Search

Purpose of reviewThis mini-review is aimed to provide an overview and discuss procedural sedation and analgesia for atrial fibrillation (AF) ablation with focus at qualification of staff, patient evaluation, monitoring, medication and postprocedural care.Recent findingsSleep-disordered breathing is highly prevalent in patients with AF. Impact of often used STOP-BANG questionnaire to detect sleep-disordered breathing in AF patients is limited due to its restricted validity. Dexmedetomidine is a commonly used drug in sedation, but is shown not to be superior to propofol in sedation during AF-ablation. Alternatively use of remimazolam has characteristics that makes it a promising drug for minimal to moderate sedation for AF-ablation. High flow nasal oxygen (HFNO) has shown to reduce the risk of desaturation in adults receiving procedural sedation and analgesia.An optimal sedation strategy during AF ablation should be based on AF patient characteristics, the level of sedation needed, the procedure (duration and type of ablation) and the education and experience of the sedation provider. Patient evaluation and post procedural care are part of sedation care. More personalized care based on use of various sedation strategies and types of drugs as related to the type of AF-ablation is the way to further optimize care.

Online video gaming has become a ubiquitous aspect of modern-day video gaming. It has gained immense popularity due to its accessibility and immersive experience, resulting in millions of players worldwide participating in various online games. Depending on the type of gameplay, the players’ quality of experience (QoE) in online video gaming can be significantly affected by network factors such as high bandwidth and low latency. As such, providers of online gaming services are competing to offer the highest quality of experience to their users at reasonable prices. To achieve this objective, online game providers face two main challenges. Firstly, they must accurately estimate the network throughput capacity required to meet the servers’ demands and ensure that the QoE is not compromised. Secondly, they must be able to secure the required throughput with network providers, which, in the current conventional network infrastructure, is neither agile nor dynamic. Thus, online game providers have to prepay for extra network throughput capacity or choose a cost-effective capacity that may result in potential QoE losses during peak usage. To address these challenges, this thesis proposes a deep neural network-based model that utilizes a QoE-aware loss function for predicting the future network throughput de- mand. The model can accurately estimate the network throughput capacity required to maintain QoE levels while minimizing the cost of network resources. By doing so, on- line game providers can achieve optimal network resource allocation and effectively meet servers’ demands. Furthermore, this thesis proposes a slice optimizer module that employs 5G network slicing and a machine learning model to optimize network slices in a cost-efficient manner that satisfies both the online game provider’s and the network provider’s requirements. This module can dynamically allocate network resources based on the game provider’s QoE requirements, the network provider’s resource availability, and the cost of network resources. As a result, online game providers can efficiently manage network resources, optimize network slicing, and effectively control the cost of network resources.

ispartof: IET COLLABORATIVE INTELLIGENT MANUFACTURING vol:4 issue:3 pages:194-207 status: published

Using the non-invasive aquatic eddy covariance technique, we provide the first oxygen (O2) uptake rates from within coral gardens at the Condor seamount (Azores). To explore some of the key drivers of the benthic O2 demand, we obtained benthic images, quantified local hydrodynamics, and estimated phototrophic biomass and deposition dynamics with a long-term moored sediment trap. The coral gardens were dominated by the octocorals Viminella flagellum and Dentomuricea aff. meteor. Daily rates of O2 uptake within 3 targeted coral garden sites (203 to 206 m depth) ranged from 10.0 ± 0.88 to 18.8 ± 2.0 mmol m-2 d-1 (mean ± SE) and were up to 10 times higher than 2 local sandy reference sites within the seamount summit area. The overall mean O2 uptake rate for the garden (13.4 mmol m-2 d-1) was twice the global mean for sedimentary habitats at comparable depths. Combined with parallel ex situ incubations, the results suggest that the octocorals might contribute just ~5% of the observed O2 uptake rates. Deposition of particulate organic matter (POM) assessed by the sediment trap accounted for less than 10% of the O2 demand of the coral garden, implying a substantial POM supply circumventing the deployed traps. Our results expand the database for carbon turnover rates in cold-water coral habitats by including the first estimates from these largely understudied coral gardens.

Objectives: To evaluate the safety and efficacy of a novel wireless localization technique that uses radiofrequency identification markers for small and deep lung lesions. Methods: Preliminary use of the device was retrospectively evaluated in 2 Japanese centers. Under general anesthesia, a marker was placed as close as possible to the tumor via computed tomography-guided bronchoscopy in a hybrid operation theater. Surgeons located the marker without lung palpation using a detection probe the tone of which changed to indicate the marker-probe distance. Efficacy was defined as functional marker placement (bronchoscopy time and marker position) and deep margin distance. Results: Twelve markers were placed for 11 lesions (mean size, 6.8±2.7mm) located at a mean depth from the pleura of 11.4±8.4mm (range=0-26.0mm). Of 12 markers, 7 markers (58.3%) were placed within 10mm from the lesion in 25.5±14.4minutes. For the 11 wedge resections, markers were placed at a mean distance of 6.7mm (range, 0-13.0mm) from the lesion and a mean distance of 14.4mm (range, 3.0-42.0mm) from the pleura. All markers were recovered without complications, and all tumors were resected with negative margins. For 5 lesions >10mm deep to the pleura (mean depth, 18.9±5.5mm; range, 11.0-26.0mm), the median depth of the surgical margin was 11.6±2.1mm (range, 9.0-14.0mm). Conclusions: Radiofrequency identification marking was safe and precisely localized small lung lesions, including their depth.

Open-set recognition (OSR) has achieved significant importance in recent years. For a robust recognition system, we need to identify the right class from a myriad of knowns and unknowns. In this work, we build and compare OSR systems for patient activity recognition (PAR) using compact radar sensors in a hospital setting. Radar sensors are an important part of a privacy-preserving monitoring system. Specifically, the proposed approach is based on a deep discriminative representation network (DDRN) trained using the large margin cosine loss (LMCL) and triplet loss (TL). A probability of an inclusion model in the embedding space based on the Weibull distribution is able to separate knowns from unknowns. This overall approach limits the risk of open space and enables us to easily identify any unknown activities. Our experiments show that the proposed approach is significantly better for open-set human activity recognition (HAR) with radar when compared with the state-of-the-art open-set approaches.

In this article we establish new central limit theorems for Ruppert-Polyak averaged stochastic gradient descent schemes. Compared to previous work we do not assume that convergence occurs to an isolated attractor but instead allow convergence to a stable manifold. On the stable manifold the target function is constant and the oscillations in the tangential direction may be significantly larger than the ones in the normal direction. As we show, one still recovers a central limit theorem with the same rates as in the case of isolated attractors. Here we consider step-sizes $\gamma_n=n^{-\gamma}$ with $\gamma\in(\frac34,1)$, typically., Comment: 42 pages

Tiefe Neuronale Netze im Allgemeinen und Convolutional Neural Networks (CNNs) im Speziellen konnten in den letzten Jahren in den Bereichen Maschinelles Sehen und Bildverarbeitung in verschiedensten Domänen große Erfolge erzielen. Herausforderungen bestehen dabei vor allem in der Anpassung der Methoden an domänenspezifische Tasks und die Reduktion des im Normalfall recht hohen Ressourcenbedarfs, insbesondere für mobile Anwendungen wie das Autonome Fahren. Dort werden CNNs zur Bildverarbeitung meist für die Gewinnung relevanter Informationen über die Umgebung aus monokularen Kameras verwendet. Hierzu werden auf einem Kamerabild verschiedene Algorithmen zur Lösung unterschiedlicher Tasks ausgeführt. Im Rahmen dieser Arbeit wird mit dem MultiNet-Ansatz ein Konzept zur integrierten Bearbeitung verschiedener Tasks in einem gemeinsamen CNN-Modell vorgestellt. Der Ressourcenbedarf kann so im Vergleich zu einer getrennten Ausführung separater Modelle bei gleichbleibender Qualität der Ergebnisse deutlich reduziert werden. Zusätzlich wird ein Verfahren vorgestellt, welches bei ebenfalls gleichbleibender Ergebnisqualität durch eine Kombination der angepassten Methoden Pruning und Knowledge Distillation den Ressourcenverbrauch eines CNN-Modells signifikant reduzieren kann. Für die Domäne des Autonomen Fahrens werden CNN-Architekturen zur allgemeinen Objektdetektion an die Anforderungen der domänenspezifischen Detektion von Objekten im Fahrzeugumfeld angepasst. Hierbei findet eine getrennte Betrachtung von statischen und dynamischen Verkehrsobjekten statt. Zur Lösung der Herausforderungen bei der Detektion von statischen Verkehrsobjekten wie Ampeln oder Verkehrsschildern wird ein Ansatz zur hierarchischen Detektion gleichartiger Objekte vorgestellt. Für die Detektion von dynamischen Verkehrsobjekten wie anderen Verkehrsteilnehmern wird ein Ansatz zur direkten 3D Detektion von Objekten in einem Kamerabild eingeführt. Derartige Ansätze offenbaren durch ihre Komplexität allerdings eine Herausforderung beim Training vieler CNN-Modelle. Der Fortschritt und die Konvergenz des Trainingsprozesses werden in wesentlichen Teilen durch die Zusammensetzung der verwendeten Trainingsdaten bedingt. Um den Trainingsprozess mit den jeweils verfügbaren Trainingsdaten bestmöglich durchführen zu können, wird eine dynamische Fehlerfunktion vorgestellt, welche die einzelnen Datenpunkte angepasst an deren aktuelle Schwierigkeit im Trainingsprozess automatisch gewichtet. Eine Evaluation auf öffentlich verfügbaren Datensätzen wird für alle im Rahmen dieser Arbeit vorgestellten Konzepte durchgeführt.

The presence of plastics in seagrass meadows and within seagrass beached remains, including Neptune balls or aegagropilae (AG, natural aggregations of the seagrasses remains) (Huang et al., 2020; Sanchez-Vidal et al., 2021) has been reported. As a result, the potential of seagrasses in providing an ecosystem service by trapping marine plastics in coastal areas has been highlighted. However, there also exists evidence of the transport to the deep sea of seagrass remains. In the present study, the presence of plastics in AGs collected at great depths (1000-1200 m) on the SE Sardinian coast was characterized to evaluate the potential role of AGs in dislocating plastics. We confirmed the presence of plastics in 75% of the AGs collected, with fibres of diverse lengths (from 0.6 to 87 mm) and bundles of fibres (ranging from 0.8 to 18 mm of mean cross-section) being the most abundant items found (mean values of 47 fibres and 3 bundles per AG). In both cases, polyester was the predominant polymer, followed by polyamide and polypropylene. Overall, a positive relationship between AG size and the abundance of plastics was found, with no depth differences observed when this relationship was accounted for. Considering that the items found might have been entrapped in AGs during their formation process in shallow waters, our findings suggest that seagrasses may promote the dislocation of plastics to the deep sea. The potential impact of these entrapped plastics as well as their potential contribution to plastic exposure for deep-sea organisms remains unknown though, as little is known about the utilization of AGs and their degradation rates in this environment. Also see: https://micro2022.sciencesconf.org/426763/document, In MICRO 2022, Online Atlas Edition: Plastic Pollution from MACRO to nano

Recently, deep-sea trenches were described as trash bins and ultimate sink of anthropogenic litter, mostly represented by plastic and Microplastic (MP). The negative effects of MP on the deep environment are point of discussion in several topics, as it might be the cause imbalances of delicate ecological and biological interactions and affects the food web at all its levels. A comprehensive characterisation and realistic quantification of MP affecting these remote areas is therefore an important step required to estimate the level of environmental disturbance caused by these and the resulting negative effects. Moreover, it is not less important to establish the pathways wit with MP is reaching the deep-sea floor, for a better understanding of the nature of contamination. Within the deep MiPoll project, in collaboration with Senckenberg World of Biodiversity and the Alfred Wegener Institute for Polar and Marine Research, we contribute to the evaluation of MP contamination in the abyssal and hadal sediment of the Kuril Kamchatka trench, located in the North West Pacific Ocean. By analysing sediment samples collected during the KuramBio II expedition in summer 2016, we assessed the presence of microplastic in the trench, in sediments at a depth range of 5143-9450 m. Moreover, a heterogeneous microplastic distribution through the sediment column and different microplastic concentration and polymer types among sampling stations located in different areas of the trench reflects the high dynamics of this environment and the numerous forces that drive the deposition processes and the in situ recast of this contaminant at the trench floor. Also see: https://micro2022.sciencesconf.org/426751/document, In MICRO 2022, Online Atlas Edition: Plastic Pollution from MACRO to nano

Methane emissions are responsible for at least one-third of the total anthropogenic climate forcing and current estimations expect a significant increase in these emissions in the next decade. Consequently, methane offers a unique opportunity to mitigate climate change while addressing energy supply problems. From the five primary methane sources, residual water treatment provided 7% of the emissions in 2010. This ratio will undoubtedly increase with global population growth. Therefore, locating sources of methane emissions is a crucial step in characterizing the current distribution of GHG better. Nevertheless, there is a lack of comprehensive global and uniform databases to bind those emissions to concrete sources and there is no automatic method to accurately locate sparse human infrastructures such as wastewater treatment plants (WWTPs). WWTP detection is an open problem posing many obstacles due to the lack of freely accessible high-resolution imagery, and the variety of real-world morphologies and sizes. In this work, we tackle this state-of-the-art complex problem and go one step forward by trying to infer capacity using one end-to-end Deep Learning architecture and multi-modal remote sensing data. This goal has a groundbreaking potential impact, as it could help estimate mapped methane emissions for improving emission inventories and future scenarios prediction. We will address the problem as a combination of two parallel inference exercises by proposing a novel network to combine multimodal data based on the hypothesis that the location and the capacity can be inferred based on characteristics such as the plant situation, size, morphology, and proximity to water bodies or population centers. We explore technical documentation and literature to develop these hypotheses and validate their soundness with data analysis. To validate the architecture and the hypotheses, we develop a model and a dataset in parallel with a series of ablation tests. The process is facilitated by an automatic pipeline, also developed in this work, to create datasets and validate models leveraging those datasets. We test the best-obtained model at scale on a mosaic composed of satellite imagery covering the region of Catalonia. The goal is to find plants not previously labeled but present in wastewater treatment plant (WWTP) databases and to compare the distribution and magnitude of the inferred capacity with the ground truth. Results show that we can achieve state-of-the-art results by locating more than half of the labeled plants with the same precision ratio and by only using orthophotos from multispectral imagery. Moreover, we demonstrate that additional data sources related to water basins and population are valuable resources that the model can exploit to infer WWTP capacity. During the process, we also demonstrate the benefit of using negative instances to train our model and the impact of using an appropriate loss function such as Dice's loss.

The Bransfield Strait (Antarctica) is an important region for evaluating changes in Weddell Sea shelf waters on geological time scales because of its restricted connections to the surrounding ocean. However, the detailed oceanographic consequences of the opening of the strait remain unclear. We present bottom-current-related sedimentary features in the Bransfield Strait and examine the impact of the strait’s opening on deep-water circulation. Our findings show that the ocean circulation started to resemble that of the present day after a period of volcanic activity, possibly around the Middle Pleistocene. Coeval changes in Bransfield Strait morphology and an increase in seafloor irregularities due to the formation of volcanic chains finally determined new pathways for the Bransfield deep and bottom waters, enhanced due to the new climatic scenario of 100 k.y. cycles. The fact that “modernlike” oceanic circulation occurred only during previous interglacial periods demonstrates the significant impact of 100 k.y. climate cycles on the thermohaline changes of Antarctic deep waters. Hence, establishing a modern-day circulation model would enable researchers to assess paleoproductivity and local upwelling that have profoundly influenced the marine ecosystem of the Antarctic Peninsula after the Middle Pleistocene, SI

Galaxy clusters provide an excellent probe in various research fields in astrophysics and cosmology. However, the number of galaxy clusters detected so far in the $AKARI$ North Ecliptic Pole (NEP) field is limited. In this work, we provide galaxy cluster candidates in the $AKARI$ NEP field with the minimum requisites based only on coordinates and photometric redshift (photo-$z$) of galaxies. We used galaxies detected in 5 optical bands ($g$, $r$, $i$, $z$, and $Y$) by the Subaru Hyper Suprime-Cam (HSC), assisted with $u$-band from Canada-France-Hawaii Telescope (CFHT) MegaPrime/MegaCam, and IRAC1 and IRAC2 bands from the $Spitzer$ space telescope for photo-$z$ estimation. We calculated the local density around every galaxy using the 10$^{th}$-nearest neighbourhood. Cluster candidates were determined by applying the friends-of-friends algorithm to over-densities. 88 cluster candidates containing 4390 member galaxies below redshift 1.1 in 5.4 deg$^2$ have been detected. The reliability of our method was examined through false detection tests, redshift uncertainty tests, and applications on the COSMOS data, giving false detection rates of 0.01 to 0.05 and recovery rate of 0.9 at high richness. 3 X-ray clusters previously observed by $ROSAT$ and $Chandra$ were recovered. The cluster galaxies show higher stellar mass and lower star formation rate (SFR) compared to the field galaxies in two-sample Z-tests. These cluster candidates are useful for environmental studies of galaxy evolution and future astronomical surveys in the NEP, where $AKARI$ has performed unique 9-band mid-infrared photometry for tens of thousands of galaxies., Comment: 19 pages, 16 figures, has been accepted for publication in MNRAS

Progress in computing capabilities has enhanced science in many ways. In recent years, various branches of machine learning have been the key facilitators in forging new paths, ranging from categorizing big data to instrumental control, from materials design through image analysis. Deep learning has the ability to identify abstract characteristics embedded within a data set, subsequently using that association to categorize, identify, and isolate subsets of the data. Scanning probe microscopy measures multimodal surface properties, combining morphology with electronic, mechanical, and other characteristics. In this review, we focus on a subset of deep learning algorithms, that is, convolutional neural networks, and how it is transforming the acquisition and analysis of scanning probe data.

Multimodal steganography consists of concealing a signal into another one of a different medium, such that the latter is only very slightly distorted and the hidden information can be later recovered. A previous work employed deep learning techniques to this end by hiding an image inside an audio signal's spectrogram in a way that the encoding of one is independent of the other. In this work we explore the way in which images were being encoded previously and present a collection of improvements that produce a significant increase in the quality of the system. These mainly consist in encoding the image in a smarter way such that more information is able to be transmitted in a container of the same size. We also explore the possibility of using the short-time Fourier transform phase as an alternative to the magnitude and to randomly permute the signal to break the structure of the noise. Finally, we report results when using a larger container signal and outline possible directions for future work.

The aim of this paper is to review NATO’s Defence Education Enhancement Programme (DEEP) and to bring to the fore the measures of effectiveness regarding its implementation and to determine its shortcomings in the South Caucasus region. At the same time, the paper highlights the efforts of the DEEP teams to influence the partner educators through dialogue and encouragement. The authors refer to the best examples considering the measures of effectiveness. Considering the outcomes achieved by the partner nations, they endeavoured to apply Kirkpatrick model for measuring DEEP’s effectiveness in the South Caucasus countries. The paper discusses how they can more effectively address the challenges in their professional military education (PME) through their tailored DEEPs. The paper analyzes the processes that led by the South Caucasus nations, their national interests and what steps still need to be taken to realize their ambitions to be intellectually interoperable with NATO forces.

Focusing light through a multimode fibre (MMF) has attracted significant research interest, mainly driven by the need for miniature endoscopes in biomedicine. In recent years, digital micromirror devices (DMD) have become increasingly popular as a high-speed alternative to liquid-crystal spatial light modulators for light focusing via wavefront shaping based on binary amplitude modulations. To exploit the potentials and limitations of the state-of-the-art DMD-based wavefront shaping methods, in this study, for the first time, we compared four representative, non-holographic and DMD-based methods that are reported so far in literature with the same experimental and simulation conditions, including a real-valued intensity transmission matrix (RVITM)-based algorithm, a complex-valued transmission matrix (TM)-based algorithm, a conditional probability algorithm and a genetic algorithm. We investigated the maximum achievable peak-to-background ratio (PBR) in comparison to theoretical expectations, and further improved the performance of the RVITM-based method. With both numerical simulations and experiments, we found that the genetic algorithm offered the highest PBR but suffered from the lowest focusing speed, while the RVITM-based algorithm provided a comparable PBR to that of the genetic algorithm, and the highest focusing speed. ispartof: OPTICS EXPRESS vol:29 issue:10 pages:14269-14281 ispartof: location:United States status: published

A soundscape is the recording of all sounds present in an area, creating a holistic view of the acoustic profile in an ecosystem. Studying acoustic parameters of marine soundscapes as a whole has been shown to give an indication of the health status of the location, as well as correlate to which species may be present and using the area. With the rapid innovation of technology, especially data storage and declining cost of equipment, marine soundscape research is fast increasing, and these previous limitations have been switched for computing capacity for data analysis. Here, we perform a systematic assessment of literature of marine soundscape studies, from 1978, when the first soundscape study was reported, until 2021. We identified 200 primary research studies that recorded soundscapes and captured their geographical location, depth, habitat, duration of the study, and number of sites in each study. Using this data, we summarize the state of play in marine soundscapes studies, and identify knowledge gaps in the spatial coverage, depth profiles, habitat representation and study duration. Spatially, studies are biased towards the northern hemisphere. They are also more prevalent in more easily accessible ecosystems, in order from most to least studied, in coastal (38%), pelagic (20%), tropical coral reef (17%), rocky reef (7%), polar (5.5%), seagrass meadows, oyster reef and kelp/algal forest (4000 m. With anthropogenic noise and other pollution sources increasing globally, these gaps in research should be further addressed, especially as they pertain to vulnerable ecosystems, many of which are affected by global climate change and anthropogenic influences. It is crucial that marine soundscape studies continue to be developed and pursued, to establish baselines for healthy ecosystems and/or document recovery following management actions. info:eu-repo/semantics/publishedVersion

Semi-automatic and fully automatic contouring tools have emerged as an alternative to fully manual segmentation to reduce time spent contouring and to increase contour quality and consistency. Particularly, fully automatic segmentation has seen exceptional improvements through the use of deep learning in recent years. These fully automatic methods may not require user interactions, but the resulting contours are often not suitable to be used in clinical practice without a review by the clinician. Furthermore, they need large amounts of labeled data to be available for training. This review presents alternatives to manual or fully automatic segmentation methods along the spectrum of variable user interactivity and data availability. The challenge lies to determine how much user interaction is necessary and how this user interaction can be used most effectively. While deep learning is already widely used for fully automatic tools, interactive methods are just at the starting point to be transformed by it. Interaction between clinician and machine, via artificial intelligence, can go both ways and this review will present the avenues that are being pursued to improve medical image segmentation.

In stroke imaging, CT angiography (CTA) is used for detecting arterial occlusions. These images could also provide information on the extent of ischemia. The study aim was to develop and evaluate a convolutional neural network (CNN)–based algorithm for detecting and segmenting acute ischemic lesions from CTA images of patients with suspected middle cerebral artery stroke. These results were compared to volumes reported by widely used CT perfusion–based RAPID software (IschemaView). A 42-layer-deep CNN was trained on 50 CTA volumes with manually delineated targets. The lower bound for predicted lesion size to reliably discern stroke from false positives was estimated. The severity of false positives and false negatives was reviewed visually to assess the clinical applicability and to further guide the method development. The CNN model corresponded to the manual segmentations with voxel-wise sensitivity 0.54 (95% confidence interval: 0.44–0.63), precision 0.69 (0.60–0.76), and Sørensen–Dice coefficient 0.61 (0.52–0.67). Stroke/nonstroke differentiation accuracy 0.88 (0.81–0.94) was achieved when only considering the predicted lesion size (i.e., regardless of location). By visual estimation, 46% of cases showed some false findings, such as CNN highlighting chronic periventricular white matter changes or beam hardening artifacts, but only in 9% the errors were severe, translating to 0.91 accuracy. The CNN model had a moderately strong correlation to RAPID-reported Tmax > 10 s volumes (Pearson’s r = 0.76 (0.58–0.86)). The results suggest that detecting anterior circulation ischemic strokes from CTA using a CNN-based algorithm can be feasible when accompanied with physiological knowledge to rule out false positives.

This study focused on the possibility of implementing a vision-based architecture to monitor and detect the presence of trash or valuables in shared cars. The system was introduced to take pictures of the rear seating area of a four-door passenger car. Image capture was performed with a stationary wide-angled camera unit, and image classification was conducted with a prediction model in a remote server. For classification, a convolutional neural network (CNN) in the form of a fine-tuned VGG16 model was developed. The CNN yielded an accuracy of 91.43% on a batch of 140 test images. To determine the correlation among the predictions, a confusion matrix was used, and in addition, for each predicted image, the certainty of the distinct output classes was examined. The execution time of the system, from capturing an image to displaying the results, ranged from 5.7 to 17.2 s. Misclassifications from the prediction model were observed in the results primarily due to the variation in ambient light levels and shadows within the images, which resulted in the target items lacking contrast with their neighbouring background. Developments pertaining to the modularity of the camera unit and expanding the dataset of training images are suggested for potential future research.

Chronic limb-threatening ischemia (CLTI) is a significant facet of atherosclerotic disease that has major medical and functional consequences. With the number of diabetic patients on the rise and improved life expectancy, has also come a greater awareness and wider adoption of endovascular techniques; thus, amputation rates continue to fall.1-3 We expect that patients will be older, have more advanced comorbidities, and will have had more prior interventions. This may translate to a greater proportion of them presenting with no-option CLTI. Venous arterialization may be a viable alternative to preserving these limbs. This thesis starts with a general introduction that provides an update on the history and development of the venous arterialization. After this the thesis consists of three parts: Part 1 is entitled; the open venous arterialization Part 2 is entitled; percutaneous deep venous arterialization (pDVA, LimFlow procedure) Part 3 is entitled; postoperative care and follow-up of the venous arterialization The first part provides information about the open superficial venous arterialization. In Chapter 2, a superficial venous arterialization cohort and pedal bypass cohort of two Dutch hospitals is analysed, with promising results in a group of patients for whom no other reconstructive treatment options were available. Although the pedal bypass is still the gold standard (if a target artery is present), distal venous arterialization is a good alternative due to comparable results and its simplicity. The current evidence was analysed in a systematic review and meta-analysis in Chapter 3 and suggests that venous arterialization is a valuable treatment option in selected patients with no-option CLTI. These otherwise unsalvageable legs can be treated with acceptable morbidity and mortality. However, optimization and standardization of techniques are needed. The venous arterialization should always be considered in patients without distal arterial outflow vessels. The second part is about the percutaneous deep venous arterialization (pDVA). In Chapter 4, the first experience with the pDVA in humans is published, applied to a cohort of patients with no-option CLTI. The dual catheters, guided by ultrasound imaging, provide a reliable way to percutaneously create the AVF between a tibial artery and a deep tibial vein. Assisted by a percutaneously introduced valvulotome, arterial blood can now be directed to the veins of the foot. In this small cohort of patients, pDVA appears to be a safe and feasible procedure that effectively improves limb oxygenation, encourages wound healing, and potentially, avoids major amputation. The experience, after this first publication, from four centers between 2014 and 2018, is analysed in Chapter 5. This study presents midterm results from the largest population of patients with no-option CLTI treated with pDVA using the LimFlow device. In this complex group of patients, the LimFlow device demonstrated high technical success and amputation-free survival rates of 67% coupled with good wound healing at up to 24 months. In selected patients with no-option CLTI, pDVA is a safe and effective treatment to prevent amputation and heal wounds. A new international multicenter prospective study, PROMISE International, has been started to validate these outcomes in a larger cohort, and the protocol is described in Chapter 6. The last part of this thesis is about the postinterventional care of the venous arterialization. Chapter 7 is the first analysis of duplex ultrasound (DUS) measurements in post-pDVA patients. The venous arterialization needs time to mature before becoming effective. Stenoses or occlusions can frequently occur during this period, and therefore, there is a high need for more insight in DUS interpretation to detect failure of the arteriovenous circuit and preserve the limb. This study showed that surveillance of the arteriovenous circuit can be performed by DUS and we determined cut-off values to define the presence of stenosis or occlusions, but the small sample size of the study does not allow firm conclusions to be drawn. A guideline for post-pDVA care is given in Chapter 8. Based on literature and clinical experience of 24 patients from Alkmaar and Singapore, we summarize the different techniques proposed and provide an algorithm for the challenges that interventionalists can face in the follow-up period after a pDVA. The most important, is the acknowledgement that the pDVA needs 6 weeks to develop before becoming effective. Other key elements are edema treatment, pain control, wound care and a staged amputation strategy. The postprocedural management can be as challenging as the procedure itself but has not been a priority for further investigation.

Chronic limb-threatening ischemia (CLTI) is a significant facet of atherosclerotic disease that has major medical and functional consequences. With the number of diabetic patients on the rise and improved life expectancy, has also come a greater awareness and wider adoption of endovascular techniques; thus, amputation rates continue to fall.1-3 We expect that patients will be older, have more advanced comorbidities, and will have had more prior interventions. This may translate to a greater proportion of them presenting with no-option CLTI. Venous arterialization may be a viable alternative to preserving these limbs. This thesis starts with a general introduction that provides an update on the history and development of the venous arterialization. After this the thesis consists of three parts: Part 1 is entitled; the open venous arterialization Part 2 is entitled; percutaneous deep venous arterialization (pDVA, LimFlow procedure) Part 3 is entitled; postoperative care and follow-up of the venous arterialization The first part provides information about the open superficial venous arterialization. In Chapter 2, a superficial venous arterialization cohort and pedal bypass cohort of two Dutch hospitals is analysed, with promising results in a group of patients for whom no other reconstructive treatment options were available. Although the pedal bypass is still the gold standard (if a target artery is present), distal venous arterialization is a good alternative due to comparable results and its simplicity. The current evidence was analysed in a systematic review and meta-analysis in Chapter 3 and suggests that venous arterialization is a valuable treatment option in selected patients with no-option CLTI. These otherwise unsalvageable legs can be treated with acceptable morbidity and mortality. However, optimization and standardization of techniques are needed. The venous arterialization should always be considered in patients without distal arterial outflow vessels. The second part is about the percutaneous deep venous arterialization (pDVA). In Chapter 4, the first experience with the pDVA in humans is published, applied to a cohort of patients with no-option CLTI. The dual catheters, guided by ultrasound imaging, provide a reliable way to percutaneously create the AVF between a tibial artery and a deep tibial vein. Assisted by a percutaneously introduced valvulotome, arterial blood can now be directed to the veins of the foot. In this small cohort of patients, pDVA appears to be a safe and feasible procedure that effectively improves limb oxygenation, encourages wound healing, and potentially, avoids major amputation. The experience, after this first publication, from four centers between 2014 and 2018, is analysed in Chapter 5. This study presents midterm results from the largest population of patients with no-option CLTI treated with pDVA using the LimFlow device. In this complex group of patients, the LimFlow device demonstrated high technical success and amputation-free survival rates of 67% coupled with good wound healing at up to 24 months. In selected patients with no-option CLTI, pDVA is a safe and effective treatment to prevent amputation and heal wounds. A new international multicenter prospective study, PROMISE International, has been started to validate these outcomes in a larger cohort, and the protocol is described in Chapter 6. The last part of this thesis is about the postinterventional care of the venous arterialization. Chapter 7 is the first analysis of duplex ultrasound (DUS) measurements in post-pDVA patients. The venous arterialization needs time to mature before becoming effective. Stenoses or occlusions can frequently occur during this period, and therefore, there is a high need for more insight in DUS interpretation to detect failure of the arteriovenous circuit and preserve the limb. This study showed that surveillance of the arteriovenous circuit can be performed by DUS and we determined cut-off values to define the presence of stenosis or occlusions, but the small sample size of the study does not allow firm conclusions to be drawn. A guideline for post-pDVA care is given in Chapter 8. Based on literature and clinical experience of 24 patients from Alkmaar and Singapore, we summarize the different techniques proposed and provide an algorithm for the challenges that interventionalists can face in the follow-up period after a pDVA. The most important, is the acknowledgement that the pDVA needs 6 weeks to develop before becoming effective. Other key elements are edema treatment, pain control, wound care and a staged amputation strategy. The postprocedural management can be as challenging as the procedure itself but has not been a priority for further investigation.

Today, with the evolution of social behavior models, European regulations on emissions due to climate change and new transport alternatives available on the market, there has been a massive increase in the so-called personal mobility vehicles (PMV) in large cities. The fact that this increase has been so fast has meant that in many cases these vehicles do not have their own lanes or detailed regulations. This means that the public highway is often a shared space between pedestrians and PMV, in the case of sidewalks. Many of these PMVs have in their design a traffic mode that limits their speed to circulate on roads shared with pedestrians, but the reality is that in most cases the PMVs users circulate at a higher speed, putting the rest of the users at risk. This thesis aims to take a first step towards being able to create, in the future, PMVs capable of adapting their speed autonomously depending on the road they travel on.

This report aims to summarize the fundamental concepts of Artificial Intelligence (AI), and to provide a non-exhaustive overview of AI applications in dental imaging, comprising diagnostics, forensics, image processing and image reconstruction. AI has arguably become the hottest topic in radiology in recent years owing to the increased computational power available to researchers, the continuing collection of digital data, as well as the development of highly efficient algorithms for machine learning and deep learning. It is now feasible to develop highly robust AI applications that make use of the vast amount of data available to us, and that keep learning and improving over time.

Methane is produced microbially in vast quantities in sediments throughout the world’s oceans. However, anaerobic oxidation of methane (AOM) provides a near-quantitative sink for the produced methane and is primarily responsible for preventing methane emissions from the oceans to the atmosphere. AOM is a complex microbial process that involves several different microbial groups and metabolic pathways. The role of different electron acceptors in AOM has been studied for decades, yet large uncertainties remain, especially in terms of understanding the processes in natural settings. This study reports whole-core incubation methane oxidation rates along an estuarine gradient ranging from near fresh water to brackish conditions, and investigates the potential role of different electron acceptors in AOM. Microbial community structure involved in different methane processes is also studied in the same estuarine system using high throughput sequencing tools. Methane oxidation in the sediments was active in three distinct depth layers throughout the studied transect, with total oxidation rates increasing seawards. We find extensive evidence of non-sulphate AOM throughout the transect. The highest absolute AOM rates were observed below the sulphate-methane transition zone (SMTZ), strongly implicating the role of alternative electron acceptors (most likely iron and manganese oxides). However, oxidation rates were ultimately limited by methane availability. ANME-2a/b were the most abundant microbial phyla associated with AOM throughout the study sites, followed by ANME-2d in much lower abundances. Similarly to oxidation rates, highest abundances of microbial groups commonly associated with AOM were found well below the SMTZ, further reinforcing the importance of non-sulphate AOM in this system.

3D Concrete Printing (3DCP) has been gaining popularity in the past few years. Due to the nature of line-by-line printing and the slump of the material deposition in each extruded line, 3D printed structures exhibit obvious lines or marks at the layer interface, which affects surface finish quality and potentially affect bonding strength between layers. This makes it necessary to control the extrudate formation in 3DCP. However, it is difficult to directly analyse the extrudate formation process because the extrudate shape depends on many parameters. In this paper, a machine learning technique is applied to correlate the formation of the extrudate to the printing parameters using an Artificial Neural Network model. The training data for the model development was obtained from extrudates printed in 3DCP experiments. The performance of the trained model was experimentally validated and the predicted extrudate geometry resulting from the developed model showed good agreement to the actual extrudate geometry. Subsequently, the developed model was used to find proper nozzle shapes to produce designated extrudate geometries. Significant improvement on the printing quality was demonstrated using nozzle shapes generated from the model on 3D printed objects consisting a vertical wall, an inclined wall and a curved part. National Research Foundation (NRF) Accepted version

Accurate segmentation of pediatric echocardiography images is essential for a wide range of diagnostic and pre-interventional planning, but remains challenging (e.g., low signal to noise ratio and internal variability in heart appearance). To address these problems, in this paper, we propose a novel Cardiac Attention-guided Dual-path Network (i.e., AIDAN). AIDAN comprises a convolutional block attention module (CBAM) attached to a spatial (i.e., SPA) and context paths (i.e., CPA), which can guide the network and learn the most discriminative features. The spatial path captures low-level spatial features, and the context path is designed to exploit high-level context. Finally, features learned from the two paths are fused efficiently using a specially designed feature fusion module (FFM), and these are used to predict the final segmentation map. We experiment on a self-collected dataset of 127 pediatric echocardiography cases which are videos containing at least a complete cardiac cycle, and obtain a Dice coefficient of 0.951 and 0.914, in the left ventricle and atrium segments, respectively. AIDAN outperforms other state-of-the-art methods and has great potential for pediatric echocardiography images analysis.

Jingmen block is an essential location for the development and exploration of shale gas. The regional structure, the tectonic characteristics, and their influence on shale gas preservation in the Jingmen block were summarized based on observations from drilled cores, seismic data interpretation, FMI imaging logging results, and other methods of subsequent analysis. The study results show that: 1) the overall structure of the study area is simple. The shale corresponding to the Wufeng and Longmaxi formations of the Upper Ordovician and Lower Silurian periods were subjected to compression deformation in Indosinian and early Yanshanianand, following extension and strike-slip transformation in the late Yanshanian period. Structures, including back-thrust, echelon, parallel, inversion structures, and rifts, could also be observed 2) Fractures with different tectonic stages and scales have different effects on the preservation degree of shale gas in the study area, resulting in differences in the nature, scale and zone of influence of faults in different regions and 3) The crack associated with fault structure include interlayer low-angle detachment fracture, conjugate shear fracture, reticular fracture, high-angle shear fracture, high-angle strike-slip shear fracture, and high-angle fracture zone. The location and type of fracture development in different stages vary, which affect the shale gas preservation and migration 4) Based on the influence of structures on shale gas preservation, the study area is divided into four categories viz. two favorable areas and two more favorable areas.

ObjectiveTo establish the relationship between hematoma sites of involvement and hematoma expansion (HE) in patients with deep intracerebral hemorrhage (ICH).MethodsEligible patients with deep ICH admitted to hospital within 6 hours of onset between 2018 and 2020 were included in this retrospective multi-center study. Individuals with secondary ICH were excluded. The volume of HE was evaluated based on admission and follow-up computed tomography scans. Associations between deep ICH sites of involvement and HE were examined using multivariable logistic regression analysis while adjusting for confounding covariates of HE.ResultsWe enrolled 583 individuals from three stroke centers. Data from a final total of 460 patients were used in the analysis; of these patients, 159 (34.6%) had HE. In the crude model without adjustment, external capsule, anterior limb of the internal capsule, and posterior limb of the internal capsule (PLIC) involvement were correlated with HE. After fully adjusted models for sex, age, intraventricular hemorrhage, Glasgow Coma Scale admission score, baseline ICH volume, and time from onset to initial computed tomography, multivariable logistic regression revealed that the PLIC is a robust predictor of HE in patients with deep ICH (adjusted odds ratio = 2.73; 95% confidence interval = 1.75–4.26; p < 0.001).ConclusionInvolvement of the posterior limb of the internal capsule in deep hemorrhage could be a promising predictor of HE.

Zebrafish (Danio rerio) are rapidly emerging in biomedicine as promising tools for disease modelling and drug discovery. The use of zebrafish for neuroscience research is also growing rapidly, necessitating novel reliable and unbiased methods of neurophenotypic data collection and analyses. Here, we applied the artificial intelligence (AI) neural network-based algorithms to a large dataset of adult zebrafish locomotor tracks collected previously in a series of in vivo experiments with multiple established psychotropic drugs. We first trained AI to recognize various drugs from a wide range of psychotropic agents tested, and then confirmed prediction accuracy of trained AI by comparing several agents with known similar behavioral and pharmacological profiles. Presenting a framework for innovative neurophenotyping, this proof-of-concept study aims to improve AI-driven movement pattern classification in zebrafish, thereby fostering drug discovery and development utilizing this key model organism.

Precise prescription of medication dosing is crucial to patients, especially among Intensive Care Unit (ICU) patients. However, improper administration of some sensitive therapeutic medications (e.g., heparin) might place patients at unneeded risk, even life-threatening. Numerous factors such as a patient's clinical phenotype, genotype, and environmental factors will affect the heparin dose response. As a result, it is challenging to prescribe the optimal initial dose of heparin. In this paper, an individualized dosing policy is proposed to determine the optimal initial dose and minimize the risk of mis-dosing, as well as preventing the patients from late complications associated with medications dosing. A latent batch-constrained deep reinforcement learning (RL) algorithm is proposed to guarantee the safety of the medication recommendation system. The agent can observe a latent representation of patents and generate medication dosing solutions in successive and limited action spaces. The individualized dosing policy aims to reduce the extrapolation errors in the off-policy algorithms, by evaluating over-dosing and under-dosing of heparin in patients. Our results evaluated on Medical Information Mart for Intensive Care III (MIMIC-III) database demonstrate that the latent batch-constrained RL algorithm can work effectively from the retrospective data, showing promise to be used in future medication dosing policies.(C)& nbsp;2021 Elsevier B.V. All rights reserved.

Тема выпускной квалификационной работы: Модификация правила мутации в методе разностной эволюции с использованием Ð°Ñ€Ñ Ð¸Ð²Ð° разностей. Данная работа посвящена модификации правила мутации в методе разностной эволюции, реализованном в программном пакете DEEP, путём внедрения Ð°Ñ€Ñ Ð¸Ð²Ð° Ñ€Ð°Ð·Ð½Ð¾ÑÑ‚Ð½Ñ‹Ñ Ð²ÐµÐºÑ‚Ð¾Ñ€Ð¾Ð². В Ñ Ð¾Ð´Ðµ работы был разработан программный код на языке C, реализующий данную модификацию. Для проверки влияния данной модификации на производительность алгоритма были проведены численные эксперименты по оптимизации параметров тестовой функции Растригина и имитационной модели роста, развития и урожайности нута, разработанной Афширом Солтани и проанализированы полученные результаты. Показано, что модификация правила мутации позволяет в среднем снизить число вычислений целевой функции, что позволяет судить о перспективности алгоритма для оптимизации Ð¼Ð°Ñ‚ÐµÐ¼Ð°Ñ‚Ð¸Ñ‡ÐµÑÐºÐ¸Ñ Ð¼Ð¾Ð´ÐµÐ»ÐµÐ¹ в целом и Ð¸Ð¼Ð¸Ñ‚Ð°Ñ†Ð¸Ð¾Ð½Ð½Ñ‹Ñ Ð¼Ð¾Ð´ÐµÐ»ÐµÐ¹ Ð±Ð¸Ð¾Ð»Ð¾Ð³Ð¸Ñ‡ÐµÑÐºÐ¸Ñ Ð¿Ñ€Ð¾Ñ†ÐµÑÑÐ¾Ð² в частности., The subject of the graduate of qualification work is «Modification of mutation rule in Differential Evolution method using archive of difference vectors». This work is devoted to the modification of the mutation rule in the difference evolution method implemented in the DEEP software package by introducing an archive of difference vectors. In the course of the work a program code in the C language was developed that implements this modification. To test the effect of this modification on the performance of the algorithm, numerical experiments were carried out to optimize the parameters of the Rastrigin test function and the chickpea growth, development and yield simulation model developed by Afshir Soltani and analyzed the results. It is shown that the modification of the mutation rule allows, on average, to reduce the number of objective function calculations, which makes it possible to judge the prospects of the algorithm for optimizing mathematical models in general as well as simulation models of biological processes.

This study aims at timing the eclipses of the binary star TU UMi. The times of minima are taken from the literature, from our observations in April 2004 and from TESS observations between 2019 and 2022. The orbital period analysis of the system indicates that there is a cyclic oscillation with an amplitude of 0.0081d and a period of 9.03 yr, accompanied by a continuous decrease at a rate of dP/dt = ???1.12 ?? 10???7 d yr???1. We study the secular evolution of the orbital period of the system and the possibility of the existence of a third companion or the magnetic activity cycle of the primary component in the system.

Many software systems today face uncertain operating conditions, such as sudden changes in the availability of resources or unexpected user behavior. Without proper mitigation these uncertainties can jeopardize the system goals. Self-adaptation is a common approach to tackle such uncertainties. When the system goals may be compromised, the self-adaptive system has to select the best adaptation option to reconfigure by analyzing the possible adaptation options, i.e., the adaptation space. Yet, analyzing large adaptation spaces using rigorous methods can be resource- and time-consuming, or even be infeasible. One approach to tackle this problem is by using online machine learning to reduce adaptation spaces. However, existing approaches require domain expertise to perform feature engineering to define the learner, and support online adaptation space reduction only for specific goals. To tackle these limitations, we present 'Deep Learning for Adaptation Space Reduction Plus' -- DLASeR+ in short. DLASeR+ offers an extendable learning framework for online adaptation space reduction that does not require feature engineering, while supporting three common types of adaptation goals: threshold, optimization, and set-point goals. We evaluate DLASeR+ on two instances of an Internet-of-Things application with increasing sizes of adaptation spaces for different combinations of adaptation goals. We compare DLASeR+ with a baseline that applies exhaustive analysis and two state-of-the-art approaches for adaptation space reduction that rely on learning. Results show that DLASeR+ is effective with a negligible effect on the realization of the adaptation goals compared to an exhaustive analysis approach, and supports three common types of adaptation goals beyond the state-of-the-art approaches., Comment: 41 pages

Full list of authors: Borlaff, A. S.; Gómez-Alvarez, P.; Altieri, B.; Marcum, P. M.; Vavrek, R.; Laureijs, R.; Kohley, R.; Buitrago, F.; Cuillandre, J. -C.; Duc, P. -A.; Gaspar Venancio, L. M.; Amara, A.; Andreon, S.; Auricchio, N.; Azzollini, R.; Baccigalupi, C.; Balaguera-Antolínez, A.; Baldi, M.; Bardelli, S.; Bender, R.; Biviano, A.; Bodendorf, C.; Bonino, D.; Bozzo, E.; Branchini, E.; Brescia, M.; Brinchmann, J.; Burigana, C.; Cabanac, R.; Camera, S.; Candini, G. P.; Capobianco, V.; Cappi, A.; Carbone, C.; Carretero, J.; Carvalho, C. S.; Casas, S.; Castander, F. J.; Castellano, M.; Castignani, G.; Cavuoti, S.; Cimatti, A.; Cledassou, R.; Colodro-Conde, C.; Congedo, G.; Conselice, C. J.; Conversi, L.; Copin, Y.; Corcione, L.; Coupon, J.; Courtois, H. M.; Cropper, M.; Da Silva, A.; Degaudenzi, H.; Di Ferdinando, D.; Douspis, M.; Dubath, F.; Duncan, C. A. J.; Dupac, X.; Dusini, S.; Ealet, A.; Fabricius, M.; Farina, M. ; Farrens, S.; Ferreira, P. G.; Ferriol, S.; Finelli, F.; Flose-Reimberg, P.; Fosalba, P.; Frailis, M.; Franceschi, E.; Fumana, M.; Galeotta, S.; Ganga, K.; Garilli, B.; Gillis, B.; Giocoli, C.; Gozaliasl, G.; Graciá-Carpio, J.; Grazian, A.; Grupp, F.; Haugan, S. V. H.; Holmes, W.; Hormuth, F.; Jahnke, K.; Keihanen, E.; Kermiche, S.; Kiessling, A.; Kilbinger, M.; Kirkpatrick, C. C.; Kitching, T.; Knapen, J. H.; Kubik, B.; Kümmel, M.; Kunz, M.; Kurki-Suonio, H.; Liebing, P.; Ligori, S.; Lilje, P. B.; Lindholm, V.; Lloro, I.; Mainetti, G.; Maino, D.; Mansutti, O.; Marggraf, O.; Markovic, K.; Martinelli, M.; Martinet, N.; Martínez-Delgado, D.; Marulli, F.; Massey, R.; Maturi, M.; Maurogordato, S.; Medinaceli, E.; Mei, S.; Meneghetti, M.; Merlin, E.; Metcalf, R. B.; Meylan, G.; Moresco, M.; Morgante, G.; Moscardini, L.; Munari, E.; Nakajima, R.; Neissner, C.; Niemi, S. M.; Nightingale, J. W.; Nucita, A.; Padilla, C.; Paltani, S.; Pasian, F.; Patrizii, L.; Pedersen, K.; Percival, W. J.; Pettorino, V.; Pires, S.; Poncet, M.; Popa, L.; Potter, D.; Pozzetti, L.; Raison, F.; Rebolo, R.; Renzi, A.; Rhodes, J.; Riccio, G.; Romelli, E.; Roncarelli, M.; Rosset, C.; Rossetti, E.; Saglia, R.; Sánchez, A. G.; Sapone, D.; Sauvage, M.; Schneider, P.; Scottez, V.; Secroun, A.; Seidel, G.; Serrano, S.; Sirignano, C.; Sirri, G.; Skottfelt, J.; Stanco, L.; Starck, J. L.; Sureau, F.; Tallada-Crespí, P.; Taylor, A. N.; Tenti, M.; Tereno, I.; Teyssier, R.; Toledo-Moreo, R.; Torradeflot, F.; Tutusaus, I.; Valentijn, E. A.; Valenziano, L.; Valiviita, J.; Vassallo, T.; Viel, M.; Wang, Y.; Weller, J.; Whittaker, L.; Zacchei, A.; Zamorani, G.; Zucca, E.; Euclid Collaboration., Context. While Euclid is an ESA mission specifically designed to investigate the nature of dark energy and dark matter, the planned unprecedented combination of survey area (∼15 000 deg2), spatial resolution, low sky-background, and depth also make Euclid an excellent space observatory for the study of the low surface brightness Universe. Scientific exploitation of the extended low surface brightness structures requires dedicated calibration procedures that are yet to be tested. Aims. We investigate the capabilities of Euclid to detect extended low surface brightness structure by identifying and quantifying sky-background sources and stray-light contamination. We test the feasibility of generating sky flat-fields to reduce large-scale residual gradients in order to reveal the extended emission of galaxies observed in the Euclid survey. Methods. We simulated a realistic set of Euclid/VIS observations, taking into account both instrumental and astronomical sources of contamination, including cosmic rays, stray-light, zodiacal light, interstellar medium, and the cosmic infrared background, while simulating the effects of background sources in the field of view. Results. We demonstrate that a combination of calibration lamps, sky flats, and self-calibration would enable recovery of emission at a limiting surface brightness magnitude of μlim = 29.5−0.27+0.08 mag arcsec−2 (3σ, 10 × 10 arcsec2) in the Wide Survey, and it would reach regions deeper by 2 mag in the Deep Surveys. Conclusions.Euclid/VIS has the potential to be an excellent low surface brightness observatory. Covering the gap between pixel-to-pixel calibration lamp flats and self-calibration observations for large scales, the application of sky flat-fielding will enhance the sensitivity of the VIS detector at scales larger than 1″, up to the size of the field of view, enabling Euclid to detect extended surface brightness structures below μlim = 31 mag arcsec−2 and beyond. © ESO 2022., A. B. was supported by an appointment to the NASA Postdoctoral Program at the NASA Ames Research Center, administered by Universities Space Research Association under contract with NASA, and the European Space Agency (ESA), through the European Space Astronomy Center Faculty. We acknowledge a number of agencies and institutes that have supported the development of Euclid, in particular the Academy of Finland, the Agenzia Spaziale Italiana, the Belgian Science Policy, the Canadian Euclid Consortium, the Centre National d’Etudes Spatiales, the Deutsches Zentrum für Luft- und Raumfahrt, the Danish Space Research Institute, the Fundação para a Ciência e a Tecnologia, the Ministerio de Economia y Competitividad, the National Aeronautics and Space Administration, the National Astronomical Observatory of Japan, the Netherlandse Onderzoekschool Voor Astronomie, the Norwegian Space Agency, the Romanian Space Agency, the State Secretariat for Education, Research and Innovation (SERI) at the Swiss Space Office (SSO), and the United Kingdom Space Agency. A complete and detailed list is available on the Euclid website (http://www.euclid-ec.org). This work has made use of data from the ESA mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This research made use of NumPy (Van Der Walt et al. 2011), Astropy, a community-developed core Python package for Astronomy (Astropy Collaboration 2013). All of the figures on this publication were generated using Matplotlib (Hunter 2007). This work was partly done using GNU Astronomy Utilities (Gnuastro, ascl.net/1801.009) version 0.11.22-dc86., With funding from the Spanish government through the Severo Ochoa Centre of Excellence accreditation SEV-2017-0709.

In this paper, a new method has been proposed to optimize the hyper parameters in Deep neural network (DNN). For this purpose, a new version of Grey Wolf Optimizer named New Balance Grey Wolf Optimizer (NB-GWO) is successfully developed for the first time. NB-GWO introduces an equation to control the movement strategies at each iteration. Based on this equation, the exploitation ability will be prioritized during the first few iterations to improve the convergence rate and gain quick access to potential search spaces. Meanwhile, during the last few iterations, the exploration ability will be biased toward exploitation to increase the opportunities of escaping the local optima's problem or exploring new spaces with the hope of finding better solutions. Because of the diversity of movement strategies, the NB-GWO has established a better balance between the ability of exploitation and exploration than that of in the original GWO. To demonstrate the effectiveness of NB-GWO, 23 classical benchmark functions combined with eight well-known optimization algorithms are used to evaluate the performance of NB-GWO as the first example. Then, NB-GWO will be used to optimize the hyper parameters in deep neural networks (DNN) for damage detection in 2D concrete frame. The results show that NB-GWO is a grown-up version of GWO, the results obtained in this work have proved the effectiveness and reliability of NB-GWO in solving optimization problems, especially, for optimizing the hyper parameters in DNN.

Semi-automatic and fully automatic contouring tools have emerged as an alternative to fully manual segmentation to reduce time spent contouring and to increase contour quality and consistency. Particularly, fully automatic segmentation has seen exceptional improvements through the use of deep learning in recent years. These fully automatic methods may not require user interactions, but the resulting contours are often not suitable to be used in clinical practice without a review by the clinician. Furthermore, they need large amounts of labelled data to be available for training. This review presents alternatives to manual or fully automatic segmentation methods along the spectrum of variable user interactivity and data availability. The challenge lies to determine how much user interaction is necessary and how this user interaction can be used most effectively. While deep learning is already widely used for fully automatic tools, interactive methods are just at the starting point to be transformed by it. Interaction between clinician and machine, via artificial intelligence, can go both ways and this review will present the avenues that are being pursued to improve medical image segmentation.

[EN] Cancerous carcinomas are the most common kidney tumours, accounting for more than 90% of clinical cases. Studies show that more and more cases are being diagnosed with cancerous masses, with more than 400,000 new cases reported worldwide in 2020. The difficulty of diagnosis between renal tumours and renal cysts implies the need to develop an accurate diagnostic system. This paper proposes a CNN model for detecting pathological kidneys containing cysts or cancers. Several topologies were studied, focusing mainly on the comparison of the models versus their size, selecting the large one as the best of them. The model, trained and validated with a 66% - 33% percentage split, is able to classify tumours, cysts and healthy kidneys with 99.91% accuracy. When tested on clinical cases, the model obtains a maximum accuracy of 31.58%. The study conducted to justify such variation in performance indicates that the training images are not representative of the real cases. Either because of the alterations they may have undergone or because of the specific characteristics they follow, the model is limited and conditioned by the images it can classify., [ES] Este trabajo propone un modelo de CNN para la detección de riñones patológicos que contienen quistes o cánceres. Se recorre todo el proceso de creación del mismo, estudiando el mejor rendimiento posible, y se analizan los resultados obtenidos en relación con el trasfondo ético que lo sustenta.

Els models de Deep Learning (DL) s'estan convertint en un dels camps del Machine Learning (ML) més extesos en el seu ús. Tot i això, la gran quantitat de paràmetres i operacions que realitzen les xarxes neuronals les fan massa complexes per a que un humà pugui entendre les accions i decisions que prenen. L'opacitat del funcionament d'aquests models requereix d'algun sistema que expliqui i justifiqui les prediccions que se n'obtenen. Aquesta justificació és necessària en molts casos del món real, des del camp de la medicina al legislatiu, on es tracta informació sensible que requereix una confiança en que les decisions del model siguin correctes. En el cas d'aquest projecte, la motivació de les explicacions ve donada per l'aplicacació del DL al camp de les companyies asseguradores, entre les quals hi ha el Grup Catalana Occident (GCO) que és on s'ha desenvolupat aquest projecte. L'explicabilitat en models d'aprenentatge automàtic, i concretament de DL ha evolucionat molt al llarg dels darrers anys, i en l'actualitat hi ha una gran quantitat de mètodes per a obtenir explicacions. L'objectiu d'aquest projecte és la recopilació i homogeneització de diversos algoritmes d'explicabilitat enfocats al Deep Learning, principalment agrupats en mètodes Globals (explicacions generals del model) i mètodes Locals (explicacions de prediccions concretes). Amb les definicions i la unificació de notació d'aquests mètodes s'ha fet una implementació bàsica de tots ells sobre el conjunt de dades públic ''Adult Data Set'', i una posterior adaptació per un dataset privat de l'empresa, que serveixi com a guia per a la futura implementació d'una llibreria que permeti aplicar out-of-the-box aquests algoritmes en nous projectes que utilitzin models de DL. Deep Learning (DL) models are becoming one of the most widespread fields of Machine Learning (ML) in use. However, due to the large number of parameters and operations performed by neural networks, they make them too complex for a human to understand the actions and decisions they make. The opacity of the networks decision making requires some system that explains and justifies the predictions obtained. This justification is necessary in many cases in the real world, from the field of medicine to legislature, where sensitive information is involved that requires user trust in the model's decision making. In the case of this project, the motivation for the explanations is given by the application of DL in the field of insurance companies, among which is the Grup Catalana Occidente (GCO) which is where this project has been developed. The explainability in machine learning models, and specifically in DL, has evolved a lot over the last few years, and today there are a lot of methods to get explanations. The aim of this project is to compile and unify various explainability algorithms focused on deep learning, mainly grouped into Global methods (general explanations of the model) and Local methods (explanations of specific predictions). Given the definitions and unification of these methods, a basic implementation of all of them has been made on a public data set ''Adult Data Set'', and a subsequent adaptation on a private dataset that belongs to the company, which will serve as a guide for future implementation of explainable DL algorithms for projects that are built with neural networks.

We present an elegant scheme for providing multi-directional illumination in selective plane illumination microscopy (SPIM). Light sheets can be delivered from one of two opposed directions at a time and pivoted about their center for efficient stripe artifact suppression using only a single galvanometric scanning mirror to perform both functions. The scheme results in a much smaller instrument footprint and allows multi-directional illumination with reduced expense compared with comparable schemes. Switching between the illumination paths is near instantaneous and the whole-plane illumination scheme of SPIM maintains the lowest rates of photodamage, which is often sacrificed by other recently reported destriping strategies. The ease of synchronization allows this scheme to be used at higher speeds than resonant mirrors typically used in this regard. We provide validation of this approach in the dynamic environment of the zebrafish beating heart, where imaging at up to 800 frames per second is demonstrated alongside efficient suppression of artifacts. (c) 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

Beside the minimization of the prediction error, two of the most desirable properties of a regression scheme are stability and interpretability. Driven by these principles, we propose continuous-domain formulations for one-dimensional regression problems. In our first approach, we use the Lipschitz constant as a regularizer, which results in an implicit tuning of the overall robustness of the learned mapping. In our second approach, we control the Lipschitz constant explicitly using a user-defined upper-bound and make use of a sparsity-promoting regularizer to favor simpler (and, hence, more interpretable) solutions. The theoretical study of the latter formulation is motivated in part by its equivalence, which we prove, with the training of a Lipschitz-constrained two-layer univariate neural network with rectified linear unit (ReLU) activations and weight decay. By proving representer theorems, we show that both problems admit global minimizers that are continuous and piecewise-linear (CPWL) functions. Moreover, we propose efficient algorithms that find the sparsest solution of each problem: the CPWL mapping with the least number of linear regions. Finally, we illustrate numerically the outcome of our formulations.

Structure function data provide insight into the nucleon quark distribution. They are relatively straightforward to extract from the world's vast, and growing, amount of inclusive leptoproduction data. In turn, structure functions can be used to model the physical processes needed for planning and optimizing future experiments. In this paper a machine learning algorithm capable of predicting, using a unique set of parameters, the F2 structure function, for four-momentum transfer 0.055≤Q2≤800.0GeV2 and for Bjorken x from 2.8×10−5 to the pion threshold, is presented. The model was trained and reproduces the hydrogen and the deuterium data at a level comparable with the average uncertainty of the experimental data. Extending the model to heavier nuclei or expanding the kinematic range is straightforward. The model is at least ten times faster than existing grid-based structure functions parametrizations that rely on interpolation and a hundred times faster than models requiring convolutions, making it an ideal candidate for event generators and systematic studies.

Compressed sensing (CS) has been playing a key role in accelerating the magnetic resonance imaging (MRI) acquisition process. With the resurgence of artificial intelligence, deep neural networks and CS algorithms are being integrated to redefine the state of the art of fast MRI. The past several years have witnessed substantial growth in the complexity, diversity, and performance of deep learning-based CS techniques that are dedicated to fast MRI. In this meta-analysis, we systematically review the deep learning-based CS techniques for fast MRI, describe key model designs, highlight breakthroughs, and discuss promising directions. We have also introduced a comprehensive analysis framework and a classification system to assess the pivotal role of deep learning in CS-based acceleration for MRI., Comment: 42 pages, 5 figures, Proceedings of the IEEE

Y Artificial intelligence (AI) coupled with promising machine learning (ML) techniques well known from computer science is broadly affecting many aspects of various fields including science and technology, industry, and even our day-to-day life. The ML techniques have been developed to analyze high-throughput data with a view to obtaining useful insights, categorizing, predicting, and making evidence-based decisions in novel ways, which will promote the growth of novel applications and fuel the sustainable booming of AI. This paper undertakes a comprehensive survey on the development and application of AI in different aspects of fundamental sciences, including information science, mathematics, medical science, materials science, geoscience, life science, physics, and chemistry. The challenges that each discipline of science meets, and the potentials of AI techniques to handle these challenges, are discussed in detail. Moreover, we shed light on new research trends entailing the integration of AI into each scientific discipline. The aim of this paper is to provide a broad research guideline on fundamental sciences with potential infusion of AI, to help motivate researchers to deeply understand the state-of-the-art applications of AI-based fundamental sciences, and thereby to help promote the continuous development of these fundamental sciences.

Along with electromagnetic radiation, the Sun also emits a constant stream of charged particles in the form of solar wind. When these particles enter Earth’s atmosphere through a process known as particle precipitation, they can through a series of chemical reactions produce N Ox and HOx gases. These gases are greenhouse gases and deplete the ozone in the mesosphere and upper stratosphere. It is important to quantify the rate of production of these gases to model the potential climate impact. Existing particle detectors in space are suboptimal because they cannot determine the energy flux and pitch angle distribution of precipitating particles. The primary scientific objective of the DEEP project is to design a particle detector instrument that is specifically designed for particle precipitation measurements. This thesis investigates different data acquisition schemes for handling the signal from a pixel detector. The chosen approach is measuring the width of a shaped pulse to quantify the energy of the particle. Known as Time-over-Threshold, a detector circuit board is designed featuring high-speed comparators as threshold discriminators and the NG-MEDIUM FPGA from NanoXplore to implement the data acquisition. Digitizing the comparator pulse width is done with a Time-to-Digital converter (TDC) implemented in the FPGA fabric. Since the difference in pulse width is small for different energies, a high conversion resolution is required. Two high-resolution TDCs are designed and compared, both of which feature a digital counter and a method of interpolating the counter clock period. The first interpolation method applies the use of a multitapped delay line implemented with hard carry chain resources, and the second method oversamples the input with several equally off-phase sampling clocks. A resolution of 302 ps and a differential non-linearity of 3.26 was achieved with the delay line TDC clocked at 100 MHz. An automatic statistical calibration scheme is included to determine the actual delays of the delay line, utilizing a second asynchronous clock to generate uniformly distributed hits. The asynchronous oversampler resolution is clock frequency dependent and provides a 4-fold improvement to the clock period. The differential nonlinearity approaches zero with close matching of the off-phase clocks and operating frequency. A complete firmware design for the data acquisition and rocket telemetry of the detector is proposed and demonstrated. A simulation of the firmware utilizing each TDC topology is conducted and the delay line TDC is demonstrated to be the most accurate at all operating frequencies and thus the recommended TDC for the DEEP data acquisition. Masteroppgave i fysikk PHYS399 MAMN-PHYS

Photoacoustic tomography (PAT) with genetically encoded near‐infrared probes enables visualization of specific cell populations in vivo at high resolution deeply in biological tissues. However, because of a lack of proper probes, PAT of cellular dynamics remains unexplored. Here, the authors report a near‐infrared Forster resonance energy transfer (FRET) biosensor based on a miRFP670‐iRFP720 pair of the near‐infrared fluorescent proteins, which enables dynamic functional imaging of active biological processes in deep tissues. By photoacoustically detecting the changes in the optical absorption of the miRFP670 FRET‐donor, they monitored cell apoptosis in deep tissue at high spatiotemporal resolution using PAT. Specifically, they detected apoptosis in single cells at a resolution of ≈3 µm in a mouse ear tumor, and in deep brain tumors (>3 mm beneath the scalp) of living mice at a spatial resolution of ≈150 µm with a 20 Hz frame rate. These results open the way for high‐resolution photoacoustic imaging of dynamic biological processes in deep tissues using NIR biosensors and PAT., Photoacoustic tomography with genetically encoded probes allows for the imaging of specific cell populations in vivo. The authors report near‐infrared biosensor enabling dynamic functional imaging of active biological processes in deep tissues. The author also detect apoptosis in single cells and in deep‐brain tumors of living mice. The results pave the way for high‐resolution photoacoustic imaging of biological processes in deep tissues.