SfN2019文章列表及点评（排名不分先后）
在我心目中，脑机接口的top顶会，就是EMBS和SfN。当然，如果细分领域，如算法，还有NeuralPIS（但是NPIS更吸引眼球的是人工智能，当然脑机接口可以算其中一部分，毕竟是混合智能，即生物智能和人工智能结合）等。这些顶会是大佬们展示肌肉的重要时刻，也是领域内科研工作者交流想法的重要机会。从成果列表中，我们也能看出接下来几年，这个方向的科研热点，如果在这些顶会上没有你所研究的课题方向，建议，个人建议，还是换方向吧，不然到时找同行评议都找不到大佬级别的人，毕竟大佬们手握各大顶刊的评议权，落不到他们手上，就不要指望文章出现在那些顶刊上来。只希望于混迹各类水刊的同学（如我～）可以忽略。大家看以下内容时重点读一下题目即可（请忽略我很随意的中英文叙述）。
-
Simultaneous exploration and exploitation of neural strategies during neuroprosthetic learning
基于的背景是previous work has shown that BMIs can be “two-learner” systems, in which both neurons and decoders co-adapt over the course of learning. -
A novel bi-directional implantable brain machine interface with versatile stimulation capabilities
从刺激电流的大小、频率，还有刺激位点进行来option，从而versatile。The Brain Interchange system developed by the authors generates current-controlled, charge balanced pulses with amplitude of up to 6 mA and pulse widths of up to 2.5 ms. The compliance voltages available for driving the current are asymmetric -12 V as negative and +6 V as positive rail. The current can be directed to any of its 32 electrode contacts the system also uses for brain recording. The current return electrode can be selected to also be any of the 32 electrode contacts or subsets of them, or an additional counter electrode. These options make the system very versatile to use. -
In vivo characterization of the stability of cortical response to electrical microstimulation
个人认为这个研究是个很重要的基础工作，研究设备是荧光显微镜和双光子成像
While efficacious, it is unclear how the cortex dynamically responds to electrical stimulation over time, and how different electrode materials differ in effectiveness and safety. Traditionally, In vivo characterizations of stimulation performance have been largely based on electrophysiological, behavioral, or end point histological outputs.
This stimulation was applied on day 1 and weekly thereafter. -
Neural dynamics in primary motor cortex during BCI calibration in a person with tetraplegia
这篇文章是BCI领域大神们的工作，看内容没什么特别，最主要是在人颅内植入来两片犹他电击。记住这些大神的名字就行：J.E. Downey; S.J. Bensmaia; J.L. Collinger; J.M. Goodman; A.K. Suresh，以后见到他们的名字一般都是Nature级的文章。 -
Temporally-precise vagus nerve stimulation increases motor learning of a dexterous reach task through cholinergic neuromodulation
别的不说，就记住这篇文章的行为学是由deepcut完成的，另外，stimulate的parameters是30hz, 100us pulse width, 0.6mA, 0.5s train -
Neural dynamics underlying generalization in motor cortex
这篇是BCI领域大神J.M. Carmena实验室的作品，值得期待后续全文。
文章的科学问题是Are the neural activity patterns used to generate a specific motor command re-used when the command appears in a new behavioral sequence?
实验范式 a center-out task composed of straight movements and an obstacle avoidance task composed of curved movements
结果：in motor cortex, a neural activity pattern generating a motor command is not a static map that is re-used in the middle of diverse neural and behavioral trajectories, but rather that motor cortex flexibly generates neural activity patterns for the same motor commands in a way that leverages its high dimensional dynamics. -
Exploring the neural basis underlying locomotor activities and leg reaching in macaque motor, premotor, and sensory cortex
终于迎来COURTINE大神实验室的作品了～～～
不出意外，在neuroscience方面继续发力（以前比较偏重neurotechology），他们是真的换了细分的方向了。
背景比较low，（通俗点就是以前的行为难度低，所以研究不出啥来，这不是打脸自己那么多篇nature吗？）：However, previous studies have primarily been conducted over motorized treadmills, a relatively simple and automated locomotor task. Consequently, cortical dynamics underlying the production of basic and skilled locomotion remains unclear.
然后做了些复杂设计：To address this question, we trained Macaca fascicularis monkeys to walk on a moving treadmill at different speeds, along corridors, uneven ladders, obstacles, and stairs. We also trained them to perform a center-out reach and grasp task with the leg. Finally, we introduced a novel, untrained task requiring object avoidance during stepping.
脑电和肌电一起记录：Three animals were implanted with 48-channel intracortical arrays in the leg motor (M1) and premotor (PMd) cortices to record multi-unit activity, in conjunction with 14 pairs of electrodes in leg muscles to record electromyographic (EMG) activity. One animal was additionally implanted with a 64-channel array in the leg sensory (S1) cortex.
怎么评价呢？如果这个势头不扭转，后续应该会上马光基因、神经迁移手术等等措施，摊子越铺越大，还让我们这些小虾米?活不活，我们只能喝点粥了。当然，很期待Courtine大神后续spinal cord injury，然后利用BCI来stimulation，应该能再刷几篇nature。 -
Experimental framework for the assessment of the efficacy of spinal cord stimulation to restore voluntary arm movement in monkeys after spinal cord injury
依然是Courtine大神组的成果，这次是来抢forelimb的生意。
背景：However, direct muscle stimulation induces rapid muscle fatigue, limiting the applicability for functional three-dimensional arm reaching movements that are critical for activities of daily living.
提出的解决方法：epidural electrical stimulation (EES) of the spinal cord can elicit large muscle forces with reduced fatigue by recruiting motoneurons pre-synaptically through the activation of proprioceptive afferents. -
Dynamical encoding of swing and stance phases as orthogonal states in hindlimb motor cortex
还是Courtine大神组的成果，这次比较规规矩矩
Here we apply dynamical system analysis to show that hindlimb motor cortex encodes the swing and stance phases of the gait cycle as two distinct states.
This means swing and stance are encoded by two distinct patterns of neural activity that do not overlap in time. Each pattern fades in and out before transitioning to the next, as opposed to smoothly transitioning between the two. Therefore the brain enters a unique state during the swing phase, then switches states during the stance phase.
Results also show that the swing phase is encoded in a localized region of hindlimb primary motor cortex, while the stance phase is more distributed. 这个说法还是第一次见，可以留意下。 -
Neurotechnologies for the restoration of three-dimensional arm movements after cervical spinal cord injury
哎，继续Courtine大神，他们组是要霸占SfN了！！！
这篇跟第8貌似类似的工作，区别在哪呢？we performed a partial hemisection of the spinal cord at the C6 level and evaluated the effects of alternated EES bursts timed with reaching (arm extension) and grasp (hand flexion) on arm/hand kinematics. -
Artificial somatosensory kinaesthetic feedback of arm movements
这篇是出自Graz的成果，别的不管，来自的Graz就是招牌！
The main focus of neuroprosthesis research naturally lies on restoring the forward part of the feedback loop, i.e. control, while the loop is usually closed via visual feedback.
方法：tactors driven by a custom device are arranged in a sparse grid attached to a custom shirt in the area of the shoulder blade, and their intensities are modulated such that the stimulation is perceived as a smoothly moving sensation
结果：Average peak amplitudes of motor-related cortical potentials were found to be higher in the feedback condition, and lowest in the condition with static stimulation. -
A neural-machine interface for control of a lower-limb prosthesis
背景： The incorporation of neural signals, specifically muscle and brain, may offer a viable method for improved volitional control by directly interpreting signals from muscle (measured through electromyography—EMG) and cortical brain activations (measured through electroencephalography—EEG).
方法：多模态，(1) identify neural correlates of movement during isolated limb movements and walking in the amputee population, and (2) demonstrate the feasibility of control of a powered LL prosthesis using neural signal from EEG and EMG.
Transfemoral amputees were recruited to perform a series of isolated limb movements of the intact and phantom knee and ankle while undergoing a functional magnetic resonance imaging (fMRI) scans. The subjects subsequently repeated the experiment while instrumented with EEG, EMG, and motion capture. fMRI data were used to extract regions of interest during individual limb movements, and a systematic EEG processing approach was implemented to reduce artifacts (e.g., eye blinks/movements, muscle artifacts) and estimate the underlying source activity. We investigate the temporal and spectral dynamics associated with activation of the phantom limb as compared to the intact limb. Secondly, we investigate the ability to leverage EEG signals for control of a lower limb prosthesis and how this may supplement control schemes utilizing neural signals from the muscles of the residual limb. -
Non invasive kinesthetic feedback in a bidirectional hand system for upper limb amputees
Micera大神来了！这篇沿续了他课题组的强势方向。肌肉震动提供反馈，具体就不说了。 -
Optimal post-injury time window to apply activity-dependent stimulation to drive functional recovery after traumatic brain injury in the rat
Cleveland的工作，必属精品，继续rehabilitation的强势方向。
方法：We hypothesized that delaying the initiation of ADS by 1, 2 or 3 weeks after an M1 injury will lead to improved skilled reaching performance compared to non-stimulated controls.
结果：Behavioral recovery was assessed weekly using a skilled reaching task. Preliminary results indicate that the application of ADS, regardless of treatment delay, leads to significant improvement in skilled reaching performance compared to non-stimulated controls. The results demonstrate that delaying the application of ADS does not impair the ability to bridge communication between spared cortical areas in the sensorimotor network and can drive functional behavioral recovery following a TBI.
换句话说，就是脑机接口在任何时候开始用上，进行训练康复，效果都差不多？值得怀疑。期待他们的正式结果吧，到时再研读。 -
Multiscale spike-field network causality identification during a motor task
这个方法不知道好不好使，得学会，毕竟文中说比Granger Causality强！找到个IEEE NSRE的原作者近似题目的文章，我会拜读一下。Pubmed：30932842 -
Dynamical characteristics of simultaneously-recorded spike and LFP activities underlying movement
背景：Investigating low-dimensional neural dynamics underlying movements has so far focused on a single scale of brain activity, in particular neuronal spikes. However, movements are encoded across multiple spatial and temporal scales of neural activity, from spikes to local field potentials. Thus, examining the dynamical characteristics of spike-field network activity is important in understanding the neural mechanisms underlying movements.
方法：Here we developed multiscale dynamical models and a modal analysis to investigate the low-dimensional dynamical characteristics of simultaneously-recorded spike and LFP activities from non-human primates performing a 3D reach-to-grasp task. Our modal analysis allowed us to dissociate the different components of neural dynamics each with a distinct dynamical characteristic. We refer to these components as dynamical modes.
结果：We found that while spike and LFP networks had some distinct dynamical modes, they had a strongly present shared mode. This mode was the dominant mode in predicting the movement. In addition, these shared dynamical characteristics were preserved across different experimental sessions and monkeys. Further, these shared modes were not a replicate of the modes in the behavior.
我个人更喜欢这篇文章的工作，相比于上面那篇搞算法的，毕竟有神经生理学的科学发现 -
Adaptation of neural activity in parietal cortex of rhesus monkey during control of 3D reaches in a brain-computer interface
背景：Posterior parietal cortex (PPC) is known to integrate visual and proprioceptive sensory information
e.g. about the position of the own hand during reaching. In a joystick task with curved trajectories, activity in PPC was previously shown to correlate better with the immediate (zero time-lag) cursor direction than with its previous positions (Mulliken et al. 2008), arguing against mere sensory encoding. Such findings support the idea that PPC contributes to internal forward model computations and real-time state estimation for online motor control: it could combine expected with actual sensory feedback to form an inner belief of how motor commands affect the state of the body.
问题和方法：From previous studies it is unclear if neural encoding is more compatible with the idea of an efference copy, the predicted sensory feedback (visual or proprioceptive) or a state estimation, since in manual task these parameters are typically confounded. We here used a brain-computer interface (BCI) to study motor control with the cursor motion (the motor command) being fully experimentally controlled and proprioceptive feedback being held constant.没什么特别的，这些都是脑机接口常见的，只是换了角度，从功能角度来反推工程。
结果：也不出意外的是和以往的研究一致，PPC是用于视觉和动作实时在线整合（矫正、决策）的区域， -
Behavioral assessment of sensory percepts induced by dorsal thoracic epidural spinal cord stimulation in primates
用Nicolelis大神的作品来结束吧，其实BCI方面在SfN的论文还是很多的，大家有兴趣可以看看。
Lack of somatosensory feedback is a major obstacle to optimally integrating neural prostheses into an individual’s body schema. Electrical stimulation of the somatosensory cortex and thalamic nuclei have been proposed to generate sensory percepts in individuals with loss of sensorimotor function; however, these targets require open intradural cranial surgery and accurate targeting of tiny brain structures. It is plausible that alternative target structures along the somatosensory pathway might be capable of encoding meaningful sensory percepts induced by electrical stimulation in a less invasive and more readily clinically translatable fashion. Previously, our group demonstrated that stimulation of the dorsal surface of the spinal cord in rats can be used to transmit sensory information to the brain, and that rats learn to discriminate temporally varying patterns of dorsal column stimulation. In the present study, we generated sensory percepts using percutaneous leads implanted on the dorsal surface of the spinal cord in 3 rhesus macaques, using a routine procedure that is performed in over 50,000 human patients annually and is FDA-approved for certain pain-control indications. Three monkeys were trained to move a joystick-controlled cursor into one of two visual targets. After basic training on the joystick task, they were implanted with epidural percutaneous leads on the dorsal surface of the spinal cord, and externalized leads connected to a custom external pulse generator. Monkeys had to hold a cursor inside a center target during a brief preparatory period while electrical stimulation was presented. Using a two-alternative forced choice task, monkeys were trained to detect artificial stimuli generated by electrical signals delivered at the lead contacts in a charge-balanced bipolar biphasic manner. Using psychometric analysis, we then studied the behavioral sensitivity to detection of sensory percepts at various stimulation parameters such as frequency, pulse width, and duration. We observed that the current threshold for detection of sensory percepts significantly decreased with an increase in frequency, or pulse-width, or duration of stimulation. We also showed that monkeys were able to discriminate between two stimuli that varied in stimulation frequency as well as those that varied in spatial location. Our results suggest that thoracic epidural spinal cord stimulation can be used to provide valuable information from a prosthetic device back to the central nervous system through a non-visual route, and that the relationship between sensory percepts and stimulation parameters can be explored to create discriminable percepts.
1565
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
