Matching Recall and Storage in Sequence Learning with Spiking Neural Networks

Storing and recalling spiking sequences is a general problem the brain needs to solve. It is, however, unclear what type of biologically plausible learning rule is suited to learn a wide class of spatiotemporal activity patterns in a robust way. Here we consider a recurrent network of stochastic spiking neurons composed of both visible and hidden neurons. We derive a generic learning rule that is matched to the neural dynamics by minimizing an upper bound on the Kullback–Leibler divergence from the target distribution to the model distribution. The derived learning rule is consistent with spike-timing dependent plasticity in that a presynaptic spike preceding a postsynaptic spike elicits potentiation while otherwise depression emerges. Furthermore, the learning rule for synapses that target visible neurons can be matched to the recently proposed voltage-triplet rule. The learning rule for synapses that target hidden neurons is modulated by a global factor, which shares properties with astrocytes and gives rise to testable predictions.

THE MULTIPLE LOCATION TIME WEIGHTED INDEX: USING PATIENT ACTIVITY SPACES TO CALCULATE PRIMARY CARE SERVICE AREAS

Geographic health planning analyses, such as service area calculations, are hampered by a lack of patient-specific geographic data. Using the limited patient address information in patient management systems, planners analyze patient origin based on home address. But activity space research done sparingly in public health and extensively in non-health related arenas uses multiple addresses per person when analyzing accessibility. Also, health care access research has shown that there are many non-geographic factors that influence choice of provider. Most planning methods, however, overlook non-geographic factors influencing choice of provider, and the limited data mean the analyses can only be related to home address. This research attempted to determine to what extent geography plays a part in patient choice of provider and to determine if activity space data can be used to calculate service areas for primary care providers. During Spring 2008, a convenience sample of 384 patients of a locally-funded Community Health Center in Houston, Texas, completed a survey that asked about what factors are important when he or she selects a health care provider. A subset of this group (336) also completed an activity space log that captured location and time data on the places where the patient regularly goes. Survey results indicate that for this patient population, geography plays a role in their choice of health care provider, but it is not the most important reason for choosing a provider. Other factors for choosing a health care provider such as the provider offering “free or low cost visits”, meeting “all of the patient’s health care needs”, and seeing “the patient quickly” were all ranked higher than geographic reasons. Analysis of the patient activity locations shows that activity spaces can be used to create service areas for a single primary care provider. Weighted activity-space-based service areas have the potential to include more patients in the service area since more than one location per patient is used. Further analysis of the logs shows that a reduced set of locations by time and type could be used for this methodology, facilitating ongoing data collection for activity-space-based planning efforts.

UNSUPERVISED INDEXING OF MEDLINE ARTICLES THROUGH GRAPH-BASED RANKING

The biomedical literature is extensively catalogued and indexed in MEDLINE. MEDLINE indexing is done by trained human indexers, who identify the most important concepts in each article, and is expensive and inconsistent. Automating the indexing task is difficult: the National Library of Medicine produces the Medical Text Indexer (MTI), which suggests potential indexing terms to the indexers. MTI’s output is not good enough to work unattended. In my thesis, I propose a different way to approach the indexing task called MEDRank. MEDRank creates graphs representing the concepts in biomedical articles and their relationships within the text, and applies graph-based ranking algorithms to identify the most important concepts in each article. I evaluate the performance of several automated indexing solutions, including my own, by comparing their output to the indexing terms selected by the human indexers. MEDRank outperformed all other evaluated indexing solutions, including MTI, in general indexing performance and precision. MEDRank can be used to cluster documents, index any kind of biomedical text with standard vocabularies, or could become part of MTI itself.

Dirichlet-Neumann inverse spectral problem for a star graph of Stieltjes strings

We solve two inverse spectral problems for star graphs of Stieltjes strings with Dirichlet and Neumann boundary conditions, respectively, at a selected vertex called root. The root is either the central vertex or, in the more challenging problem, a pendant vertex of the star graph. At all other pendant vertices Dirichlet conditions are imposed; at the central vertex, at which a mass may be placed, continuity and Kirchhoff conditions are assumed. We derive conditions on two sets of real numbers to be the spectra of the above Dirichlet and Neumann problems. Our solution for the inverse problems is constructive: we establish algorithms to recover the mass distribution on the star graph (i.e. the point masses and lengths of subintervals between them) from these two spectra and from the lengths of the separate strings. If the root is a pendant vertex, the two spectra uniquely determine the parameters on the main string (i.e. the string incident to the root) if the length of the main string is known. The mass distribution on the other edges need not be unique; the reason for this is the non-uniqueness caused by the non-strict interlacing of the given data in the case when the root is the central vertex. Finally, we relate of our results to tree-patterned matrix inverse problems.

Feasibility of post mortem cardiac proton density weighted fast field echo imaging in two cases of sudden death

OBJECTIVE The aim of this work is to investigate and compare cardiac proton density (PD) weighted fast field echo (FFE) post-mortem magnetic resonance (PMMR) imaging with standard cardiac PMMR imaging (T1-weighted and T2-weighted turbo spin-echo (TSE)), postmortem CT (PMCT) as well as autopsy. MATERIALS AND METHODS Two human cadavers sequentially underwent cardiac PMCT and PMMR imaging (PD-weighted FFE, T1-weighted and T2-weighted TSE) and autopsy. The cardiac PMMR images were compared to each other as well as to PMCT and autopsy findings. RESULTS For the first case, cardiac PMMR exhibited a focal region of low signal in PD-weighted FFE and T2-weighted TSE images, surrounded by a signal intense rim in the T2-weighted images. T1-weighted TSE and PMCT did not appear to identify any focal abnormality. Macroscopic inspection identified a blood clot; histology confirmed this to be a thrombus with an adhering myocardial infarction. In the second case, a myocardial rupture with heart tamponade was identified in all PMMR images, located at the anterior wall of the left ventricle; PMCT excluded additional ruptures. In PD-weighted FFE and T2-weighted TSE images, it occurred hypo-intense, while resulting in small clustered hyper-intense spots in T1-weighted TSE. Autopsy confirmed the PMMR and PMCT findings. CONCLUSIONS Presented initial results have shown PD-weighted FFE to be a valuable imaging sequence in addition to traditional T2-weighted TSE imaging for blood clots and myocardial haemorrhage with clearer contrast between affected and healthy myocardium.

Preexisting Cerebral Microbleeds on Susceptibility-Weighted Magnetic Resonance Imaging and Post-Thrombolysis Bleeding Risk in 392 Patients

Background and Purpose—The question whether cerebral microbleeds (CMBs) visible on MRI in acute stroke increase the risk for intracerebral hemorrhages (ICHs) or worse outcome after thrombolysis is unresolved. The aim of this study was to analyze the impact of CMB detected with pretreatment susceptibility-weighted MRI on ICH occurrence and outcome. Methods—From 2010 to 2013 we treated 724 patients with intravenous thrombolysis, endovascular therapy, or intravenous thrombolysis followed by endovascular therapy. A total of 392 of the 724 patients were examined with susceptibility-weighted MRI before treatment. CMBs were rated retrospectively. Multivariable regression analysis was used to determine the impact of CMB on ICH and outcome. Results—Of 392 patients, 174 were treated with intravenous thrombolysis, 150 with endovascular therapy, and 68 with intravenous thrombolysis followed by endovascular therapy. CMBs were detected in 79 (20.2%) patients. Symptomatic ICH occurred in 21 (5.4%) and asymptomatic in 75 (19.1%) patients, thereof 61 (15.6%) bleedings within and 35 (8.9%) outside the infarct. Neither the existence of CMB, their burden, predominant location nor their presumed pathogenesis influenced the risk for symptomatic or asymptomatic ICH. A higher CMB burden marginally increased the risk for ICH outside the infarct (P=0.048; odds ratio, 1.004; 95% confidence interval, 1.000–1.008). Conclusions—CMB detected on pretreatment susceptibility-weighted MRI did not increase the risk for ICH or worsen outcome, even when CMB burden, predominant location, or presumed pathogenesis was considered. There was only a small increased risk for ICH outside the infarct with increasing CMB burden that does not advise against thrombolysis in such patients.

Thrombus imaging in acute stroke: correlation of thrombus length on susceptibility-weighted imaging with endovascular reperfusion success

OBJECTIVES Susceptibility-weighted imaging (SWI) enables visualization of thrombotic material in acute ischemic stroke. We aimed to validate the accuracy of thrombus depiction on SWI compared to time-of-flight MRA (TOF-MRA), first-pass gadolinium-enhanced MRA (GE-MRA) and digital subtraction angiography (DSA). Furthermore, we analysed the impact of thrombus length on reperfusion success with endovascular therapy. METHODS Consecutive patients with acute ischemic stroke due to middle cerebral artery (MCA) occlusions undergoing endovascular recanalization were screened. Only patients with a pretreatment SWI were included. Thrombus visibility and location on SWI were compared to those on TOF-MRA, GE-MRA and DSA. The association between thrombus length on SWI and reperfusion success was studied. RESULTS Eighty-four of the 88 patients included (95.5 %) showed an MCA thrombus on SWI. Strong correlations between thrombus location on SWI and that on TOF-MRA (Pearson's correlation coefficient 0.918, P < 0.001), GE-MRA (0.887, P < 0.001) and DSA (0.841, P < 0.001) were observed. Successful reperfusion was not significantly related to thrombus length on SWI (P = 0.153; binary logistic regression). CONCLUSIONS In MCA occlusion thrombus location as seen on SWI correlates well with angiographic findings. In contrast to intravenous thrombolysis, thrombus length appears to have no impact on reperfusion success of endovascular therapy. KEY POINTS • SWI helps in assessing location and length of thrombi in the MCA • SWI, MRA and DSA are equivalent in detecting the MCA occlusion site • SWI is superior in identifying the distal end of the thrombus • Stent retrievers should be deployed over the distal thrombus end • Thrombus length did not affect success of endovascular reperfusion guided by SWI.

Leptomeningeal collateralization in acute ischemic stroke: Impact on prominent cortical veins in susceptibility-weighted imaging.

BACKGROUND The extent of hypoperfusion is an important prognostic factor in acute ischemic stroke. Previous studies have postulated that the extent of prominent cortical veins (PCV) on susceptibility-weighted imaging (SWI) reflects the extent of hypoperfusion. Our aim was to investigate, whether there is an association between PCV and the grade of leptomeningeal arterial collateralization in acute ischemic stroke. In addition, we analyzed the correlation between SWI and perfusion-MRI findings. METHODS 33 patients with acute ischemic stroke due to a thromboembolic M1-segment occlusion underwent MRI followed by digital subtraction angiography (DSA) and were subdivided into two groups with very good to good and moderate to no leptomeningeal collaterals according to the DSA. The extent of PCV on SWI, diffusion restriction (DR) on diffusion-weighted imaging (DWI) and prolonged mean transit time (MTT) on perfusion-imaging were graded according to the Alberta Stroke Program Early CT Score (ASPECTS). The National Institutes of Health Stroke Scale (NIHSS) scores at admission and the time between symptom onset and MRI were documented. RESULTS 20 patients showed very good to good and 13 patients poor to no collateralization. PCV-ASPECTS was significantly higher for cases with good leptomeningeal collaterals versus those with poor leptomeningeal collaterals (mean 4.1 versus 2.69; p=0.039). MTT-ASPECTS was significantly lower than PCV-ASPECTS in all 33 patients (mean 1.0 versus 3.5; p<0.00). CONCLUSIONS In our small study the grade of leptomeningeal collateralization correlates with the extent of PCV in SWI in acute ischemic stroke, due to the deoxyhemoglobin to oxyhemoglobin ratio. Consequently, extensive PCV correlate with poor leptomeningeal collateralization while less pronounced PCV correlate with good leptomeningeal collateralization. Further SWI is a very helpful tool in detecting tissue at risk but cannot replace PWI since MTT detects significantly more ill-perfused areas than SWI, especially in good collateralized subjects.

Focal hemodynamic patterns of status epilepticus detected by susceptibility weighted imaging (SWI).

OBJECTIVE To investigate pathological findings in the susceptibility weighted imaging (SWI) of patients experiencing convulsive (CSE) or non-convulsive status epilepticus (NCSE) with focal hyperperfusion in the acute setting. METHODS Twelve patients (six with NCSE confirmed by electroencephalogram (EEG) and six patients with CSE with seizure event clinically diagnosed) underwent MRI in this acute setting (mean time between onset of symptoms and MRI was 3 h 8 min), including SWI, dynamic susceptibility contrast MR imaging (DSC) and diffusion-weighted imaging (DWI). MRI sequences were retrospectively evaluated and compared with EEG findings (10/12 patients), and clinical symptoms. RESULTS Twelve out of 12 (100 %) patients showed a focal parenchymal area with pseudo-narrowed cortical veins on SWI, associated with focal hyperperfused areas (increased cerebral blood flow (CBF) and mean transit time (MTT) shortening), and cortical DWI restriction in 6/12 patients (50 %). Additionally, these areas were associated with ictal or postical EEG patterns in 8/10 patients (80 %). Most frequent acute clinical findings were aphasia and/or hemiparesis in eight patients, and all of them showed pseudo-narrowed veins in those parenchymal areas responsible for these symptoms. CONCLUSION In this study series with CSE and NCSE patients, SWI showed focally pseudo-narrowed cortical veins in hyperperfused and ictal parenchymal areas. Therefore, SWI might have the potential to identify an ictal region in CSE/NCSE. KEY POINTS • The focal ictal brain regions show hyperperfusion in DSC MR-perfusion imaging. • SWI shows focally diminished cortical veins in hyperperfused ictal regions. • SWI has the potential to identify a focal ictal region in CSE/NCSE.

The Web KnowARR Framework: Orchestrating Computational Intelligence with Graph Databases

This chapter presents fuzzy cognitive maps (FCM) as a vehicle for Web knowledge aggregation, representation, and reasoning. The corresponding Web KnowARR framework incorporates findings from fuzzy logic. To this end, a first emphasis is particularly on the Web KnowARR framework along with a stakeholder management use case to illustrate the framework’s usefulness as a second focal point. This management form is to help projects to acceptance and assertiveness where claims for company decisions are actively involved in the management process. Stakeholder maps visually (re-) present these claims. On one hand, they resort to non-public content and on the other they resort to content that is available to the public (mostly on the Web). The Semantic Web offers opportunities not only to present public content descriptively but also to show relationships. The proposed framework can serve as the basis for the public content of stakeholder maps.