermicrostate networks to identify information interactions among microstate networks. We demonstrate that, beyond abnormal parameters, disrupted organization of the microstate networks plays a crucial role in different stages of the disease by 128-channel EEG data collected from individuals with first-episode schizophrenia, ultrahigh-risk, familial high-risk, and healthy controls. According to the characteristics of the microstates of patients at different stages, the parameters of microstate class A are reduced, those of class C are increased, and the transitions from intra- to intermicrostate functional connectivity are gradually disrupted. Furthermore, decreased integration of intermicrostate information might lead to cognitive deficits in individuals with schizophrenia and those in high-risk states. Taken together, these findings illustrate that the dynamic functional connectivity of intra- and intermicrostate networks captures more components of disease pathophysiology. Our work sheds new light on the characterization of dynamic functional brain networks based on EEG signals and provides a new interpretation of aberrant brain function in different stages of schizophrenia from the perspective of microstates.

Ultrasound-induced cell disintegration and its ultrastructure characterization for the valorisation of Chlorella pyrenoidosa protein.

Chlorella pyrenoidosa (CP) has great potential for feeding future demands in food, environment, energy, and pharmaceuticals. To achieve this goal, the exploitation of emerging efficient technique such as ultrasound-assisted extraction (UAE) for CP nutrient enrichment is crucial. Here, UAE is deployed for high-efficient CP protein (CPP) valorisation. Compared to conventional solvent extraction (CSE), remarkable mass transfer enhancements with 9-time protein yields and 3-time extraction rate are achieved by ultrasonic cavitation in UAE, indicating UAE can drastically shift intracellular nutrients including proteins and pigments to solvent. Cell morphology and ultrastructure show the different responses of cell wall and membrane, indicating that the cell membrane may play a role in the extraction process, based on which the extremely-low efficiency of CSE and high efficiency of UAE are highlighted. This study provides a solution for future food crisis by extracting CPP and may open a new discussion field in ultrasonic extraction.

Individual-level functional connectomes predict the motor symptoms of Parkinson's disease.

Abnormalities in functional connectivity networks are associated with sensorimotor networks in Parkinson's disease (PD) based on group-level mapping studies, but these results are controversial. Using individual-level cortical segmentation to construct individual brain atlases can supplement the individual information covered by group-level cortical segmentation. Functional connectivity analyses at the individual level are helpful for obtaining clinically useful markers and predicting treatment response. Based on the functional connectivity of individualized regions of interest, a support vector regression model was trained to estimate the severity of motor symptoms for each subject, and a correlation analysis between the estimated scores and clinical symptom scores was performed. Forty-six PD patients aged 50-75 years were included from the Parkinson's Progression Markers Initiative database, and 63 PD patients were included from the Beijing Rehabilitation Hospital database. Only patients below Hoehn and Yahr stage III were included. The analysis showed that the severity of motor symptoms could be estimated by the individualized functional connectivity between the visual network and sensorimotor network in early-stage disease. The results reveal individual-level connectivity biomarkers related to motor symptoms and emphasize the importance of individual differences in the prediction of the treatment response of PD.

Altered Neurovascular Coupling for Multidisciplinary Intensive Rehabilitation in Parkinson's Disease.

Effective rehabilitation in Parkinson's disease (PD) is related to brain reorganization with restoration of cortico-subcortical networks and compensation of frontoparietal networks; however, further neural rehabilitation evidence from a multidimensional perspective is needed. To investigate how multidisciplinary intensive rehabilitation treatment affects neurovascular coupling, 31 PD patients (20 female) before and after treatment and 30 healthy controls (17 female) underwent blood oxygenation level-dependent functional magnetic resonance imaging and arterial spin labeling scans. Cerebral blood flow (CBF) was used to measure perfusion, and fractional amplitude of low-frequency fluctuation (fALFF) was used to measure neural activity. The global CBF-fALFF correlation and regional CBF/fALFF ratio were calculated as neurovascular coupling. Dynamic causal modeling (DCM) was used to evaluate treatment-related alterations in the strength and directionality of information flow. Treatment reduced CBF-fALFF correlations. The altered CBF/fALFF exhibited increases in the left angular gyrus and the right inferior parietal gyrus and decreases in the bilateral thalamus and the right superior frontal gyrus. The CBF/fALFF alteration in right superior frontal gyrus showed correlations with motor improvement. Further, DCM indicated increases in connectivity from the superior frontal gyrus and decreases from the thalamus to the inferior parietal gyrus. The benefits of rehabilitation were reflected in the dual mechanism, with restoration of executive control occurring in the initial phase of motor learning and compensation of information integration occurring in the latter phase. These findings may yield multimodal insights into the role of rehabilitation in disease modification and identify the dorsolateral superior frontal gyrus as a potential target for noninvasive neuromodulation in PD.SIGNIFICANCE STATEMENT Although rehabilitation has been proposed as a promising supplemental treatment for PD as it results in brain reorganization, restoring cortico-subcortical networks and eliciting compensatory activation of frontoparietal networks, further multimodal evidence of the neural mechanisms underlying rehabilitation is needed. We measured the ratio of perfusion and neural activity derived from arterial spin labeling and blood oxygenation level-dependent fMRI data and found that benefits of rehabilitation seem to be related to the dual mechanism, restoring executive control in the initial phase of motor learning and compensating for information integration in the latter phase. We also identified the dorsolateral superior frontal gyrus as a potential target for noninvasive neuromodulation in PD patients.