Search

Newborn infants with respiratory difficulties frequently require nasal respiratory support such as nasal continuous positive airway pressure (nCPAP) or high-flow nasal cannulae (HFNC). Oral feeding of these infants under nasal respiratory support remains controversial out of fear of aspiration and cardiorespiratory events. The main objective of this study was to evaluate the safety of oral feeding under different types of nasal respiratory support in newborn lambs without or with tachypnea. Eight lambs aged 4-5 days were instrumented to record sucking, swallowing, respiration, ECG, oxygen saturation, and arterial blood gases. Each lamb was given two bottles of 30 mL of milk with a pause of 30 s under videofluoroscopy in four conditions [no respiratory support, nCPAP 6 cmH2O, HFNC 7 L/min, HFNCCPAP (= HFNC 7 L/min + CPAP 6 cmH2O)] administered in random order. The study was conducted in random order over 2 days, with or without standardized tachypnea induced by thoracic compression with a blood pressure cuff. Generalized linear mixed models were used to compare the four nasal respiratory supports in terms of safety (cardiorespiratory events and aspiration), sucking-swallowing-breathing coordination, and efficacy of oral feeding. Results reveal that no nasal respiratory support impaired the safety of oral feeding. Most of the few laryngeal penetrations we observed occurred with HFNCCPAP. Nasal CPAP modified sucking-swallowing-breathing coordination, whereas the efficiency of oral feeding decreased under HFNCCPAP. Results were similar with or without tachypnea. In conclusion, oral feeding under nasal respiratory support is generally safe in a term lamb, even with tachypnea.NEW & NOTEWORTHY The practice of orally feeding newborns suffering from respiratory problems while on nCPAP or HFNC remains controversial for fear of triggering cardiorespiratory events or aspiration pneumonia, or aggravating chronic lung disease. The present results show that bottle-feeding is generally safe in full-term lambs under nasal respiratory support, both without and with tachypnea.

Prophylaxis treatment is considered as the reference approach for children with severe haemophilia A or B. However, no consensus about the best prophylaxis protocol has yet been identified in term of dosage and timing of infusions. Guidelines were drawn up in France in the early 2000s by an expert group. The objective of this 16-year study (2001 to 2016) was to describe the clotting factor concentrates (CFCs) use in haemophiac outpatients. This is a retrospective monocentric study. Pharmaceutical and clinical data were captured using medical records. Main outcome measures are CFCs use and clinical data in paediatrics. Eighty haemophiliacs A or B under 12 years old with a factor level less than 2% were included (74% of HA), from pharmaceutical outpatient data. Global use of CFCs followed the evolution of the patients' number and regimen type introduced: increase of prophylaxis and decrease of on demand regimen. The average age at the prophylaxis introduction is significantly different according to the birth year. Prophylaxis introduction was made earlier with an increase of prophylactic regimen joined to an increase of CFCs use. The significant reduction of haemarthrosis in our cohort can be linked to a first infusion age and a prophylaxis introduction much earlier.

GPS and more recently multi-GNSS carrier phase measurements have been used to measure the dynamic displacements of large structures, such as long-span bridges, in both the time and frequency domains. Such measurements can be used as part of a structural health monitoring system. Additionally, fibre optic-based systems have been used to measure long-term deformations of structures, such as tunnels and roads. The research presented in this paper brings together the ideas and technologies in the two aforementioned areas of research, resulting in dynamic displacement measurements of the hull of a ship at high frequencies, and indoor environments. Field trials using kinematic GPS and FBG sensors on the 138-m long Smyril passenger and vehicle ferry operating in the North Atlantic Ocean on the Faroe Islands are presented. FBG sensors were in the bow and engine room of the ship, gathering data at 1 kHz. The configuration of the surveys and the results from the FBG and GPS sensors are presented, in both the time and frequency domains. Various frequencies were measured, due to the movements of the ship in the ocean as well as the vibrations caused through the ship mainly due to the engine. One dominant frequency was of 12.25 Hz, measured at all the FBG locations, due to the engines’ operating speed of 735 RPM. Common frequencies were evident for both the FBG and GPS results for lower frequency displacement, caused by the movements in the sea. Such measurements could be used to monitor the long-term displacement characteristics and changes, in both the time and frequency domains, and used to help understand the health characteristics of the ship. Further, such measurements could be used to analyse the noise characteristics of the ship for both operational and environmental reasons.

Introduction With current molecular diagnosis, about 1 to 5% of haemophilia A (HA) patients remain genetically unresolved. In these cases, deep intronic variation or structural variation disrupting the F8 gene could be causal. Aim To identify the causal variation in four genetically unresolved mild-to-severe HA patients using an F8 mRNA analysis approach. Methods Ectopic F8 mRNA analysis was performed in four unrelated HA patients. An in vitro minigene assay was performed in order to confirm the deleterious splicing impact of each variation identified. Results In all probands, mRNA analysis revealed an aberrant splicing pattern, and sequencing of the corresponding intronic region found a deep intronic substitution. Two of these were new variations: c.2113+601G>A and c.1443+602A>G, while the c.143+1567A>G, found in two patients, has previously been reported. The c.1443+602A>G and the c.143+1567A>G variants both led to the creation of a de novo acceptor or donor splice site, respectively. Moreover, the c.143+1567A>G was found in 3/6 patients with genetically unresolved moderate HA registered in our laboratory. Haplotype analysis performed in all patients carrying the c.143+1567A>G variation suggests that this variation could be a recurrent variation. The c.2113+601G>A led to the exonization of a 122-bp antisense AluY element by increasing the strength of a pre-existing cryptic 5' splice site. For each point variation, in vitro splicing analysis confirmed its deleterious impact on splicing of the F8 transcript. Conclusion We identified three deep intronic variations, leading to an aberrant mRNA splicing process as HA causing variation.

The papers in this special section is to broadly engage the machine learning and multimedia communities on the emerging yet challenging interpretable machine learning. Multimedia is increasingly becoming the “biggest big data,” among the most important and valuable source for insight and information. Many powerful machine learning algorithms, especially deep learning models such as convolutional neural networks (CNNs), have recently achieved outstanding predictive performance in a wide range of multimedia applications, including visual object classification, scene understanding, speech recognition, and activity prediction. Nevertheless, most deep learning algorithms are generally conceived as blackbox methods, and it is difficult to intuitively and quantitatively understand the results of their prediction and inference. Since this lack of interpretability is a major bottleneck in designing more successful predictive models and exploring wider-range useful applications, there has been an explosion of interest in interpreting the representations learned by these models, with profound implications for research into interpretable machine learning in the multimedia community.

Efficient and economic foundations are essential to ensure the long-term integrity of structures. Driven ductile piles offer a safe and quick solution for foundations, which can be individually customized to changing soil conditions. Geotechnical load tests on a small subset of piles can be performed at large construction sites to examine the bearing capacity for optimization purposes. Arising deformations during these statical tests are usually measured using electrical sensors at the top, which, however, do not deliver information about the stress distribution along the pile. This paper presents a fiber optic monitoring approach, which provides distributed strain profiles with a spatial resolution of up to 10 mm along driven ductile piles. The high measurement resolution of about $$1~{\mu}m/m$$ 1 μ m / m enables the detection of local effects in the load transfer from the pile to the surrounding grout and soil. The critical sensor installation on-site as well as results of various field applications with pile lengths of up to 25 m are presented. Verification measurements at the pile’s head and internal measurements of strain gauges prove the suitability of the developed monitoring approach and demonstrate the high potential of distributed fiber optic sensing for applications in soil mechanics.

Aim For patients with severe haemophilia A, guidelines recommend prophylactic treatment with FVIII, with dose calculations targeting a predetermined FVIII trough level. However, this pharmacokinetic (PK) approach is suboptimal, with some patients experiencing breakthrough bleeds. We aimed to improve FVIII dosing by incorporating the thrombin generation assay, a global haemostasis assay whose main pharmacodynamic (PD) parameter, endogenous thrombin potential (ETP), predicts spontaneous bleeding risk. Methods We performed post hoc combined PK‐PD modelling using data from 66 adults who received human‐cl rhFVIII (Nuwiq®, Octapharma AG) in a phase IIIb study. Time‐to‐event analyses simulated the probability of spontaneous bleeding for different FVIII exposures and baseline ETPs. Results Ninety‐one spontaneous bleeds occurred in 20/66 patients. The relationship between FVIII:C and ETP was non‐linear, and the sigmoid Emax model adequately described the data. Individual PK‐PD Bayesian estimation significantly improved predictive performance. Simulations showed that the mean spontaneous annual bleeding rate decreased with increasing baseline ETP or dosing: with ETP values of 200, 400 and 600 (nmol/L)·min annual bleeding rates were 2.36, 1.25 and 0.66, respectively, on 40 IU/kg human‐cl rhFVIII every 3 days; and annual bleeding rates were 2.09, 1.10, and 0.60, respectively, on 60 IU/kg every 3 days. Conclusion Prophylactic FVIII dosing is more clinically meaningful when incorporating ETP alongside FVIII level. For the first time, FVIII dosing can be personalized with the aim of eliminating spontaneous breakthrough bleeds.

Introduction Recombinant FVIIa (rFVIIa) is widely used to manage bleeding risk during and after surgery in patients with haemophilia complicated by inhibitors. In the postoperative period, rFVIIa must be delivered frequently and regularly to maintain haemostasis, considering its short half-life. Preparation and manual administration of bolus doses of rFVIIa at regular intervals may place a strain on available nursing resources. A programmable mini-pump may offer an approach to facilitate regular administration of bolus doses of rFVIIa at specified intervals. Aim To investigate if a mini-pump is a practical and effective way to deliver rFVIIa in the postoperative period. Methods It was first necessary to establish that rFVIIa remains stable and sterile in the mini-pump reservoir for an extended period. Four days after loading the mini-pump under sterile conditions no evidence of bacterial or fungal growth was observed and in vitro procoagulant activity of rFVIIa remained stable. The mini-pump was used to deliver rFVIIa as bolus doses to two patients with inhibitors who had undergone surgery. Nurses were asked to report their satisfaction with the use of the mini-pump using a specific questionnaire. Results Haemostasis was evaluated as excellent in both cases; nurses were satisfied with use of the mini-pump. Conclusion This pilot study shows that intermittent delivery of rFVIIa at fixed intervals using an automated mini-pump offers accurate and reliable administration in the postoperative setting. This approach may reduce burden on nursing staff, potentially minimize the risk of human error and avoid delay in administration of rFVIIa.

Grouted steel anchors combined with shotcrete layers at the surface are a common method to stabilize slopes and construction pits. Their design is based on the equilibrium consideration of a rigid body motion, and therefore, only tensile stresses are considered and accompanying forces, such as bending, are neglected. Modern approaches substitute the steel with wooden anchors, which show a sufficient tensile, but poor shear strength capacity. Distributed bending curves along the embedded anchors are required to design the optimal anchoring system, which, however, may not be measured using conventional sensing techniques. This paper reports about a distributed fiber optic sensing approach that enables the assessment of curvature and bending characteristics along grouted anchors. The designed system was tested within comprehensive laboratory investigations, in which various loading scenarios were applied to an instrumented, real-scale steel anchor. The suitability of the fiber optic sensing system was proven using image-based and geodetic total station measurements at the surface as well as by comparisons to theoretical beam models. In addition, three instrumented anchors were installed at a highway construction site and autonomously monitored over several weeks. The presented studies demonstrate the high potential of distributed fiber optic monitoring systems and their capability to extend traditional measurement methods in foundation engineering.

This paper introduces the successful implementation of a fibre-optic sensing system for direct and distributed strain measurement within the shotcrete lining of a conventional tunnel drive. The shotcrete lining of the top-heading and the invert are equipped with two layers (rock side and cavity side) of fibre-optic sensing cables installed in circumferential and longitudinal direction. All cables are measured autonomously for several weeks to capture the strain evolution inside the lining from the day of the construction to a posterior, well-hardened state. An additional follow-up measurement is conducted 2 months afterwards. The measurements enable an assessment of the strain distribution inside the lining with a spatial resolution in the range of some centimetres and a measurement resolution of up to 1 µm/m. Besides the conventional monitoring targets used for displacement recordings, measurement equipment like strain gages and pressure cells are also installed in the cross-section under investigation. Back-calculated strain from absolute displacements and the readings from the strain gages show good agreement with the results of the conducted fibre-optic measurements and verify the suitability of the used system.

New forms of multimedia data (such as text, numbers, tags, networking, signals, geo-tagged information, graphs/relationships, three-dimensional/VR/AR and sensor data, etc.) have emerged in many applications in addition to traditional multimedia data (image, video, audio). Multimedia has become the biggest of big data as the foundation of today's data-driven discoveries. Almost all disciplines of science and engineering, as well as social sciences, involve multimedia data in some forms, such as recording experiments, driverless cars, unmanned aerial vehicles, smart communities, biomedical instruments, security surveillance. Some recent events demonstrate the power of real-time broadcast of unfolding events on social networks. Multimedia data is not just big in volume, but also multi-modal and mostly unstructured. Storing, indexing, searching, integrating, and recognizing from the vast amounts of data create unprecedented challenges. Even though significant progress has been made processing multimedia data, today's solutions are inadequate in handling data from millions of sources simultaneously. The IEEE International Conference on Multimedia Information Processing and Retrieval (IEEE-MIPR) aims to provide a forum for original research contributions and practical system design, implementation, and applications of multimedia information processing and retrieval for single modality or multiple modalities. The target audiences are university researchers, scientists, industry practitioners, software engineers, and graduate students who want to become acquainted with technologies for big data analytics, machine intelligence, information fusion in multimedia information processing and retrieval. Following a successful debut in Miami, Florida in 2018, the second MIPR, held in San Jose, California, attracted over 140 submissions and more than 100 attendees—a big growth from its inaugural event and an indication of its growing visibility and popularity in the multimedia research society. In addition to four keynote talks, MIPR'19 also featured six innovation multimedia forums organized by both academic and industrial researchers, with topics ranging from computer vision, self-driving, Natural Language Processing, next generation video display, to cutting edge video/audio coding techniques.

This paper deals with the problem of semantic image segmentation of street scenes at night, as the recent advances in semantic image segmentation are mainly related to daytime images. We propose a method to extend the learned domain of daytime images to nighttime images based on an extended version of the CycleGAN framework and its integration into a self-supervised learning framework. The aim of the method is to reduce the cost of human annotation of night images by robustly transferring images from day to night and training the segmentation network to make consistent predictions in both domains, allowing the usage of completely unlabelled images in training. Experiments show that our approach significantly improves the performance on nighttime images while keeping the performance on daytime images stable. Furthermore, our method can be applied to many other problem formulations and is not specifically designed for semantic segmentation.

Human pose estimation (HPE) is a commonly used technique to determine derived parameters that are important to improve the performance of athletes in many sports disciplines. This paper proposes two methods to fine-tune a HPE system trained on general poses to a sports discipline specific HPE model using only a few labeled images. We show that 50 labeled 2D poses and additionally unlabeled videos are sufficient to achieve a Percentage of Correct Keypoints (PCK) of 88.6% at a threshold of 0.1 in the disciplines of triple and long jump, closing the gap between the supervised fine-tuning on the same 50 images and the fully supervised training on $60 \times$ more images by 60%. The first proposed method uses pseudo labels as a self-supervised training technique together with a filtering method of the pseudo labels. Furthermore, this paper shows that a mean teacher approach, which is based on consistency between a teacher and a student model, can also improve the results.

Introduction Lightening the injection burden is commonly believed to improve prophylaxis adherence. Efmoroctocog alfa (rFVIIIFc) is the first recombinant FVIII-Fc fusion protein available in France. This clotting factor with an extended half-life could thus improve medication adherence. Aim The study primarily aimed to assess the real-life impact on prophylaxis adherence of haemophilia A patients, when switching from a standard to an extended half-life FVIII. Methods This study was an observational, monocentre, non-interventional study aiming at assessing haemophilia A patients' real-life adherence during the first-year post-rFVIIIFc prophylaxis initiation. Medication adherence was assessed using two methods: the medication possession ratio (MPR), which is based on the hospital pharmacy dispensing data, and self-reported VERITAS-Pro® questionnaire. Patients on rFVIIIFc prophylaxis for at least 12 months, following a 12-month standard FVIII prophylaxis, were eligible for inclusion. Results In 2019, 47 male patients were undergoing rFVIIIFc prophylaxis in our Hemophilia Center, among which 36 meeting the inclusion criteria. Switching from standard to extended half-life FVIII prophylaxis resulted in increased mean dosing, while the mean number of weekly prophylactic injections (2.6 ± 0.5 vs 1.8 ± 0.3) decreased. Following rFVIIIFc initiation, a non-significant increase in median MPR occurred and the self-reported VERITAS-Pro® questionnaire demonstrated improved adherence to rFVIIIFc prophylaxis. Comparing adherent and non-adherent patients revealed age as the only factor likely to impact adherence (p = .07). Conclusion Our patient cohort exhibited high adherence levels before and after FVIII switching, based on MPR and VERITAS-Pro® questionnaire. The latter is likely a useful tool to quantity prophylaxis adherence from a patient's perspective in daily use.

The usage of VR gear in mixed reality applications demands a high position and orientation accuracy of all devices to achieve a satisfying user experience. This paper investigates the system behaviour of the VR system HTC Vive Pro at a testing facility that is designed for the calibration of highly accurate positioning instruments like geodetic total stations, tilt sensors, geodetic gyroscopes or industrial laser scanners. Although the experiments show a high reproducibility of the position readings within a few millimetres, the VR system has systematic effects with magnitudes of several centimetres. A tilt of about 0.4∘ of the reference plane with respect to the horizontal plane was detected. Moreover, our results demonstrate that the tracking algorithm faces problems when several lighthouses are used.

International audience

Global Navigation Satellite Systems (GNSS) like GPS, Galileo, GLONASS and Beidou are used for monitoring of civil structures and natural phenomena since the 1980s. However, most of the implementations were either short term or realized in a very controlled environment. In this article we report about the experiences of more than 20 years of GNSS based monitoring of a large-scale object. In 1999 the GNSS monitoring system was developed and installed at a deep-seated gravitational mass movement in Austria and it is still operating. The system gradually evolved during the years of operation to maintain a reliable operation and to take new developments of information and communication technology (ICT) into account. This included the change from radio communication to UTMS and continuous maintenance of the hardware (revision of GNSS stations, revision of solar collectors, exchange of batteries) as well as software maintenance (Firmware updates of GNSS receivers, software updates, change to new operating systems). Especially the constant changes in the used software system results in maintenance time, which is often underestimated. From these experiences we derive lessons learned which can be a guideline for other long-term GNSS based monitoring systems.

The COVID-19 pandemic has modified practice for patients with symptomatic aortic stenosis and could result in higher mortality due to treatment delays. In this clinical case series, three patients underwent ambulatory transcatheter aortic valve replacement (TAVR) thanks to patient and entourage willingness, careful patient selection (including a prior history of permanent pacemaker) and a minimalist procedural approach. No complications occurred during the 30-day follow-up. Performing ambulatory TAVR could reduce the clinical consequences of wait times, minimize exposure to coronavirus contamination and reduce the use of hospital resources that might be needed for COVID-19 patients., We report a case series of patients undergoing ambulatory transcatheter aortic valve replacement thanks to a careful patient selection (including a prior history of permanent pacemaker) and a minimalist procedural approach. This strategy could reduce the clinical consequences of wait times, minimize exposure to coronavirus contamination and reduce the use of hospital resources that might be needed for COVID-19 patients.

In this paper we address the problem of motion event detection in athlete recordings from individual sports. In contrast to recent end-to-end approaches, we propose to use 2D human pose sequences as an intermediate representation that decouples human motion from the raw video information. Combined with domain-adapted athlete tracking, we describe two approaches to event detection on pose sequences and evaluate them in complementary domains: swimming and athletics. For swimming, we show how robust decision rules on pose statistics can detect different motion events during swim starts, with a F1 score of over 91% despite limited data. For athletics, we use a convolutional sequence model to infer stride-related events in long and triple jump recordings, leading to highly accurate detections with 96% in F1 score at only +/- 5ms temporal deviation. Our approach is not limited to these domains and shows the flexibility of pose-based motion event detection., Accepted at 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshop (CVPRW)

In this paper we address the problem of dynamic scene motion deblurring. We present a model that combines high-resolution processing with a multi-resolution feature aggregation method for single frame and video deblurring. Our proposed model consists of 2 stages. In the first stage, single image deblurring is performed at a very high-resolution. For this purpose, we propose a novel network building block that employs multiple atrous convolutions in parallel. We carefully tune the atrous rate of each of these convolutions to achieve complete coverage of a rectangular area of the input. In this way we obtain a large receptive field at a high spatial resolution. The second stage aggregates information across multiple consecutive frames of a video sequence. Here we maintain a high-resolution, but also use multi-resolution features to mitigate the effects of large movements of objects between images. The presented models rank first and fourth in the NTIRE2020 challenges for single image deblurring and video deblurring, respectively. We apply our framework on current benchmarks and challenges and show that our model provides state-of-the art results.

The thrombin generation (TG) assay evaluates haemostatic balance, which is influenced by the levels of many coagulation factors and inhibitors. Our objective was to identify the determinant factors of TG in haemophilia A (HA) and haemophilia B (HB) patients and to compare them to those in healthy controls. Coagulation factor and inhibitor levels, and TG, were measured in platelet-poor plasma from 40 patients with HA, 32 patients with HB and 40 healthy subjects. Data were analysed using multiple regression models. In HA patients, factor VIII was a positive determinant of endogenous thrombin potential (ETP) and peak, whereas tissue factor pathway inhibitor (TFPI) and factor V were negative determinants of ETP and peak. In HB patients, FIX was a positive determinant of ETP and peak, FVII being a positive determinant of peak. Antithrombin and protein S (PS) were negative determinants of ETP while FX was a negative determinant of peak. Above all, in HB patients, TFPI was a negative determinant of ETP and peak. In healthy subjects, FVIII was a positive determinant of ETP and peak, whereas FX and protein S were negative determinants of these parameters. TFPI was not a negative determinant of either peak or ETP. In haemophilic patients, the determinant factors of TG are all implicated in FXa generation and inhibition, the crucial determinant factor being TFPI whatever the type of haemophilia, A or B. These findings contribute to the rationale that recently place TFPI as a target for innovative therapies of haemophilia.

This paper presents a model that robustly estimates important flight parameters for ski jumpers during their flight phase based on several camera views from the side along the jumpers’ typical flight trajectories. A convolutional neural network for pose estimation, but also trained to detect skis, serves as a base model. It identifies 98.0% of the relevant flight parameters correctly within an angle threshold of 5 degrees, improving by 11.6% over previous work. In postprocessing, a pose checker first removes all wrong poses by using comparisons of distances and relative positions of the detected keypoints. A second step executes two RANSAC variants. One robustly estimates the average pose and another one the average pose angles. This model lifts the detection performance to 99.3% of the relevant flight parameters within a threshold of 5 degrees.

Civil structural health monitoring (CSHM) has become significantly more important within the last decades due to rapidly growing construction volume worldwide as well as aging infrastructure and longer service lifetimes of the structures. The utilization of distributed fiber optic sensing (DFOS) allows the assessment of strain and temperature distributions continuously along the installed sensing fiber and is widely used for testing of concrete structures to detect and quantify local deficiencies like cracks. Relations to the curvature and bending behavior are however mostly excluded. This paper presents a comprehensive study of different approaches for distributed fiber optic shape sensing of concrete structures. Different DFOS sensors and installation techniques were tested within load tests of concrete beams as well as real-scale tunnel lining segments, where the installations were interrogated using fully-distributed sensing units as well as by fiber Bragg grating interrogators. The results point out significant deviations between the capabilities of the different sensing systems, but demonstrate that DFOS can enable highly reliable shape sensing of concrete structures, if the system is appropriately designed depending on the CSHM application.

As part of the Stuttgart 21 infrastructure project, German Railways is building a four-line railway bridge across the River Neckar in Stuttgart, work started in 2016. The Neckar railway bridge has an overall length of 345 m and has two abutments and six support axes. The structure crosses rail lines from the urban transit (Stadtbahn), the main trunk road B 10, inner city roads, the Neckar River, pedestrian and cycle ways as well as diverse public and private utilities. The bridge structure is located within the core mineral water protection zone, with its strict restrictions regarding the permissible depth of penetration into the subsoil and groundwater reservoirs. The mineral water springs lie within the construction site 26 m below the surface as artesian aquifers with an artesian pressure of ca. 10 m above ground level. From a groundwater management perspective, and following a program of five ground investigations and their evaluation, a raft foundation at the mid-level of the lower Keuper, built using cut-and-cover technique with a compressed air caisson, was chosen as the most economical foundation type for the three primary supports. A further programme of hydrochemical and geomechanical ground investigations was undertaken during the construction design phase. Following an evaluation of these additional ground parameters, it was possible to replace the compressed air caisson with a more economical deep foundation using large diameter piles. A management concept for possible scenarios was needed for the large diameter pile construction to achieve the required consent for exceptions from the ban on ground penetration in the area of the mineral springs.

Der vorliegende Fall beschreibt die fruhklinische Versorgung eines Patienten mit schwerer Methyl-endioxymetamphetamin-Intoxikation (Ecstasy). Es wird die Historie stimulierender psychotroper Substanzen und die Wirkweise moderner Psychostimulanzien dargestellt. Daten aus dem osterreichischen Drogenbericht zeigen die Verlagerung des Konsums „harter Drogen“ wie z. B. Opioiden hin zu Cannabis und Amphetaminderivaten. Der letale Verlauf im beschriebenen Fall zeigt das hohe Gefahrenpotenzial derartiger Substanzen und weist auf die Notwendigkeit rechtzeitiger, aggressiver intensivmedizinischer Therapie hin.

The importance of automated prism tracking is increasingly triggered by the rising automation of total station measurements in machine control, monitoring and one-person operation. In this article we summarize and explain the different techniques that are used to coarsely search a prism, to precisely aim at a prism, and to identify whether the correct prism is tracked. Along with the state-of-the-art review, we discuss and experimentally evaluate possible improvements based on the image data of an additional wide-angle camera which is available for many total stations today. In cases in which the total station’s fine aiming module loses the prism, the tracked object may still be visible to the wide-angle camera because of its larger field of view. The theodolite angles towards the target can then be derived from its image coordinates which facilitates a fast reacquisition of the prism. In experimental measurements we demonstrate that our image-based approach for the coarse target search is 4 to 10-times faster than conventional approaches.

Robotic total stations (RTS) are frequently used for the measurement of temperature induced bridge deformations or during load testing of bridges. In experimental setups, total stations have also been used for the measurement of dynamic bridge deformations. However, with standard configurations the measurement rate is not constant and on average an update rate of 7–10Hz can be achieved. This is not sufficient for the vibration monitoring of bridges considering their natural frequencies which are also in the same range. In this paper, we present different approaches to overcome these problems. In the first two approaches we demonstrate how the measurement rate to prisms can be increased to 20Hz to determine vertical deformations of bridges. Critical aspects like the measurement resolution of the automated target tracking and the correct sequence of steering commands are discussed. In another approach we demonstrate how vertical bridge vibrations can be measured using an image assisted total station (IATS) and corresponding processing techniques. The advantage of image-based methods is that structural features of a bridge like bolts can be used as targets. Therefore, no expensive prisms have to be mounted and access to the bridge is not required. All approaches are verified by laboratory investigations and their suitability is proven in a field experiment on a 74m long footbridge. In this field experiment the natural frequencies derived from the total station measurements are compared to the results of accelerometer measurements.

The current state-of-the-art in monocular 3D human pose estimation is heavily influenced by weakly supervised methods. These allow 2D labels to be used to learn effective 3D human pose recovery either directly from images or via 2D-to-3D pose uplifting. In this paper we present a detailed analysis of the most commonly used simplified projection models, which relate the estimated 3D pose representation to 2D labels: normalized perspective and weak perspective projections. Specifically, we derive theoretical lower bound errors for those projection models under the commonly used mean per-joint position error (MPJPE). Additionally, we show how the normalized perspective projection can be replaced to avoid this guaranteed minimal error. We evaluate the derived lower bounds on the most commonly used 3D human pose estimation benchmark datasets. Our results show that both projection models lead to an inherent minimal error between 19.3mm and 54.7mm, even after alignment in position and scale. This is a considerable share when comparing with recent state-of-the-art results. Our paper thus establishes a theoretical baseline that shows the importance of suitable projection models in weakly supervised 3D human pose estimation., Comment: Accepted at 3DV 2020

We propose Color Shift GAN(CSGAN), a method that allows learning to segment an object class without the need for pixel-wise annotations. We exploit a single textual annotation of the basic object color per image to learn the semantics of an object class. By using only a textual basic color annotation of each object, we are able to drastically reduce labeling efforts. We created a dataset of 29,910 images of cars and annotated the basic color of their body works. Our model reaches 61.4 percent IoU on our test data. CSGAN trained with additional 128 pixel-wise annotations reaches 62.0 percent. By adding 45,150 unlabeled images to the training of CSGAN we are able to increase IoU to 65.0 percent without using a single pixel-wise annotation. This verifies that our weak objective is sufficient for learning segmentation.

Emicizumab is a recombinant, humanized, bispecific, monoclonal antibody that bridges activated factor IX and factor X to restore the function of deficient factor VIII. Treatment of bleeding in patients with hemophilia and inhibitors involves the use of bypassing agents (BPA).These molecules are also used for the management of breakthrough bleeds in patients on prophylaxis with emicizumab, with increased concerns about the risks of combining two procoagulant drugs. Using thrombin generation assay, we tailored the dosage of activated prothrombin complex concentrate (APCC) and successfully treated an acute spontaneous arterial bleeding in a patient on prophylaxis with emicizumab 1.5mg.kg-1. Low dosages of APCC 15-25U.kg-1 were sufficient to normalize thrombin generation. The patient safely recovered with no thrombotic complications despite multiple infusions of APCC given during a 3-week period. We also showed that recombinant factor IX concentrates might represent an alternative to BPA in treating breakthrough bleeds in patients receiving prophylaxis with emicizumab.

In this paper, we present a method to automatically identify diseases from videos of gastrointestinal (GI) tract examinations using a Deep Convolutional Neural Network (DCNN) that processes images from digital endoscopes. Our goal is to aid domain experts by automatically detecting abnormalities and generating a report that summarizes the main findings. We have implemented a model that uses two different DCNN architectures to generate our predictions, which are also capable of running on a mobile device. Using this architecture, we are able to predict findings on individual images. Combined with class activations maps (CAM), we can also automatically generate a textual report describing a video in detail while giving hints about the spatial location of findings and anatomical landmarks. Our work shows one way to use a multi-disease detection pipeline to also generate video reports that summarize key findings.

In this paper we address the problem of automatic event detection in athlete motion for automated performance analysis in athletics. We specifically consider the detection of stride-, jump- and landing related events from monocular recordings in long and triple jump. Existing work on event detection in sports often uses manually designed features on body and pose configurations of the athlete to infer the occurrence of events. We present a two-step approach, where temporal 2D pose sequences extracted from the videos form the basis for learning an event detection model. We formulate the detection of discrete events as a sequence translation task and propose a convolutional sequence network that can accurately predict the timing of event occurrences. Our best performing architecture achieves a precision/recall of 92.3%/89.0% in detecting start and end of ground contact during the run-up and jump of an athlete at a temporal precision of +/- 1 frame at 200Hz. The results show that 2D pose sequences are a suitable motion representation for learning event detection in a sequence-to-sequence framework.

Acquired von Willebrand syndrome (AVWS) is a rare bleeding disorder. We report herein a case of AVWS due to a monoclonal gammopathy of undetermined significance, in which a transient but prolonged response to a treatment by intravenous immunoglobulin (IVIG) was observed. The diagnosis was fortuitously made in a preoperative setting for neurosurgery, after biological exploration of an isolated prolonged activated partial thromboplastin time. AVWS was confirmed by an accelerated clearance of an infused plasma-derived von Willebrand factor (VWF) concentrate. High doses of IVIG were used to perform the neurosurgery. Fifty-four days after IVIG, the patient was still responding to treatment with normal levels of factor VIII and VWF.

The automated extraction of kinematic parameters from athletes in video footage allows for direct training feedback and continuous quantitative assessment of an athlete's performance. Recent developments in the field of deep learning enable the measurement of kinematic coefficients directly from human pose estimates. However, the detection quality decreases while errors and noise increase with the complexity of the scene. In aquatic training scenarios, for instance, continuous pose estimation suffers from several orthogonal errors like switched joint predictions between the left and right sides of the body. In this paper, we analyze different error modes and present a rectification pipeline for improving the pose predictions using merely joint coordinates. We show experimentally that joint rectification equally improves the detection of key-poses, which are essential for a continuous qualitative performance assessment and pose retrieval, as well as posture visualization for quantitative training feedback.

Automatically generating descriptive captions for images is a well-researched area in computer vision. However, existing evaluation approaches focus on measuring the similarity between two sentences disregarding fine-grained semantics of the captions. In our setting of images depicting persons interacting with branded products, the subject, predicate, object and the name of the branded product are important evaluation criteria of the generated captions. Generating image captions with these constraints is a new challenge, which we tackle in this work. By simultaneously predicting integer-valued ratings that describe attributes of the human-product interaction, we optimize a deep neural network architecture in a multi-task learning setting, which considerably improves the caption quality. Furthermore, we introduce a novel metric that allows us to assess whether the generated captions meet our requirements (i.e., subject, predicate, object, and product name) and describe a series of experiments on caption quality and how to address annotator disagreements for the image ratings with an approach called soft targets. We also show that our novel clause-focused metrics are also applicable to other image captioning datasets, such as the popular MSCOCO dataset., 6 pages, accepted at MIPR 2019

In mountainous regions, the usability of roads and railway tracks strongly depends on the safety and the condition of retaining structures. Consequently, an early detection and identification of structural degradation, damages and potential failure mechanisms is of major interest. Currently, single point measurements using total stations or other sensors depict the state-of-the-art in deformation monitoring of retaining structures. Due to the large number of such structures in mountainous regions, e.g. the Alps, establishing a monitoring system for every object is unfeasible. This paper presents an approach for a large-scale deformation monitoring with a mobile mapping system (MMS). A MMS with two highquality laser scanners has been installed on the roof of a standard car. A newly developed algorithm processes the gathered data in a partially automated manner in order to perform deformation analysis on the one hand and to detect structural deficiencies (e.g. concrete spalling) on the other hand. Data of multiple (periodical) measurement campaigns of selected retaining structures have been used to evaluate the proposed approach. It is demonstrated, that mobile mapping in combination with targeted processing algorithms is a promising, efficient and comprehensive alternative to traditional static, single point monitoring solutions.

BACKGROUND Transcatheter aortic valve replacement (TAVR) has revolutionized management of high-risk patients with severe aortic stenosis. However, survival and the incidence of severe complications have been assessed in relatively small populations and/or with limited follow-up. OBJECTIVES This report details late clinical outcome and its determinants in the FRANCE-2 (FRench Aortic National CoreValve and Edwards) registry. METHODS The FRANCE-2 registry prospectively included all TAVRs performed in France. Follow-up was scheduled at 30 days, at 6 months, and annually from 1 to 5 years. Standardized VARC (Valve Academic Research Consortium) outcome definitions were used. RESULTS A total of 4,201 patients were enrolled between January 2010 and January 2012 in 34 centers. Approaches were transarterial (transfemoral 73%, transapical 18%, subclavian 6%, and transaortic or transcarotid 3%) or, in 18% of patients, transapical. Median follow-up was 3.8 years. Vital status was available for 97.2% of patients at 3 years. The 3-year all-cause mortality was 42.0% and cardiovascular mortality was 17.5%. In a multivariate model, predictors of 3-year all-cause mortality were male sex (p = 2 of 4 (p < 0.001). Severe events according to VARC criteria occurred mainly during the first month and subsequently in < 2% of patients/year. Mean gradient, valve area, and residual aortic regurgitation were stable during follow-up. CONCLUSIONS The FRANCE-2 registry represents the largest database available on late results of TAVR. Late mortality is largely related to noncardiac causes. Incidence rates of severe events are low after the first month. Valve performance remains stable over time. (J Am Coll Cardiol 2016; 68: 1637-47) (C) 2016 by the American College of Cardiology Foundation.

We present the case of an accidental extubation in a prone-positioned patient with a challenging airway because of Klippel-Feil syndrome and previous cervical spine fusions. The surgical procedure was well underway when this occurred, which added substantially to the difficulties produced by this event. We herein highlight the corrective steps we took in our case. We also recommend the need for a comprehensive preoperative briefing with all operating room personnel together with an action plan for how to prevent this particular scenario.

With the development of ambulatory surgery, there may be questions about the legal risk of this procedure. Indeed, the discharge of the patient from the hospital on the same day as the medical treatment raises the problem of monitoring and supervising potential complications, with a substantial delay in medical care, and the anaesthesiologists can be confronted with new areas of liability. This article specifies the French statutory and legal framework of the ambulatory surgery, and shows how the responsibility of the anaesthesiologist can be involved during patient care at all steps. The analysis of judicial precedent shows that the legal risk for the anaesthesiologist also exists in outpatient surgery. Surgery and anaesthesia are medical procedures involving a relatively high risk of damage for the patient. The damage can be attributed to malpractice from one or several health care professionals or to a medical complication (abnormal damage not related to malpractice and independent of past medical history of the patient). In the light of the ongoing and significant development in ambulatory surgery, there may be questions about the legal risk of this procedure. Indeed, the discharge of the patient from the hospital on the same day as the medical treatment raises the problem of monitoring and supervising potential complications, with a substantial delay in medical care. If the patient suffers any damage, the surgeon, the anaesthesiologist and in some cases, the hospital will have to answer in courts: the surgeon for the surgical procedure, the anaesthesiologist for the medical care and the hospital as the liable institution. After having specified the statutory framework of ambulatory surgery, we will see how the responsibility of the anaesthesiologist can be involved during patient care at all steps.