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.

Background: Alzheimer’s disease (AD) is a common neurodegenerative disease in which the brain undergoes alterations for decades before symptoms become obvious. Subjective cognitive decline (SCD) have self-complain of persistent decline in cognitive function especially in memory but perform normally on standard neuropsychological tests. SCD with the presence of AD pathology is the transitional stage 2 of Alzheimer’s continuum, earlier than the prodromal stage, mild cognitive impairment (MCI), which seems to be the best target to research AD. In this study, we aimed to detect the transformational patterns of the intrinsic brain activity as the disease burden got heavy. Method: In this study, we enrolled 44 SCD, 55 amnestic MCI (aMCI), 47 AD dementia (d-AD) patients and 57 normal controls (NC) in total. A machine learning classification was utilized to detect identification accuracies between groups by using ALFF, fALFF, and fusing ALFF with fALFF features. Then, we measured the amplitude of the low-frequency fluctuation (ALFF) and fractional ALFF (fALFF) levels in three frequency bands (classic: 0.01–0.1 Hz; slow-5: 0.01–0.027 Hz; and slow-4: 0.027–0.073 Hz) and compared alterations in patients with NC. Results: In the machine learning verification, the identification accuracy of SCD, aMCI, d-AD from NC was higher when fused ALFF and fALFF features (76.44, 81.94, and 91.83%, respectively) than only using ALFF or fALFF features. Several brain regions showed significant differences in ALFF/fALFF within these bands among four groups: brain regions presented decreasing trend of values, including the Cingulum_Mid_R (aal), bilateral inferior cerebellum lobe, bilateral precuneus, and the Cingulum_Ant_R (aal); increasing trend of values were detected in the Hippocampus_L (aal), Frontal_Mid_Orb_R (aal), Frontal_Sup_R (aal) and Paracentral_Lobule_R (aal) as disease progressed. The normalized ALFF/fALFF values of these features were significantly correlated with the neuropsychological test scores. Conclusion: This study revealed gradual disturbances in intrinsic brain activity as the disease progressed: the normal objective performance in SCD may be dependent on compensation; as disease advanced, the cognitive function gradually impaired and decompensated in aMCI, severer in d-AD. Our results indicated that the ALFF and fALFF may help detect the underlying pathological mechanism in AD continuum. Clinical Trial Registration: ClinicalTrials.gov, identifier NCT02353884 and NCT02225964.

Alzheimer's disease (AD) has a preclinical phase that can last for decades prior to clinical dementia onset. Subjective cognitive decline (SCD) is regarded as the last preclinical AD stage prior to the development of amnestic mild cognitive decline (aMCI) and AD dementia (d-AD). The analysis of brain structural networks based on diffusion tensor imaging (DTI) has identified the so-called 'rich club', a set of cortical regions highly connected to each other, with other regions referred to as peripheral. It has been reported that rich club architecture is affected by regional atrophy and connectivity, which are reduced in patients with aMCI and d-AD.

Alzheimer’s disease (AD) is a frequently observed, irreversible brain function disorder among elderly individuals. Resting-state functional magnetic resonance imaging (rs-fMRI) has been introduced as an alternative approach to assessing brain functional abnormalities in AD patients. However, alterations in the brain rs-fMRI signal complexities in mild cognitive impairment (MCI) and AD patients remain unclear. Here, we described the novel application of permutation entropy (PE) to investigate the abnormal complexity of rs-fMRI signals in MCI and AD patients. The rs-fMRI signals of 30 normal controls (NCs), 33 early MCI (EMCI), 32 late MCI (LMCI) and 29 AD patients were obtained from the Alzheimer's disease Neuroimaging Initiative (ADNI) database. After preprocessing, whole-brain entropy maps of the four groups were extracted and subjected to Gaussian smoothing. We performed a one-way analysis of variance (ANOVA) on the brain entropy maps of the four groups. The results after adjusting for age and sex differences together revealed that the patients with AD exhibited lower complexity than did the MCI and NC controls. We found five clusters that exhibited significant differences and were distributed primarily in the occipital, frontal and temporal lobes. The average PE of the five clusters exhibited a decreasing trend from MCI to AD. The AD group exhibited the least complexity. Additionally, the average PE of the five clusters was significantly positively correlated with the Mini-Mental State Examination (MMSE) scores and significantly negatively correlated with Functional Assessment Questionnaire (FAQ) scores and global Clinical Dementia Rating (CDR) scores in the patient groups. Significant correlations were also found between the PE and regional homogeneity (ReHo) in the patient groups. These results indicated that declines in PE might be related to changes in regional functional homogeneity in AD. These findings suggested that complexity analyses using PE in rs-fMRI signals can provide important information about the fMRI characteristics of cognitive impairments in MCI and AD.

Background: Multiple sclerosis (MS) biomarker identification is important for pathogenesis research and diagnosis in routine clinical practice. Cerebrospinal fluid (CSF) and blood cytokines as potential biomarkers that can inform MS pathogenesis, diagnosis and response to treatment have been assessed in numerous studies. However, there have been no comprehensive meta-analyses to pool cytokine data and to address their diagnostic performance. We systematically reviewed literature with meta-analyses to assess the alteration levels of cytokines and chemokines in MS. Methods: We searched PubMed and Web of Science for articles published between January 1, 1990 and April 30, 2018 for this systematic review and meta-analysis. Data were extracted from 226 included studies encompassing 13,526 MS patients and 8,428 controls. Biomarker performance was rated by a random-effects meta-analysis based on the standard mean difference between cytokine concentration in patients with MS and controls, or patients before and after treatments. Results: Of the 26 CSF cytokines and 37 blood cytokines for potential differentiation between MS patients and controls, the random-effects meta-analysis showed that 13 CSF cytokines and 21 blood cytokines were significantly increased in MS patients in comparison to the controls. Interestingly, TNF-α, CXCL8, IL-15, IL-12p40, and CXCL13 were increased in both blood and CSF of MS patients. For those cytokines analyzed in at least 10 studies, differentiation between case and control was strong for CSF CXCL13, blood IL-2R, and blood IL-23; CSF CXCL8, blood IL-2, and blood IL-17 also performed well in differentiating between MS patients and controls, whereas those of CSF TNF-α and blood TNF-α, CXCL8, IL-12, IFN-γ were moderate. Furthermore, CSF IL-15, CCL19, CCL11, CCL-3, and blood CCL20, IL-12p40, IL-21, IL-17F, IL-22 had large effective sizes when differentiating between MS patients and controls but had a relatively small number of studies (three to seven studies). Conclusion: Our findings clarified the circulating cytokine profile in MS, which provide targets for disease modifying treatments, and suggest that cytokines have the potential to be used as biomarkers for MS.