Neural Spike Sorting Using Unsupervised Adversarial Learning

Unsupervised spike sorting, a vital processing step in real-time brain-implantable microsystems, is faced with the prominent challenge of managing nonstationarity in neural signals. In long-term recordings, spike waveforms gradually change and new source neurons are likely to become activated. Adaptive spike sorters combined with on-implant training units effectively process the nonstationary signals at the cost of high hardware resource utilization. On the other hand, static approaches, while being hardware-friendly, are subjected to decreased processing performance in such recordings where the neural signal characteristics gradually change. To strike a balance between the hardware cost and processing performance, this study proposes a hardware-efficient novelty-aware spike sorting approach that is capable of dealing with both variated spike waveforms and spike waveforms generated from new source neurons. Its improved hardware efficiency compared to adaptive ones and capability of dealing with nonstationary signals make it attractive for implantable applications. The proposed novelty-aware spike sorting especially would be a good fit for brain-computer interfaces where long-term, real-time interaction with the brain is required, and the available on-implant hardware resources are limited. Our unsupervised spike sorting benefits from a novelty detection process to deal with neural signal variations. It tracks the spike features so that in case of detecting an unexpected change (novelty detection) both on and off-implant parameters are updated to preserve the performance in new state. To make the proposed approach agile enough to be suitable for brain implants, the on-implant computations are reduced while the computational burden is realized off-implant. The performance of our proposed approach is evaluated using both synthetic and real datasets. The results demonstrate that, in the mean, it is capable of detecting 94.31% of novel spikes (wave-drifted or emerged spikes) with a classification accuracy (CA) of 96.31%. Moreover, an FPGA prototype of the on-implant circuit is implemented and tested. It is shown that in comparison to the OSORT algorithm, a pivotal spike sorting method, our spike sorting provides a higher CA at significantly lower hardware resources. The proposed circuit is also implemented in a 180-nm standard CMOS process, achieving a power consumption of 1.78[Formula: see text][Formula: see text] per channel and a chip area of 0.07[Formula: see text]mm2 per channel.

Exploring magneto-electric nanoparticles (MENPs): a platform for implanted deep brain stimulation

Orientation-selective neurons shift their preferred orientation after being adapted to a nonpreferred orientation. These shifts of the peaks of tuning curves may be in the attractive or repulsive direction in relation to the adapter orientation. In anesthetized cats, we recorded evoked electrical responses from the visual cortex in a conventional manner. The recorded spikes in cortex may present two typical waveforms: regular spikes or fast spikes. However, there is no evidence whether the shapes of spikes are related to the attractive or repulsive shifts of orientation tuning curves of cells. Our results show that after adaptation the recorded cells with both attractive and repulsive shifts display one or the other shape of spike. However, the magnitude of shifts is systematically higher for regular spikes, which is attributed to putative pyramidal cells, whereas tuning curves for fast spikes have smaller magnitudes and are evoked by putative interneurons.

I read Nagel et al. with interest (Nagel et al., 2015). In theory, the potential to modify cognitive function in Huntington's disease (HD) with electrical stimulation or specifically Deep Brain Stimulation (DBS) is exciting. HD is a devastating and terminal disease in need of better treatments, both symptomatic and neuroprotective. Cognitive dysfunction itself is becoming more recognized as a severely disabling aspect of the disease as well (Paulsen and Long, 2014). Nagel et al. outline an intriguing argument for pursuing the globus pallidus externa as a DBS target for cognitive improvement while also acknowledging that the current evidence is based on a small number of animal models and very limited human (n = 2) experience. In actuality, the practical nature of pursing DBS for the treatment of cognition in HD is more complicated. The ethics of pursuing DBS for new indications is a hotly debated topic (Unterrainer and Oduncu, 2015). We still do not yet fully understand the mechanism of DBS in treating disorders where its efficacy is established. Trialing it for other neurologic conditions without sound rational can be unnecessarily dangerous and not in the best interest of those we are trying to treat. While DBS was relatively free of complications in limited trials for HD chorea, it is not a benign procedure. It is a brain surgery with known irreversible surgical complications including infection, stroke, intracranial bleeding and even death (Bronstein et al., 2011). Some authors cite complication rates in Parkinson's disease as high as 10% for hemorrhage, 2% for stroke, 15% for infection and death rates up to 4.5% (Bronstein et al., 2011). Hardware complications requiring repeat surgery also occur. Finally, cognition and behavior have also been known to be negatively impacted after DBS surgery, albeit more often cited in the subthalamic target for Parkinson's disease. Still, there are accounts of cognitive decline and even mania after pallidal DBS as well (Miyawaki et al., 2000; Rothlind et al., 2014). Considering these known potential neuropsychiatric complications, trials of DBS for HD would need to proceed very cautiously given the high incidence of pre-existing cognitive dysfunction and increased risk for suicide. Lastly, I would caution against making any statements alluding to slowing disease progression or neuroprotection given the current data, especially to perspective future participants in any DBS trials. This is too premature. This is indeed a desperate population and the ability to identify subjects early in the prodromal stage of the disease increases the potential for inappropriate and improper recruitment to any future trials. Any research using DBS in this population needs to be done slowly with the highest ethical standards and rigorous consenting protocols. Personally, while I commonly recommend DBS for my patients for treatment of currently approved indications and participate in a high volume of surgeries, I am still holding out hope for a medicinal, neuroprotective therapy that would have reversible adverse effects and not carry the risk of a brain surgery.

Deep brain stimulation (DBS) is effective in managing motor symptoms of Parkinson's disease in well-selected individuals. Recently, research has shown that DBS in the basal ganglia (BG) can alter neural circuits beyond the traditional basal ganglia-thalamus-cortical (BG-TH-CX) loop. For instance, functional imaging showed alterations in cerebellar activity with DBS in the subthalamic nucleus (STN). However, these imaging studies revealed very little about how cell-specific cerebellar activity responds to STN stimulation or if these changes contribute to its efficacy. In this study, we assess whether STN-DBS provides efficacy in managing motor symptoms in Parkinson's disease by recruiting cerebellar activity. We do this by applying STN-DBS in hemiparkinsonian rats and simultaneously recording neuronal activity from the STN, brainstem and cerebellum. We found that STN neurons decreased spiking activity by 55% during DBS (P = 0.038), which coincided with a decrease in most pedunculopontine tegmental nucleus and Purkinje neurons by 29% (P < 0.001) and 28% (P = 0.003), respectively. In contrast, spike activity in the deep cerebellar nuclei increased 45% during DBS (P < 0.001), which was likely from reduced afferent activity of Purkinje cells. Then, we applied STN-DBS at sub-therapeutic current along with stimulation of the deep cerebellar nuclei and found similar improvement in forelimb akinesia as with therapeutic STN-DBS alone. This suggests that STN-DBS can engage cerebellar activity to improve parkinsonian motor symptoms. Our study is the first to describe how STN-DBS in Parkinson's disease alters cerebellar activity using electrophysiology in vivo and reveal a potential for stimulating the cerebellum to potentiate deep brain stimulation of the subthalamic nucleus.

Deep Brain Stimulation and Movement Disorder Treatment

Deep brain stimulation (DBS) is an FDA-approved neurosurgical procedure for treating patients with movement disorders such as Parkinson’s disease. Patient-specific computational modeling and visualization play a key role for efficient surgical and therapeutic decision-making relevant to DBS. The computational models analyze DBS post-operative brain imaging, e.g., computed tomography (CT), to understand the DBS electrode positions within the patient’s brain. The DBS stimulation settings for optimal patient response depend upon a physician’s knowledge regarding precise electrode positions. The finite resolution of brain imaging, however, restricts our understanding regarding precise DBS electrode positions. In our contribution, we study the problem of the quantification of positional uncertainty in the DBS electrodes caused by the finite resolution of post-operative imaging. We propose a Monte Carlo statistical framework, which takes the advantage of our analytical characterization of the DBS electrode geometry to understand the spatial uncertainty in DBS electrodes. Our statistical framework quantifies the uncertainty in two positional parameters of the DBS electrodes, namely, the longitudinal axis direction and the positions at sub-voxel levels. We interactively visualize quantified uncertainties by employing volume rendering and isosurfaces. We show that the spatial variations in the DBS electrode positions are significant for finite resolution imaging, and interactive visualization can be instrumental for efficient interpretation of the positional variations in the DBS lead.

Spike sorting, i.e., the separation of the firing activity of different neurons from extracellular measurements, is a crucial but often error-prone step in the analysis of neuronal responses. Usually, three different problems have to be solved: the detection of spikes in the extracellular recordings, the estimation of the number of neurons and their prototypical (template) spike waveforms, and the assignment of individual spikes to those putative neurons. If the template spike waveforms are known, template matching can be used to solve the detection and classification problem. Here, we show that for the colored Gaussian noise case the optimal template matching is given by a form of linear filtering, which can be derived via linear discriminant analysis. This provides a Bayesian interpretation for the well-known matched filter output. Moreover, with this approach it is possible to compute a spike detection threshold analytically. The method can be implemented by a linear filter bank derived from the templates, and can be used for online spike sorting of multielectrode recordings. It may also be applicable to detection and classification problems of transient signals in general. Its application significantly decreases the error rate on two publicly available spike-sorting benchmark data sets in comparison to state-of-the-art template matching procedures. Finally, we explore the possibility to resolve overlapping spikes using the template matching outputs and show that they can be resolved with high accuracy.

Objective To investigate the short-term outcome of deep brain stimulation (DBS) for the treatment of pain associated with PD (Parkinson′s disease). Methods The clinical data of 47 cases of PD patients with pain symptoms admitted to Neurosurgery Department of Northern Theater Command General Hospital from January 2017 to June 2018 were retrospectively analyzed. A total of 35 patients underwent STN (subthalamic nucleus)-DBS and 12 patients underwent GPi (globus pallidus internus)-DBS. The stimulation contact selection and parameter adjustment were carried out at 3 weeks post operation. Visual analogue scale (VAS) was used to evaluate the degree of preoperative and postoperative pain severity. Results The VAS [median (upper and lower quartiles)] were 1.0 (0.5-1.4) and 1.0(0.7-1.6) respectively at 3 months and 6 months post operation in 47 patients, which was statistically different from that of pre-operation 4.0(4.0-8.0) (both P<0.001). At 3 months post operation, the improvement rate of 35 patients undergoing STN-DBS was (79.5±26.3)%, which was not statistically different from that of 12 patients undergoing GPi-DBS [(76.9±28.4)%] (P=0.78). The improvement rate of 35 patients receiving STN-DBS was (75.1±27.4)% 6 months after operation, which was not statistically different from that of 12 patients receiving GPi-DBS [(82.8±28.9)%] (P=0.86). Conclusion Deep brain stimulation could effectively relieve PD-related pain. There seems no significant difference between GPi-DBS and STN-DBS in the treatment outcome of PD-related pain. Key words: Parkinson disease; Pain; Deep brain stimulation; Treatment outcome

This study tested whether firing rate and spike shape could be used to distinguish projection cells from interneurons in extracellular recordings of basolateral amygdala (BLA) neurons. To this end, we recorded BLA neurons in isoflurane-anesthetized animals with tungsten microelectrodes. Projection cells were identified by antidromic activation from cortical projection sites of the BLA. Although most projection cells fired spontaneously at low rates (<1 Hz), an important subset fired at higher rates (up to 6.8 Hz). In fact, the distribution of firing rates in projection cells and unidentified BLA neurons overlapped extensively, even though the latter cell group presumably contains a higher proportion of interneurons. The only difference between the two distributions was a small subset (5.1%) of unidentified neurons with unusually high firing rates (9-16 Hz). Similarly, distributions of spike durations in both cell groups were indistinguishable, although most of the fast-firing neurons had spike durations at the low end of the distribution. However, we observed that spike durations depended on the exact position of the electrode with respect to the recorded cell, varying by as much as 0.7 ms. Thus neither firing rate nor spike waveform allowed for unequivocal separation of projection cells from interneurons. Nevertheless, we propose the use of two firing rate cutoffs to obtain relatively pure samples of projection cells and interneurons: < or =1 Hz for projection cells and > or =7 Hz for fast-spiking interneurons. Supplemented with spike-duration cutoffs of > or =0.7 ms for projection cells and < or =0.5 ms for interneurons, this approach should keep instances of misclassifications to a minimum.

Deep brain stimulation (DBS) is a neurosurgical procedure that involves implanting a pulse generator in patient's chest to send electric pulses via subcutaneous leads to electrodes implanted in nuclei deep within the brain to modulate aberrant neural behavior. DBS is the gold standard treatment for drug resistant Parkinson's disease, and its applications are rapidly expanding for other neurologic and psychiatric disorders. It is estimated that 70% of patients with DBS implants will need to undergo magnetic resonance imaging (MRI) exams during their lifetime. Unfortunately, however, access to MRI is restricted to DBS patients due to safety risks associated with radiofrequency (RF) heating of implanted leads. The major risk is due to the antenna effect , where the electric field of MRI transmitter couples with long DBS leads and amplifies the specific absorption rate (SAR) of radiofrequency energy in the tissue surrounding tips of implanted leads. The resultant RF heating is highly patient specific: up to 2 orders of magnitude difference has been observed in temperature rise in the tissue around individual DBS leads (0.1°C to 20°C) depending on the specific lead trajectory and patient's body composition [2] .

Protocol for machine learning-based classification of neuronal cell types from spike waveforms

Objective To explore the therapeutic effect of bilateral subthalamic nucleus (STN) deep brain stimulation (DBS) in the treatment of refractory Meige syndrome (MS). Methods A total of 32 patients with MS who underwent STN-DBS at Department of Functional Neurosurgery, Aviation General Hospital of China Medical University from May 2014 to July 2017 were retrospectively enrolled into this study. The Burke-Fahn-Marsden Dystonia Disorder Scale (BFMDRS) was used to assess the improvement of clinical symptoms. The patient′s programmed parameters were documented as well. Results All 32 patients were successfully operated. No serious complications such as intracranial hemorrhage, wound infection, allergic reaction or electrical stimulator equipment failure occurred during perioperative period. Ten patients (31.3%) had varying degrees of dyskinesia during postoperative programing, which was improved at a certain level after adjustment of programmed parameters. One patient (3.1%) had psychiatric symptoms after surgery, and the mental symptoms disappeared 3 days after treatment. The postoperative follow-up time was >12 months. The improvement rates (median) of BFMDRS scores at 3, 6, and 12 months after surgery were 76.9%, 81.4%, and 82.8%, respectively. Among them, 18 patients were followed up for >24 months, and the improvement rate (median) was 82.5% at 24 months after surgery. All 32 patients were treated with unipolar stimulation mode, which was changed to double negative stimulation mode in 2 cases due to poor effects. The stimulation voltage was 0.8-3.4 V at 12 months postoperatively, the stimulation pulse width was 50-90 μs, and the stimulation frequency was 100-175 Hz. There was no significant increase or decrease in the programmed parameters of 32 patients between 3 and 12 months post operation. Conclusions STN-DBS seems safe and effective for patients with drug-resistant MS. Postoperative programming is relatively easy with parameters varying in a small scope. Key words: Meige syndrome; Dystonia; Subthalamic nucleus; Deep brain stimulation; Prognosis

Objective To analyze the changes in stimulation parameters post subthalamic nucleus(STN) deep brain stimulation (DBS) for the treatment of Parkinson′s disease. Methods Ninety-five cases of patients with Parkinson′s disease treated with STN-DBS at Department of Neurosurgery, Zhongnan Hospital of Wuhan University from November 2011 to June 2017 were analyzed retrospectively. The stimulation parameters of 95 patients were as follows: the monopolar stimulation was main (98.3%, 175/178), The stimulus contacts were mainly in the upper STN (90.9%, 171/188), and most of them were single contact stimulation (94.4%, 168/178). Patients were evaluated in 3 months, 6 months, 1 year and 2 years. The primary outcome was assessed included the post-operation parameters, the unified Parkinson′s disease rating scale Ⅲ(UPDRS-Ⅲ), the improvement rate[(preoperative score-postoperative score) / preoperative score × 100%], the stimulus energy(voltage2×frequency×pulsewidth/impedance×1 s), the stimulation-related adverse events. Results The electrodes of 95 patients were implanted into preset targets based on intraoperative microelectrode recording and postoperative MRI images.The score of UPDRS-Ⅲ was (12.7 ±6.7)points, which was better than that before operation(P<0.05), and the improvement rate was (70.7±12.1)%. During the follow-up period, the improvement rate of UPDRS-Ⅲ decreased gradually (F=2.421, P=0.048). The voltage increased gradually (F=24.633, P<0.001), the pulse width increased slightly (F=3.118, P=0.015), but the frequency did not change significantly (F=1.332, P=0.257). The energy increased gradually (P<0.05), the contacts in zona incerta (ZI) increased gradually (P<0.05), the bipolar stimulation increased slightly (P<0.05), and the double contacts gradually increased gradually (P<0.05). During the follow-up period, a total of 35 patients developed stimulation-related adverse events, including 14 patients with troublesome dyskinesia, 4 patients with gait disorder or disequilibrium, 1 patients with anxiety and depression, 2 patients with dysphagia or drinking cough, 8 patients with hallucinations, 1 patient with hypophrenia, 2 patients with increased excitability, 1 patient with eye opening difficulty. Conclusions With the passage of time in the treatment of Parkinson′s disease with STN-DBS, the stimulation voltage and energy increased gradually, and the stimulation mode of double contacts increased gradually, but it was still dominated by unipolar stimulation, and the upper part of STN is the best stimulus site. Key words: Parkinson disease; Subthalamic nucleus; Deep brain stimulation; Post-operation programming

Recent progress in multi-electrode spike sorting methods