Neural network surrogates of Bayesian diagnostic models for fast inference of plasma parameters

We present a framework for training artificial neural networks (ANNs) as surrogate Bayesian models for the inference of plasma parameters from diagnostic data collected at nuclear fusion experiments, with the purpose of providing a fast approximation of conventional Bayesian inference. Because of the complexity of the models involved, conventional Bayesian inference can require tens of minutes for analyzing one single measurement, while hundreds of thousands can be collected during a single plasma discharge. The ANN surrogates can reduce the analysis time down to tens/hundreds of microseconds per single measurement. The core idea is to generate the training data by sampling them from the joint probability distribution of the parameters and observations of the original Bayesian model. The network can be trained to learn the reconstruction of plasma parameters from observations and the model joint probability distribution from plasma parameters and observations. Previous work has validated the application of such a framework to the former case at the Wendelstein 7-X and Joint European Torus experiments. Here, we first give a description of the general methodological principles allowing us to generate the training data, and then we show an example application of the reconstruction of the joint probability distribution of an effective ion charge Z_eff-bremsstrahlung model from data collected at the latest W7-X experimental campaign. One key feature of such an approach is that the network is trained exclusively on data generated with the Bayesian model, requiring no experimental data. This allows us to replicate the training scheme and generate fast, surrogate ANNs for any validated Bayesian diagnostic model.

First Observation of a Stable Highly Dissipative Divertor Plasma Regime on the Wendelstein 7-X Stellarator

For the first time, the optimized stellarator Wendelstein 7-X has operated with an island divertor. An operation regime in hydrogen was found in which the total plasma radiation approached the absorbed heating power without noticeable loss of stored energy. The divertor thermography recorded simultaneously a strong reduction of the heat load on all divertor targets, indicating almost complete power detachment. This operation regime was stably sustained over several energy confinement times until the preprogrammed end of the discharge. The plasma radiation is mainly due to oxygen and is located at the plasma edge. This plasma scenario is reproducible and robust at various heating powers, plasma densities, and gas fueling locations. These experimental results show that the island divertor concept actually works and displays good power dissipation potential, producing a promising exhaust concept for the stellarator reactor line.

Aneurysmal bone cyst of the zygomatic arch

We present a case of aneurysmal bone cyst in the jaw investigated with plain film, US, computed tomography (axial, precontrast) and magnetic resonance (1 Tesla, axial T1 weighted before and after Gadolinium administration, axial and coronal T2-weighted after Gadolinium administration). The zygomatic arch is a very rare location for an aneurysmal bone cyst: as far as we know, only a few cases are found in the literature.

Schwannoma of the trachea

In this study we present a case of schwannoma of the trachea investigated with computer tomography (axial, precontrast) and magnetic resonance (1 Tesla, sagittal T1w before and after Gadolinium, sagittal T2w, coronal and axial after Gadolinium). The topographic abilities of MR allowed us to determine the exact location and extension of the tumor in the trachea. As far as we know, this study, based on magnetic resonance, is the first of the kind in the radiologic literature.

Neuroprotective efficacy of lifarizine (RS-87476) in a simplified rat survival model of 2 vessel occlusion

1. A new, modified rat two vessel occlusion model (with hypotension) was established and the neuroprotective efficacy of the novel agent lifarizine (RS-87476) was examined. 2. The two vessel occlusion model used in the study was a modification of the model described in the literature, whereby we have obviated the need to use a muscle relaxant and intubate the trachea to provide ventilatory support by providing a tight fitting face mask attached to the ventilator. Furthermore, the need to combine exsanguination and additional pharmacological means of inducing the mandatory hypotension (50 mmHg), required to decrease brain blood perfusion pressure, has been removed by simply manipulating the concentration of the already present halothane anaesthetic. 3. The appropriate level of hypotension having been reached, microvascular clips were applied to bilaterally occlude the common carotid arteries for 12 min. This resulted in a loss of the cortical EEG activity. Local cerebral blood flow was measured 6 min into the occlusion period, using the fully quantitative [14C]-iodoantipyrine autoradiographic technique, in a separate group of rats (n = 5). This illustrated the lack of any blood flow, in the areas under study, during the period when there was an isoelectric cortical EEG pattern. 4. The high grade global ischaemic lesion which occurred gave quantifiable neuronal damage in several vulnerable regions of the brain, namely, the hippocampal CA1 sub-field, cortex, thalamus, striatum, and cerebellar brain stem (Purkinje cells). 5. Following the global ischaemic insult the rats were allowed to recover for 72 h before assessment of the damage, during which time one group of rats (n = 11) received 100 micrograms kg-1 lifarizine i.a. 5 min post-occlusion, 500 micrograms kg-1 lifarizine i.p. 15 min post-occlusion, and 500 micrograms kg-1 lifarizine i.p. twice daily for 72 h. A second group of rats (n = 12) was treated with appropriate volumes of vehicle (0.4 ml kg-1 i.a. and 2 ml kg-1 i.p.) at identical time points. 6. Histopathological damage was assessed, from cresyl violet and haematoxyline/eosin stained sections, using a scoring system of 0-6 (no damage-complete neuronal death). The dosing regimen of lifarizine gave reduced damage in the hippocampal CA1 sub-field (4.1 +/- 0.3 to 2.8 +/- 0.6) and striatum (1.7 +/- 0.3 to 1.2 +/- 0.3) and significant neuroprotection in the anterior cortex (2.0 +/- 0.2 to 1.2 +/- 0.2; p < 0.05), thalamus (1.5 +/- 0.2 to 0.8 +/- 0.2; p < 0.01), posterior cortex (1.5 +/- 0.2 to 1.0 +/- 0.2; p < 0.05) and cerebellar brain stem (0.9 +/- 0.2 to 0.4 +/- 0.1; p < 0.01). The overall mean brain score was significantly reduced (from 1.5 +/- 0.1 to 0.9 +/- 0.2). 7. These data show that the newly modified 2 vessel occlusion model produced a quantifiable level of ischaemic damage and that the novel agent lifarizine is neuroprotective in the model.

Micronucleus induction by camptothecin and amsacrine in bone marrow of male and female CD-1 mice

Camptothecin (CPT) and amsacrine (m-AMSA), specific inhibitors of topoisomerase I and II, were tested for micronucleus inducing potential in bone marrow of male and female CD-1 mice. CPT was given as intraperitoneal (i.p.) injections at 0 h in doses of 1, 2 and 4 mg/kg, and at 0 + 24 h in doses of 0.5, 1 and 2 mg/kg. Bone marrow samples were taken 30 and 48 h post 0 h dosing in both the single and split dose studies. The same i.p. regimens were used to evaluate m-AMSA at doses of 1.5, 3 and 6 in the single, and 0.75, 1.5 and 3 mg/kg in the split dose study. Both compounds were tested in two consecutive experiments using identical study designs to confirm findings or trends. The CPT experiment showed that more micronuclei (MN) were induced at 30 h than at 48 h following both 0 h and 0 + 24 h treatment, and that dose interaction occurred as two divided doses of CPT induced as much or more micronuclei than the same total single dose in both sexes. No overall sex differences were found in the 0 h dose study at 30 or 48 h. However, females had significantly more MN than males at 30 h following 0 + 24 h dosing, indicating an inducible female G2 mitotic effect in proliferating bone marrow blast cells.(ABSTRACT TRUNCATED AT 250 WORDS)

A trial with a new peripheral implanted vascular access device

In a prospective study, a new implanted vascular access device designed for peripheral placement in the arm was evaluated. Thirty-two patients requiring long-term venous access received the Port-A-Cath P.A.S. Port over a 13-month period. The access devices were used for multiple therapies including chemotherapy, antibiotics, antivirals, antifungals, and blood products. After 4,896 patient days (range 12-388), 19 complications occurred in 14 patients or 3.88/1,000 catheter days. The infection rate was 3% or 0.2/1,000 catheter days. Port pocket cellulitis was reported in 3% of patients or 0.2/1,000 catheter days. Vein phlebitis occurred at 12.5% or 0.8/1,000 catheter days. Two instances of vessel thrombosis occurred (6.2% or 0.4/1,000 catheter days). Ten incidents in six patients of inability to aspirate blood samples were noted (18.75% or 2/1,000 catheter days). No infiltrations or extravasations were reported. Nurses involved in this early trial found performance similar to the standard venous chest ports. Peripheral port placement was accepted well by patients.

Management of Hickman catheter sepsis

Hickman catheters have been shown to be efficacious and durable. We believe their performance can be safely prolonged with aggressive management of complications. One hundred sixty-five consecutive patients with catheters were followed prospectively and 28 episodes of sepsis in 24 patients were evaluated. At least 90 percent of septic catheters were salvaged with a combination of antibiotics and thrombolytic agents. When our protocol was followed, all episodes of sepsis were successfully treated. This included gram-positive and gram-negative organisms as well as one case of fungal sepsis. Since treatment can be handled mostly on an outpatient basis, there are benefits in regard to patient comfort as well as health care costs.