Evaluation of intracranial stenting in a simulated training and assessment environment for neuroendovascular procedures.

Purpose: Given the inherent complexity of neurointerventional procedures and the associated risks of ionizing radiation exposure, it is crucial to prioritize ongoing training and improve safety protocols. The aim of this study is to assess a training and evaluation in-vitro environment using a vascular model of M1 stenosis, within a clinical angiography suite, without relying on animal models or X-ray radiation. Materials and methods: Using a transparent model replicating M1 stenosis, we conducted intracranial stenting procedures with four different setups (Gateway & Wingspan, Gateway & Enterprise, Neurospeed & Acclino, and Pharos Vitesse). A video camera was integrated with the angiography system's monitor for real-time visualization, while a foot switch was employed to simulate live fluoroscopy. Three neuroradiologists with varying levels of expertise performed each procedure for three times. The total duration of fluoroscopy as well as the time from passing the stenosis with the wire to completion of the procedure were recorded using a dedicated software designed for this experimental setup. Results: Compared to the Gateway & Wingspan procedure, the total fluoroscopy time reduced significantly with the Gateway & Enterprise, Neurospeed & Acclino, and Pharos Vitesse procedures by 51.56 s, 111.33 s, and 144.89 s, respectively (p < 0.001). Additionally, physicians with under 2 years and over 5 years of experience reduced FT by 62.83 s and 106.42 s, respectively, (p < 0.001), compared to a novice physician. Similar trends were noted for the time of wire distal to stenosis, with significant reductions for Neurospeed & Acclino and Pharos Vitesse compared to both Gateway & Wingspan as well as Gateway & Enterprise (all p < 0.001). Conclusion: Procedures requiring wire exchange maneuvers exhibited nearly twice the fluoroscopy time in comparison to balloon-mounted stenting or stent-placement via PTA balloon catheters. The more experienced neuroradiologist demonstrated significantly quicker performance in line with expectations in a real-life clinical setting, when compared to the less experienced interventionalist. This in-vitro setup allowed the evaluation of alternative technical approaches and differences in experience of operators without the use of animal models or X-ray. The setup combines advantages of simulators and silicone vessel models in a realistic working environment.

Selective intra-carotid blood cooling in acute ischemic stroke: A safety and feasibility study in an ovine stroke model.

Selective therapeutic hypothermia (TH) showed promising preclinical results as a neuroprotective strategy in acute ischemic stroke. We aimed to assess safety and feasibility of an intracarotid cooling catheter conceived for fast and selective brain cooling during endovascular thrombectomy in an ovine stroke model.Transient middle cerebral artery occlusion (MCAO, 3 h) was performed in 20 sheep. In the hypothermia group (n = 10), selective TH was initiated 20 minutes before recanalization, and was maintained for another 3 h. In the normothermia control group (n = 10), a standard 8 French catheter was used instead. Primary endpoints were intranasal cooling performance (feasibility) plus vessel patency assessed by digital subtraction angiography and carotid artery wall integrity (histopathology, both safety). Secondary endpoints were neurological outcome and infarct volumes.Computed tomography perfusion demonstrated MCA territory hypoperfusion during MCAO in both groups. Intranasal temperature decreased by 1.1 °C/3.1 °C after 10/60 minutes in the TH group and 0.3 °C/0.4 °C in the normothermia group (p < 0.001). Carotid artery and branching vessel patency as well as carotid wall integrity was indifferent between groups. Infarct volumes (p = 0.74) and neurological outcome (p = 0.82) were similar in both groups.Selective TH was feasible and safe. However, a larger number of subjects might be required to demonstrate efficacy.

Comprehensive Analyses of Carbohydrates, 1,2-Dicarbonyl Compounds, and Advanced Glycation End Products in Industrial Bread Making.

The technology of bread making is characterized by three major steps: dough mixing, proofing, and baking. To follow the course of Maillard processes in an authentic food matrix, the complete manufacturing process of wheat bread rolls was assessed along all production steps with the quantitation of sugars, furfurals, 1,2-dicarbonyl compounds, and advanced glycation end products (AGEs). As a result, the AGE profile was significantly enlarged to more than 12 structures, and comprehensive mechanistic insights were provided. The analyses of five major German bread types including wheat, brown, rye bread, pumpernickel, and crispbreads led to AGE contents of 69-149 mg/kg bread or 984-1857 mg/kg protein. Major lysine protein modifications were carboxymethyl, carboxyethyl, and formyl lysine and pyrraline. Arginine was mainly modified by methylglyoxal (MGO) to give imidazolinones. A major part of MGO was confirmed to stem from microbial metabolism.

Efficient Analysis of 2-Acetyl-1-pyrroline in Foods Using a Novel Derivatization Strategy and LC-MS/MS.

2-Acetyl-1-pyrroline (2-AP) is a key odorant in many foods, such as aromatic rice and wheat bread, with a very low odor threshold of 0.05 µg/L in water. The small molecule with a popcornlike, roasty odor is generated biologically or by Strecker degradation within the Maillard-reaction cascades during thermal food processing with methylglyoxal and 1-pyrroline as the main direct precursors. Numerous gas-chromatographic methods for the analysis of 2-AP have been published, but the reactivity of the compound leads to discrimination or degradation during sample workup. We developed a novel derivatization method for 2-AP with o-phenylenediamine followed by HPLC-MS/MS analysis of the resulting stable quinoxaline. The precision (7%), repeatability (14%), recovery (92%), linearity (0.79-500 µg/kg), limit of detection (LOD, 0.26 µg/kg), and limit of quantitation (LOQ, 0.79 µg/kg) were validated for rice matrix and were excellent as compared with those of methods published before. With the novel method, 2-AP levels in typical foods like aromatic rice (131 µg/kg), wheat bread (18 µg/kg), brown bread (18 µg/kg), rye bread (18 µg/kg), and popcorn (38 µg/kg) were determined.

Analysis of Advanced Glycation Endproducts in Rat Tail Collagen and Correlation to Tendon Stiffening.

Methylglyoxal is a major 1,2-dicarbonyl compound in vivo and leads to nonenzymatic protein modifications, known as advanced glycation endproducts. Especially long-lived proteins like collagen are prone to changes of the mechanical or biological function, respectively, by accumulation of Maillard-derived modifications. Specifically, the resulting nonenzymatic cross-link structures in parallel to the natural maturation process of collagen fibrils lead to complications with age or during disease. A novel lysine-lysine amide cross-link derived from methylglyoxal, 2,15-diamino-8-methyl-9-oxo-7,10-diaza-1,16-hexadecanedioic acid, named MOLA, was synthesized and identified in vitro and in vivo. Tail tendons of young, adult, and old rats (3, 12, and 22 months) were enzymatically digested prior to analysis of acid-labile glycation products via liquid chromatography-tandem mass spectrometry (LC-MS/MS). As a result, nine monovalent amino acid modifications, mostly originating from methylglyoxal (36 µmol/mol leucine-equivalents in total), and four glycation cross-links (0.72 µmol/mol glucosepane, 0.24 µmol/mol DODIC (3-deoxyglucosone-derived imidazoline cross-link), 0.04 µmol/mol MODIC (methylglyoxal-derived imidazoline cross-link), 0.34 µmol/mol MOLA) were quantitated in senescent tendon collagen. The results correlated with increased tail tendon breaking time from 10 to 190 min and indicate that methylglyoxal is a major player in the aging process of connective tissue.

Novel α-Oxoamide Advanced-Glycation Endproducts within the N6-Carboxymethyl Lysine and N6-Carboxyethyl Lysine Reaction Cascades.

The highly reactive α-dicarbonyl compounds glyoxal and methylglyoxal are major precursors of posttranslational protein modifications in vivo. Model incubations of N2-t-Boc-lysine and either glyoxal or methylglyoxal were used to further elucidate the underlying mechanisms of the N6-carboxymethyl lysine and N6-carboxyethyl lysine reaction cascades. After independent synthesis of the authentic reference standards, we were able to detect N6-glyoxylyl lysine and N6-pyruvoyl lysine for the first time by HPLC-MS2 analyses. These two novel amide advanced-glycation endproducts were exclusively formed under aerated conditions, suggesting that they were potent markers for oxidative stress. Analogous to the well-known Strecker degradation pathway, leading from amino acids to Strecker acids, the oxidation of an enaminol intermediate is suggested to be the key mechanistic step. A highly sensitive workup for the determination of AGEs in tissues was developed. In support of our hypothesis, the levels of N6-glyoxylyl lysine and N6-pyruvoyl lysine in rat livers indeed correlated with liver cirrhosis and aging.

Deep learning with convolutional neural networks for EEG decoding and visualization.

Deep learning with convolutional neural networks (deep ConvNets) has revolutionized computer vision through end-to-end learning, that is, learning from the raw data. There is increasing interest in using deep ConvNets for end-to-end EEG analysis, but a better understanding of how to design and train ConvNets for end-to-end EEG decoding and how to visualize the informative EEG features the ConvNets learn is still needed. Here, we studied deep ConvNets with a range of different architectures, designed for decoding imagined or executed tasks from raw EEG. Our results show that recent advances from the machine learning field, including batch normalization and exponential linear units, together with a cropped training strategy, boosted the deep ConvNets decoding performance, reaching at least as good performance as the widely used filter bank common spatial patterns (FBCSP) algorithm (mean decoding accuracies 82.1% FBCSP, 84.0% deep ConvNets). While FBCSP is designed to use spectral power modulations, the features used by ConvNets are not fixed a priori. Our novel methods for visualizing the learned features demonstrated that ConvNets indeed learned to use spectral power modulations in the alpha, beta, and high gamma frequencies, and proved useful for spatially mapping the learned features by revealing the topography of the causal contributions of features in different frequency bands to the decoding decision. Our study thus shows how to design and train ConvNets to decode task-related information from the raw EEG without handcrafted features and highlights the potential of deep ConvNets combined with advanced visualization techniques for EEG-based brain mapping. Hum Brain Mapp 38:5391-5420, 2017. © 2017 Wiley Periodicals, Inc.

In vitro, contrast agent-based evaluation of the influence of flow diverter size and position on intra-aneurysmal flow dynamics using syngo iFlow.

PURPOSE: Treatment of intracranial aneurysm with flow-diverting devices has become widespread in recent years. Despite that, intra-aneurysmal flow changes are yet not fully understood and can lead to different complications. Our aim was an in vitro contrast-based evaluation of the influence of flow diverter size and position on intra-aneurysmal flow dynamics. METHODS: Flow-diverting devices with different sizes (diameters 4.0, 4.5, and 6.0 mm) were deployed in seven silicone aneurysm models at different positions relative to the aneurysm neck (proximal, central, distal). Using syngo iFlow, we defined quantitative evaluation criteria based on contrast medium intensity and performed a flow evaluation. RESULTS: Intra-aneurysmal flows were heavily dependent on both size and position of flow-diverting devices at the aneurysm neck. We observed a higher peak intensity delay and intra-aneurysmal washout delay with the centrally placed 4.0- and 4.5-mm device, respectively, compared to the proximal and distal positions. Especially distally placed 4.0-mm devices led to an earlier filling of the aneurysm and increased intra-aneurysmal contrast agent intensity compared to the parent vessel, due to a potential endoleak. CONCLUSIONS: Not only size but also position of flow-diverting devices have a considerable impact on the intra-aneurysmal flow dynamics. The suggested evaluation criteria allowed a quantitative comparison of flow-diverting effect using syngo iFlow and could represent an efficient tool for predicting flow diversion pre-procedurally.

Learning to Generate Chairs, Tables and Cars with Convolutional Networks.

We train generative 'up-convolutional' neural networks which are able to generate images of objects given object style, viewpoint, and color. We train the networks on rendered 3D models of chairs, tables, and cars. Our experiments show that the networks do not merely learn all images by heart, but rather find a meaningful representation of 3D models allowing them to assess the similarity of different models, interpolate between given views to generate the missing ones, extrapolate views, and invent new objects not present in the training set by recombining training instances, or even two different object classes. Moreover, we show that such generative networks can be used to find correspondences between different objects from the dataset, outperforming existing approaches on this task.

Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks.

Deep convolutional networks have proven to be very successful in learning task specific features that allow for unprecedented performance on various computer vision tasks. Training of such networks follows mostly the supervised learning paradigm, where sufficiently many input-output pairs are required for training. Acquisition of large training sets is one of the key challenges, when approaching a new task. In this paper, we aim for generic feature learning and present an approach for training a convolutional network using only unlabeled data. To this end, we train the network to discriminate between a set of surrogate classes. Each surrogate class is formed by applying a variety of transformations to a randomly sampled 'seed' image patch. In contrast to supervised network training, the resulting feature representation is not class specific. It rather provides robustness to the transformations that have been applied during training. This generic feature representation allows for classification results that outperform the state of the art for unsupervised learning on several popular datasets (STL-10, CIFAR-10, Caltech-101, Caltech-256). While features learned with our approach cannot compete with class specific features from supervised training on a classification task, we show that they are advantageous on geometric matching problems, where they also outperform the SIFT descriptor.