Search

Privacy regulation laws, such as GDPR, impose transparency and security as design pillars for data processing algorithms. In this context, federated learning is one of the most influential frameworks for privacy-preserving distributed machine learning, achieving astounding results in many natural language processing and computer vision tasks. Several federated learning frameworks employ differential privacy to prevent private data leakage to unauthorized parties and malicious attackers. Many studies, however, highlight the vulnerabilities of standard federated learning to poisoning and inference, thus raising concerns about potential risks for sensitive data. To address this issue, we present SGDE, a generative data exchange protocol that improves user security and machine learning performance in a cross-silo federation. The core of SGDE is to share data generators with strong differential privacy guarantees trained on private data instead of communicating explicit gradient information. These generators synthesize an arbitrarily large amount of data that retain the distinctive features of private samples but differ substantially. In this work, SGDE is tested in a cross-silo federated network on images and tabular datasets, exploiting beta-variational autoencoders as data generators. From the results, the inclusion of SGDE turns out to improve task accuracy and fairness, as well as resilience to the most influential attacks on federated learning.

Real-time six degree-of-freedom pose estimation with ground vehicles represents a relevant and well studied topic in robotics, due to its many applications, such as autonomous driving and 3D mapping. Although some systems exist already, they are either not accurate or they struggle in real-time setting. In this paper, we propose a fast, accurate and modular LiDAR SLAM system for both batch and online estimation. We first apply downsampling and outlier removal, to filter out noise and reduce the size of the input point clouds. Filtered clouds are then used for pose tracking and floor detection, to ground-optimize the estimated trajectory. The availability of a pre-tracker, working in parallel with the filtering process, allows to obtain pre-computed odometries, to be used as aids when performing tracking. Efficient loop closure and pose optimization, achieved through a g2o pose graph, are the last steps of the proposed SLAM pipeline. We compare the performance of our system with state-of-the-art point cloud based methods, LOAM, LeGO-LOAM, A-LOAM, LeGO-LOAM-BOR and HDL, and show that the proposed system achieves equal or better accuracy and can easily handle even cases without loops. The comparison is done evaluating the estimated trajectory displacement using the KITTI and RADIATE datasets., This paper is currently under review

In the emerging high mobility Vehicle-to-Everything (V2X) communications using millimeter Wave (mmWave) and sub-THz, Multiple-Input Multiple-Output (MIMO) channel estimation is an extremely challenging task. At mmWaves/sub-THz frequencies, MIMO channels exhibit few leading paths in the space-time domain (i.e., directions or arrival/departure and delays). Algebraic Low-rank (LR) channel estimation exploits space-time channel sparsity through the computation of position-dependent MIMO channel eigenmodes leveraging recurrent training vehicle passages in the coverage cell. LR requires vehicles' geographical positions and tens to hundreds of training vehicles' passages for each position, leading to significant complexity and control signalling overhead. Here we design a DL-based LR channel estimation method to infer MIMO channel eigenmodes in V2X urban settings, starting from a single LS channel estimate and without needing vehicle's position information. Numerical results show that the proposed method attains comparable Mean Squared Error (MSE) performance as the position-based LR. Moreover, we show that the proposed model can be trained on a reference scenario and be effectively transferred to urban contexts with different space-time channel features, providing comparable MSE performance without an explicit transfer learning procedure. This result eases the deployment in arbitrary dense urban scenarios.

Detecting buildings in the surroundings of an urban vehicle and matching them to building models available on map services is an emerging trend in robotics localization for urban vehicles. In this paper, we present a novel technique, which improves a previous work by detecting building façade, their positions, and finding the correspondences with their 3D models, available in OpenStreetMap. The proposed technique uses segmented point clouds produced using stereo images, processed by a convolutional neural network. The point clouds of the façades are then matched against a reference point cloud, produced extruding the buildings’ outlines, which are available on OpenStreetMap (OSM). In order to produce a lane-level localization of the vehicle, the resulting information is then fed into our probabilistic framework, called Road Layout Estimation (RLE). We prove the effectiveness of this proposal, testing it on sequences from the well-known KITTI dataset and comparing the results concerning a basic RLE version without the proposed pipeline.

Background Primary cardiac tumors (PCT) are rare lesions but have the potential to cause significant morbidity if not timely treated. We reviewed our single‐center experience in the surgical treatment of PCT with a focus on the long‐term outcome. Methods From 2001 to 2020, 57 consecutive patients underwent surgical resection of PCT at our Institution. Data including the demographic characteristics, tumor histology, surgical procedure, and postoperative outcomes were collected and analyzed. Results Mean age at presentation was 63.6 ± 11.2 years, and 33 (57.9%) of the patients were female. A total of 55 (96.5%) subjects were diagnosed with benign cardiac tumor, while the remaining had malignant tumors. The most common histopathological type was myxoma. All patients survived to hospital discharge. Main postoperative complications were: acute kidney injury (n = 3), sepsis (n = 3), and stroke (n = 2). Mean follow‐up time was 9 ± 5.9 years. Long‐term mortality was 22.8% (13/57). No tumor recurrence was observed among survivors. There was a significant relationship between mortality and pathological characteristics of the tumor, and myxomas had higher survival rates. Conclusion Surgical treatment of PCT is a safe and highly effective strategy associated with excellent short‐term outcomes. Long‐term survival remains poor for primary malignant tumors of the heart.

Benchmarking via carefully designed competitions makes it possible to provide a common framework for the rigorous comparison of intelligent and autonomous systems; competitions may play the role of scientific experiments while being appealing both to researchers and to the general public thus promoting critical analysis of systems outside the labs. This paper describes our approach to benchmarking domestic service robots through organizing recurrent competitions under the European Robotics League. It details the tools and benchmarks designed to evaluate the performance of robots at task and functionality levels. In particular, the functionality benchmarks for object perception and navigation are described and an overview of the new benchmarks to appear in the league is presented.

INTRODUCTION The latest STS guidelines recommend concomitant atrial fibrillation (AF) ablation not only during mitral surgery (Class IA) but also during other-than-mitral cardiac surgery procedures (Class IB) in patients with preoperative AF. Conventional Cox-Maze III/IV procedures are performed on both atria (BA), but several studies reported excellent results with left atrial only (LA) ablations: the scope of this study is to compare the safety and efficacy of BA vs LA approach. METHODS AND RESULTS Pubmed, Scopus, and WOS were searched from inception to November 2018: 28 studies including 7065 patients and comparing the performance of BA vs LA approaches were identified: of these, 16 (57.1%) enrolled exclusively patients with non-paroxysmal AF forms, 10 (35.7%) focused on mitral surgery as main procedure, and 16 (57.1%) regarded patients undergone Cox-Maze with radiofrequency. The 6- and 12-months prevalence of sinus rhythm were higher in the BA group (OR, 1.37, CI, 1.09-1.73, P = .008 and OR, 1.37, CI, 0.99-1.88, P = .05 respectively). Permanent pacemaker (PPM) implantation (OR, 1.85, CI, 1.38-2.49, P

Veno-arterial extracorporeal membrane oxygenation (V-A ECMO) has been used to deal with life-threatening complications as well as back-up or active cardiovascular support during high-risk procedures in patients undergoing transcatheter aortic valve implantation (TAVI). PubMed and MEDLINE electronic databases were searched in order to identify studies with emergency or prophylactic V-A ECMO application in association with TAVI procedures. From November 2012 to November 2017, 14 relevant studies were identified that included 5,115 TAVI patients of whom 102 (2%) required V-A ECMO (22 prophylactically, 66 as an emergency and 14 without a reported indication). The reason for emergency V-A ECMO institution was detailed in 64 patients: left ventricle free wall rupture (n = 14), haemodynamic instability (n = 12), ventricular arrhythmias (n = 7), aortic annulus rupture (n = 6), coronary obstruction (n = 6), low left ventricular output (ejection fraction

Extracorporeal membrane oxygenation (ECMO) in the veno-arterial (VA) configuration is an established method for the treatment of refractory cardiogenic shock. Such a condition characterizes the postoperative course of approximatively 1% of cardiac surgery patients. Although some studies have reported ECMO-related short-term results, little is known about the long-term outcomes of VA-ECMO therapy in the post-cardiotomy setting. Therefore, an extensive literature search was conducted regarding articles published after 1990 reporting postoperative ECMO use. PubMed, EMBASE and Web of Science were searched for sources. In-hospital mortality was high in post-cardiotomy VA-ECMO patients, ranging from 24.8% to 52%. Long-term results were poorly reported. However, based on the limited information available, hospital survivors showed a favorable outcome, with improvement in overall clinical condition, quality of life and limited hospital readmission for cardiac-related events. To conclude, in-hospital outcome in post-cardiotomy ECMO is often unfavorable, post-discharge results show satisfactory condition, with stable improvement of overall patient clinical status and low rate of hospital readmission and cardiac-related adverse events. Data reporting is, however, scarce and hence new and detailed studies are still warranted to investigate such aspects.

Study objective Viscoelastic point-of-care (POC) tests are commonly used to provide prompt diagnosis of coagulopathy and allow targeted treatments in bleeding patients on ECMO. We evaluated the clinical effectiveness of point-of-care (POC) testing for anticoagulation management in patients on extracorporeal membrane oxygenation (ECMO). Design Systematic review and meta-analysis. Eligible studies evaluating the use of thromboelastography- or thromboelastometry-guided algorithms, anti-factor Xa and platelet function testing were selected after screening the literature from July 1975 to January 2020. Setting Patients on ECMO support. Patients Anticoagulation management on ECMO patients. Interventions Rotational thromboelastometry, thromboelastography, alone or combined with platelet function testing. Trials monitoring the anticoagulation effects during ECMO using an anti-factor Xa assay were included in the systematic review. Measurements The primary outcomes were bleeding events, surgical revisions, thrombosis events and ECMO circuit change/failure. Secondary outcomes were blood-product transfusions, cerebrovascular accidents, mortality on ECMO, ECMO duration, intensive care unit and hospital discharge rates, and in-hospital mortality. Main results Thirty-one trials enrolling 1684 participants were included in the systematic review. Four trials enrolling 547 subjects were included in the meta-analysis. The use of a POC testing device resulted in improved detection of surgical bleeding (RR: 0.68, 95% CI 0.49 to 0.94, I2 = 0%; χ2 test for heterogeneity, P = 0.02). The use of POC-guided algorithms did not affect bleeding (RR:0.78, 95% CI 0.58 to 1.04, I2 = 47%; χ2 test for heterogeneity, P = 0.09), thrombosis events (RR:1.35, 95% CI 0.86 to 2.12, I2 = 37%; χ2 test for heterogeneity, P = 0.19), or ECMO circuit/change (RR:0.90, 95% CI 0.48 to 1.71, I2 = 28%; χ2 test for heterogeneity, P = 0.75). Conclusion Routine use of POC tests did not improve the main clinical outcomes beyond suggesting a diagnosis of surgical bleeding in ECMO patients.

One of the main components of an autonomous vehicle is the obstacle detection pipeline. Most prototypes, both from research and industry, rely on lidars for this task. Pointcloud information from lidar is usually combined with data from cameras and radars, but the backbone of the architecture is mainly based on 3D bounding boxes computed from lidar data. To retrieve an accurate representation, sensors with many planes, e.g., greater than 32 planes, are usually employed. The returned pointcloud is indeed dense and well defined, but high-resolution sensors are still expensive and often require powerful GPUs to be processed. Lidars with fewer planes are cheaper, but the returned data are not dense enough to be processed with state of the art deep learning approaches to retrieve 3D bounding boxes. In this paper, we propose two solutions based on occupancy grid and geometric refinement to retrieve a list of 3D bounding boxes employing lidar with a low number of planes (i.e., 16 and 8 planes). Our solutions have been validated on a custom acquired dataset with accurate ground truth to prove its feasibility and accuracy.

Event cameras are novel bio-inspired sensors, which asynchronously capture pixel-level intensity changes in the form of "events". The innovative way they acquire data presents several advantages over standard devices, especially in poor lighting and high-speed motion conditions. However, the novelty of these sensors results in the lack of a large amount of training data capable of fully unlocking their potential. The most common approach implemented by researchers to address this issue is to leverage simulated event data. Yet, this approach comes with an open research question: how well simulated data generalize to real data? To answer this, we propose to exploit, in the event-based context, recent Domain Adaptation (DA) advances in traditional computer vision, showing that DA techniques applied to event data help reduce the sim-to-real gap. To this purpose, we propose a novel architecture, which we call Multi-View DA4E (MV-DA4E), that better exploits the peculiarities of frame-based event representations while also promoting domain invariant characteristics in features. Through extensive experiments, we prove the effectiveness of DA methods and MV-DA4E on N-Caltech101. Moreover, we validate their soundness in a real-world scenario through a cross-domain analysis on the popular RGB-D Object Dataset (ROD), which we extended to the event modality (RGB-E)., Accepted at IROS21

Background. Ventricular septal rupture (VSR) is a rare but life-threatening complication after acute myocardial infarction. Although surgical correction is challenging and associated with high mortality, it remains the treatment of choice. This systematic review and meta-analysis aimed to evaluate the early outcome of surgical VSR repair.Methods. We searched electronic databases from January 1998 to February 2020. Studies reporting patients undergoing surgical treatment for VSR were analyzed. The primary outcome assessed was operative mortality. Differences were expressed as odds ratios (ORs) with 95% confidence intervals (CIs) to assess the relationships of predefined surgical variables and clinical prognosis.Results. A total of 6361 adult patients from 41 studies were identified. Operative mortality was 38.2%. Pooled ORs showed increased odds of operative mortality in patients with preoperative or perioperative intraaortic balloon pump insertion (OR = 3.48; 95% CI, 3.01-4.02; P >= .001), right ventricular dysfunction (OR = 2.85; 95% CI, 1.47-5.52; P = .002), posterior VSR (OR = 1.73; 95% CI, 1.30-2.31; P >= .001), and emergency surgery (OR = 3.79; 95% CI, 2.52-5.72; P >= .001). Temporal trend evaluation revealed no difference over time in the operative mortality rate; it was 34% in both time-related groups (1971-2000 versus 2001-2018).Conclusions. Ventricular septal rupture repair has a high operative mortality. Patients with preoperative or perioperative intraaortic balloon pump support, right ventricular dysfunction at presentation, or posterior defects, and those undergoing emergent VSR correction have increased odds of operative mortality. (C) 2021 by The Society of Thoracic Surgeons. Published by Elsevier Inc.

Background In-depth analysis of regulation networks of genes aberrantly expressed in cancer is essential for better understanding tumors and identifying key genes that could be therapeutically targeted. Results We developed a quantitative analysis approach to investigate the main biological relationships among different regulatory elements and target genes; we applied it to Ovarian Serous Cystadenocarcinoma and 177 target genes belonging to three main pathways (DNA REPAIR, STEM CELLS and GLUCOSE METABOLISM) relevant for this tumor. Combining data from ENCODE and TCGA datasets, we built a predictive linear model for the regulation of each target gene, assessing the relationships between its expression, promoter methylation, expression of genes in the same or in the other pathways and of putative transcription factors. We proved the reliability and significance of our approach in a similar tumor type (basal-like Breast cancer) and using a different existing algorithm (ARACNe), and we obtained experimental confirmations on potentially interesting results. Conclusions The analysis of the proposed models allowed disclosing the relations between a gene and its related biological processes, the interconnections between the different gene sets, and the evaluation of the relevant regulatory elements at single gene level. This led to the identification of already known regulators and/or gene correlations and to unveil a set of still unknown and potentially interesting biological relationships for their pharmacological and clinical use.

Neural Architecture Search (NAS) is the process of automating architecture engineering, searching for the best deep learning configuration. One of the main NAS approaches proposed in the literature, Progressive Neural Architecture Search (PNAS), seeks for the architectures with a sequential model-based optimization strategy: it defines a common recursive structure to generate the networks, whose number of building blocks rises through iterations. However, NAS algorithms are generally designed for an ideal setting without considering the needs and the technical constraints imposed by practical applications. In this paper, we propose a new architecture search named Pareto-Optimal Progressive Neural Architecture Search (POPNAS) that combines the benefits of PNAS to a time-accuracy Pareto optimization problem. POPNAS adds a new time predictor to the existing approach to carry out a joint prediction of time and accuracy for each candidate neural network, searching through the Pareto front. This allows us to reach a trade-off between accuracy and training time, identifying neural network architectures with competitive accuracy in the face of a drastically reduced training time.

Skeleton-based human action recognition has achieved a great interest in recent years, as skeleton data has been demonstrated to be robust to illumination changes, body scales, dynamic camera views, and complex background. Nevertheless, an effective encoding of the latent information underlying the 3D skeleton is still an open problem. In this work, we propose a novel Spatial-Temporal Transformer network (ST-TR) which models dependencies between joints using the Transformer self-attention operator. In our ST-TR model, a Spatial Self-Attention module (SSA) is used to understand intra-frame interactions between different body parts, and a Temporal Self-Attention module (TSA) to model inter-frame correlations. The two are combined in a two-stream network which outperforms state-of-the-art models using the same input data on both NTU-RGB+D 60 and NTU-RGB+D 120., Accepted as ICPRW2020 (FBE2020, Workshop on Facial and Body Expressions, micro-expressions and behavior recognition) 8 pages, 2 figures. arXiv admin note: substantial text overlap with arXiv:2008.07404

The aim of ACRE (Agri-food Competition for Robot Evaluation) is to provide a set of benchmarks for agricultural robots and smart implements. While involving capabilities of general application, ACRE puts a special focus on weeding, identified as one of the tasks where it is easier for robotics to demonstrate its potential. ACRE, as the other three robot competitions that are being organised by European project METRICS, is built on the established idea of benchmarking through competitions. In this paper we present the framework of ACRE and examples of its benchmarks, https://youtu.be/oTXDsJ03mA0

BackgroundDifficulties in handwriting, such as dysgraphia, impact several aspects of a child’s everyday life. Current methodologies for the detection of such difficulties in children have the following three main weaknesses: (1) they are prone to subjective evaluation; (2) they can be administered only when handwriting is mastered, thus delaying the diagnosis and the possible adoption of countermeasures; and (3) they are not always easily accessible to the entire community.ObjectiveThis work aims at developing a solution able to: (1) quantitatively measure handwriting features whose alteration is typically seen in children with dysgraphia; (2) enable their study in a preliteracy population; and (3) leverage a standard consumer technology to increase the accessibility of both early screening and longitudinal monitoring of handwriting difficulties.MethodsWe designed and developed a novel tablet-based app Play Draw Write to assess potential markers of dysgraphia through the quantification of the following three key handwriting laws: isochrony, homothety, and speed-accuracy tradeoff. To extend such an approach to a preliteracy age, the app includes the study of the laws in terms of both word writing and symbol drawing. The app was tested among healthy children with mastered handwriting (third graders) and those at a preliterate age (kindergartners).ResultsApp testing in 15 primary school children confirmed that the three laws hold on the tablet surface when both writing words and drawing symbols. We found significant speed modulation according to size (P0.28, 12 degrees of freedom). Importantly, the three laws were verified on symbols among 19 kindergartners. Results from the speed-accuracy exercise showed a significant evolution with age of the global movement time (circle: P=.003, square: PConclusionsThe designed app represents a promising solution for the screening of handwriting difficulties, since it allows (1) anticipation of the detection of alteration of handwriting principles at a preliteracy age and (2) provision of broader access to the monitoring of handwriting principles. Such a solution potentially enables the selective strengthening of lacking abilities before they exacerbate and affect the child’s whole life.

Tako-tsubo cardiomyopathy (TC) is characterized by acute but transient ventricular dysfunction without obstructive coronary artery disease, generally precipitated by emotional and physical triggers. We describe this syndrome in a 76-year-old woman who was admitted with thoracic pain secondary to TC as shown by echocardiographic assessment, with a concurrent diagnosis of giant ascending aortic aneurism. Surgical intervention was delayed to allow ventricular recovery and then to perform ascending aorta replacement. An individualized perioperative approach was applied to avoid a possible TC recurrence with an uneventful postoperative course.

Extracorporeal life support (ECLS) is indicated in refractory acute respiratory or cardiac failure. According to the need for anticoagulation, bleeding conditions (e.g., in trauma, pulmonary bleeding) have been considered a contraindication for the use of ECLS. However, there is increasing evidence for improved outcomes after ECLS support in hemorrhagic patients based on the benefits of hemodynamic support outweighing the increased risk of bleeding. We conducted a systematic literature search according to the PRISMA guidelines and reviewed publications describing ECLS support in hemorrhagic conditions. Seventy-four case reports, four case series, seven retrospective database observational studies, and one preliminary result of an ongoing study were reviewed. In total, 181 patients were identified in total of 86 manuscripts. The reports included patients suffering from bleeding caused by pulmonary hemorrhage (n = 53), trauma (n = 96), postpulmonary endarterectomy (n = 13), tracheal bleeding (n = 1), postpartum or cesarean delivery (n = 11), and intracranial hemorrhage (n = 7). Lower targeted titration of heparin infusion, heparin-free ECLS until coagulation is normalized, clamping of the endotracheal tube, and other ad hoc possibilities represent potential beneficial maneuvers in such conditions. Once the patient is cannulated and circulation restored, bleeding control surgery is performed for stabilization if indicated. The use of ECLS for temporary circulatory or respiratory support in critical patients with refractory hemorrhagic shock appears feasible considering tailored ECMO management strategies. Further investigation is needed to better elucidate the patient selection and ECLS management approaches.

BACKGROUND Difficulties in handwriting, such as dysgraphia, impact several aspects of a child’s everyday life. Current methodologies for the detection of such difficulties in children have the following three main weaknesses: (1) they are prone to subjective evaluation; (2) they can be administered only when handwriting is mastered, thus delaying the diagnosis and the possible adoption of countermeasures; and (3) they are not always easily accessible to the entire community. OBJECTIVE This work aims at developing a solution able to: (1) quantitatively measure handwriting features whose alteration is typically seen in children with dysgraphia; (2) enable their study in a preliteracy population; and (3) leverage a standard consumer technology to increase the accessibility of both early screening and longitudinal monitoring of handwriting difficulties. METHODS We designed and developed a novel tablet-based app Play Draw Write to assess potential markers of dysgraphia through the quantification of the following three key handwriting laws: isochrony, homothety, and speed-accuracy tradeoff. To extend such an approach to a preliteracy age, the app includes the study of the laws in terms of both word writing and symbol drawing. The app was tested among healthy children with mastered handwriting (third graders) and those at a preliterate age (kindergartners). RESULTS App testing in 15 primary school children confirmed that the three laws hold on the tablet surface when both writing words and drawing symbols. We found significant speed modulation according to size (PP2>0.28, 12 degrees of freedom). Importantly, the three laws were verified on symbols among 19 kindergartners. Results from the speed-accuracy exercise showed a significant evolution with age of the global movement time (circle: P=.003, square: PP=.001), the goodness of fit of the regression between movement time and accuracy constraints (square: PP=.02), and the index of performance (square: P CONCLUSIONS The designed app represents a promising solution for the screening of handwriting difficulties, since it allows (1) anticipation of the detection of alteration of handwriting principles at a preliteracy age and (2) provision of broader access to the monitoring of handwriting principles. Such a solution potentially enables the selective strengthening of lacking abilities before they exacerbate and affect the child’s whole life.

Background The use of peripheral veno-arterial extracorporeal life support (V-A ECLS) as a mechanical circulatory support in cardiogenic shock has increased dramatically over the last years. However, increased afterload may jeopardize left ventricle (LV) recovery and cause blood stasis and pulmonary edema. Therefore, several LV unloading techniques have been developed and used with limited understanding of the actual difference among them. The aim of the present study was to compare a trans-aortic suction device (Impella) and pulmonary artery (PA) drainage, for LV unloading and V-A ECLS management as well as efficacy in a porcine cardiogenic shock (CS) model Methods A dedicated CS model compared included twelve female swine (21± 1,8-weeks old and weighing 54,3 ± 4,6 kg) supported with V-A ECLS and randomized to Impella or PA-related LV drainage. LV unloading and end-organ perfusion were evaluated through the pulmonary artery catheter and the LV pressure/volume analysis. All the variables were collected at baseline, profound CS, V-A ECLS support with maximum flow and when Impella or PA cannula run on top. Results CS was successfully induced in all twelve animals. Impella resulted in a marked drop of LVEDV compared to a slight decrease in the PA cannula group, resulting in an overall stroke work (SW) and Pressure-Volume Area (PVA) reductions with both techniques. However, SW reduction was significant in the Impella CP group (VA ECMO 3998.82027.6 mmHg x mL vs VAECMO + Impella 1796.9±1033.9 mmHg x ml, p value 0,016), leading to a more consistent PVA reduction (Impella reduction 34,7% vs PA cannula reduction 9,7%) In terms of end organ perfusion, central and mixed O 2 saturation improved with V-A ECLS, and subsequently, remaining unchanged with either Impella or PA cannula as unloading strategy Conclusions Trans-aortic suction and PA drainage provided effective LV unloading during V-A ECLS while maintaining adequate end-organ perfusion. Trans-aortic suction device provides a greater LV unloading effect and reduces more effectively the total LV stroke work.

Cardiogenic shock (CS) is a challenging syndrome, associated with significant morbidity and mortality. Although pharmacological therapies are successful and can successfully control this acute cardiac illness, some patients remain refractory to drugs. Therefore, a more aggressive treatment strategy is needed. Temporary mechanical circulatory support (TCS) can be used to stabilise patients with decompensated heart failure. In the last two decades, the increased use of TCS has led to several kinds of devices becoming available. However, indications for TCS and device selection are part of a complex process. It is necessary to evaluate the severity of CS, any early and prompt haemodynamic resuscitation, prior TCS, specific patient risk factors, technical limitations and adequacy of resources and training, as well as an assessment of whether care would be futile. This article examines options for commonly used TCS devices, including intra-aortic balloon pumps, a pulsatile percutaneous ventricular assist device (the iVAC), veno-arterial extra-corporeal membrane oxygenation and Impella (Abiomed) and TandemHeart (LivaNova) percutaneous ventricular assist device.

The authors sincerely apologise for the imperfections made during the collection of the data and wish to make the following corrections to this paper [...]

Mesh refinement is a fundamental step for accurate Multi-View Stereo. It modifies the geometry of an initial manifold mesh to minimize the photometric error induced in a set of camera pairs. This initial mesh is usually the output of volumetric 3D reconstruction based on min-cut over Delaunay Triangulations. Such methods produce a significant amount of non-manifold vertices, therefore they require a vertex split step to explicitly repair them. In this paper, we extend this method to preemptively fix the non-manifold vertices by reasoning directly on the Delaunay Triangulation and avoid most vertex splits. The main contribution of this paper addresses the problem of choosing the camera pairs adopted by the refinement process. We treat the problem as a mesh labeling process, where each label corresponds to a camera pair. Differently from the state-of-the-art methods, which use each camera pair to refine all the visible parts of the mesh, we choose, for each facet, the best pair that enforces both the overall visibility and coverage. The refinement step is applied for each facet using only the camera pair selected. This facetwise refinement helps the process to be applied in the most evenly way possible., Comment: Accepted as Oral ICPR2020

Dynamic Vision Sensors (DVSs) asynchronously stream events in correspondence of pixels subject to brightness changes. Differently from classic vision devices, they produce a sparse representation of the scene. Therefore, to apply standard computer vision algorithms, events need to be integrated into a frame or event-surface. This is usually attained through hand-crafted grids that reconstruct the frame using ad-hoc heuristics. In this paper, we propose Matrix-LSTM, a grid of Long Short-Term Memory (LSTM) cells that efficiently process events and learn end-to-end task-dependent event-surfaces. Compared to existing reconstruction approaches, our learned event-surface shows good flexibility and expressiveness on optical flow estimation on the MVSEC benchmark and it improves the state-of-the-art of event-based object classification on the N-Cars dataset., Comment: 23 pages, 6 figures

Recurrent neural networks (RNN) are being extensively exploited in industry to address complex predictive tasks by leveraging on the increased availability of data from processes. However, the rationale behind model response is encoded in an implicit way, which is difficult to be explained by practitioners. If revealed, such mechanisms could provide deeper insights into RNN execution, enhancing conventional performance evaluations. We propose a new approach based on the introduction of a model-based clustering layer, constraining the network to operate on a discrete latent state representation. By processing context-input conditioned transitions between clusters, a Moore Machine characterizing the RNN computations is extracted. The proposed approach is demonstrated on both synthetic experiments from an open benchmark problem and via the application to a pilot industrial plant, by the behavior cloning of the flexible conveyor of a Remanufacturing process. The finite-state RNN attains the prediction accuracy of RNN with continuous state, providing in addition a more interpretable structure.

Introduction: Phaeochromocytoma is a catecholamine-secreting tumour associated with clinical presentation ranging from paroxysmal hypertension to intractable cardiogenic shock. Extracorporeal life support, in veno-arterial mode, application in refractory acute heart dysfunction is sharply increasing worldwide. However, its clinical utility in phaeochromocytoma-induced cardiogenic shock remains still unclear. Methods: A systematic review of published reports was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement. Searches were accomplished on PubMed, Embase and Google Scholar to identify articles describing the use of extracorporeal life support in the setting of phaeochromocytoma-induced cardiogenic shock (PROSPERO: CRD42019125225). Results: Thirty-five reports, including 62 patients supported with extracorporeal life support because of intractable phaeochromocytoma crisis, were included for the analysis. Almost all the subjects underwent peripheral cannulation for extracorporeal life support. The median duration of the mechanical circulatory support was 5 days, and most of the patients recovered normal myocardial function (left ventricular ejection fraction ⩾50%). In-hospital survival was 87%. Phaeochromocytoma was removed surgically during extracorporeal life support in 10 patients (16%), while in the remaining after haemodynamic stabilization and weaning from the mechanical support. Conclusion: Successful management of phaeochromocytoma-induced cardiogenic shock depends on prompt recognition and immediate treatment of shock. In this scenario, extracorporeal life support may play a significant role allowing cardiac and end-organ recovery and giving time for accurate diagnosis and specific treatment.

Probabilistic electricity price forecast (EPF) systems represent a fundamental tool to achieve robust production scheduling and day-ahead bidding strategies. However, most EPF methods, including recently proposed deep learning based techniques, are still targeting point predictions, following the common Gaussian assumption. In this work, we propose a novel probabilistic EPF approach based on the integration of a Gaussian Mixture layer, parametrized by a Recurrent Neural Network with Gated Recurrent Units, including an L1-norm based feature selection mechanisms. The network is conceived to approximate general conditional price distributions through learning. Moreover, we developed a multi-hours prediction approach exploiting correlations and patters both in hourly and cross-hour contexts. Experiments have been performed on the Italian market dataset, showing the capability of the proposed method to achieve accurate out-of-sample predictions while providing explicit uncertainty indications supporting enhanced decision making.

Neural networks are being broadly explored for the identification of Industrial Cyber Physical Systems (ICPS) models from data sequences. However, learned representations typically lack explainability, representing a major challenge of deep learning. Interpreting the information structured across the synaptic links is particularly challenging for recurrent neural networks (RNN), encoding input features and observed system dynamics within a continuous latent space. In this work, we target the representations built within RNNs while learning behavioral models of a class of discrete dynamical systems. To this end, we propose a method to extract the symbolic knowledge structured by the continuous state, based on Gaussian Mixture Model clustering, handling latent activations characterized by partially overlapping and non-isotropic distributions. Experiments are performed on a pilot remanufacturing plant, by learning the model of a conveyor controller from process data. We show the capability of the proposed method to extract the hidden finite state machine from the trained RNN, providing a human interpretable representation of the input conditioned computations performed through the continuous latent space.

Background: Extracorporeal membrane oxygenation (ECMO) use in patients with active bleedings or traumatic injuries may be challenging because of the risk of refractory haemorrhage related to systemic anticoagulation (SA). Technological advancements, especially heparin coated circuits, allowed ECMO application with a mild or no anticoagulation regimen in some circumstances. We herein report a limited case-series of ECMO application without anticoagulation.Methods: ECMO without SA was established in 6 patients. Four of them received veno-arterial mode. Full heparin antagonization with protamine was provided in 3 post-cardiotomy cases. Tip-to-tip heparin-coated ECMO circuit was used in 4 patients, whereas heparin-coated tubing and no heparin-coated cannulas were applied for the remaining subjects.Results: Mean duration of support and pump flow were 10 +/- 4 hours and 3.0 +/- 0.7 litres respectively. No ECMO related complication or oxygenator failure occurred, nor was clotting of the circuit observed at inspection. All patients were weaned from ECMO and half of them survived and were discharged.Conolusions: Short ECMO duration without anticoagulation was feasible and provided effective cardiopulmonary support in patients with active bleeding or high hemorrhagic risk. Further investigations are needed to evaluate its benefits or limitations for a protracted time, in a wider range of clinical conditions and in larger patient populations.

Introduction: Lactate, a product of anaerobic metabolism, is a biomarker and indicator for tissue hypoperfusion and oxygen debt. An elevated blood lactate level has been associated with poor outcome in many clinical conditions, including cardiac surgery. Nevertheless, debate exists regarding which blood lactate concentration is most indicative of poor outcomes. We evaluate the impact of hyperlactatemia, defined as a peak arterial blood concentration ⩾2.0 mmol/L during cardiopulmonary bypass, on surgical results with a focus on long-term outcome. Methods: We reviewed 1,099 consecutive adult patients who underwent cardiac surgery on pump. The patients were divided into two groups based on the presence or not of hyperlactatemia. Pre- and intraoperative risk factors for hyperlactatemia were identified, and the postoperative outcome of patients with or without hyperlactatemia was compared. Results: Hyperlactatemia was present in 372 patients (33.8%). Factors independently associated with hyperlactatemia were urgent/emergency procedure, cardiopulmonary bypass duration and aortic cross-clamp time. Patients with hyperlactatemia had significantly higher rate of prolonged mechanical ventilation time, in-hospital stay and requirement of inotropes and intra-aortic balloon pump support (p Conclusion: Hyperlactatemia during cardiopulmonary bypass has a significant association with postoperative morbidity and mortality. Correction of risk factors for hyperlactatemia, together with prompt detection and correction of this condition, may control complications and improve outcome.

Dense 3D reconstruction of the surrounding environment is one the fundamental way of perception for Advanced Driver-Assistance Systems (ADAS). In this field, accurate 3D modeling finds applications in many areas like obstacle detection, object tracking, and remote driving. This task can be performed with different sensors like cameras, LiDARs, and radars. Each one presents some advantages and disadvantages based on the precision of the depth, the sensor cost, and the accuracy in adverse weather conditions. For this reason, many researchers have explored the fusion of multiple sources to overcome each sensor limit and provide an accurate representation of the vehicle’s surroundings. This paper proposes a novel post-processing method for accurate depth estimation, based on a patch-wise depth correction approach, to fuse data from LiDAR and stereo camera. This solution allows for accurate edges and object boundaries preservation in multiple challenging scenarios.

Understanding the roles and interplays of histone marks and transcription factors in the regulation of gene expression is of great interest in the development of non-invasive and personalized therapies. Computational studies at genome-wide scale represent a powerful explorative framework, allowing to draw general conclusions. However, a genome-wide approach only identifies generic regulative motifs, and possible multi-functional or co-regulative interactions may remain concealed. In this work, we hypothesize the presence of a number of distinct subpopulations of transcriptional regulative patterns within the set of protein coding genes that explain the statistical redundancy observed at a genome-wide level. We propose the application of a K-Plane Regression algorithm to partition the set of protein coding genes into clusters with specific shared regulative mechanisms. Our approach is completely data-driven and computes clusters of genes significantly better fitted by specific linear models, in contrast to single regressions. These clusters are characterized by distinct and sharper histonic input patterns, and different mean expression values.

Frequent occurrence of paravalvular leak (PVL) after transcatheter aortic valve replacement (TAVR) was the main concern with earlier-generation devices. Current meta-analysis compared outcomes of TAVR with next-generation devices: ACURATE neo and SAPIEN 3. In random-effects meta-analysis, the pooled incidence rates of procedural, clinical and functional outcomes according to VARC-2 definitions were assessed. One randomized controlled trial and five observational studies including 2818 patients (ACURATE neo n = 1256 vs. SAPIEN 3 n = 1562) met inclusion criteria. ACURATE neo was associated with a 3.7-fold increase of moderate-to-severe PVL (RR (risk ratio): 3.70 (2.04&ndash, 6.70), P &lt, 0.0001), which was indirectly related to higher observed 30-day mortality with ACURATE valve (RR: 1.77 (1.03&ndash, 3.04), P = 0.04). Major vascular complications, acute kidney injury, periprocedural myocardial infarction, stroke and serious bleeding events were similar between devices. ACURATE neo demonstrated lower transvalvular pressure gradients both at discharge (P &lt, 0.00001) and at 30 days (P &lt, 0.00001), along with lower risk of patient&ndash, prosthesis mismatch (RR: 0.29 (0.10&ndash, 0.87), P = 0.03) and pacemaker implantation (RR: 0.64 (0.50&ndash, 0.81), P = 0.0002), but no differences were observed regarding composite endpoints early safety and device success. In conclusion, ACURATE neo, as compared with SAPIEN 3, was associated with higher rates of moderate-to-severe PVL, which were indirectly linked with increased observed 30-day all-cause mortality.

The vehicle relative pose, referenced to road boundaries or lanes, is fundamental for an autonomous driving vehicle to perform motion planning. The estimate of relative pose includes the difference between the car heading angle and road center-line direction (i.e., the relative heading angle), together with the lateral displacement of the vehicle from the road center-line. This information can be equally derived by an inertial navigation system equipped with GPS receivers or by a vision system based on cameras. However, both solutions present some disadvantages. GPS-based estimates of heading angle shows loss of accuracy during low-speed manoeuvres or within tunnels. On the other hand, a line-fitting based estimation typical of vision systems suffers from lines accuracy, and they are hard to be implemented in an urban environment or high-curvature track scenarios. This work presents an innovative integrated algorithm that fuses those two approaches, providing vehicle relative pose with respect to road lane combining data coming from an inertial navigation system and a line detection algorithm. This integrated solution is then able to always feed the panning algorithm with an accurate estimate of the vehicle pose in multiple challenging scenarios.

A novel biomedical instrument for supporting the physician in Skin Prick Test analysis was designed, developed, characterized, and is now ready for clinical trials. Skin Prick Test is the gold standard front-end analysis for diagnosis of allergies in human subjects. The forearm skin is punctured with different allergens and the resulting reaction wheals are analyzed and compared to standard reaction, with larger areas corresponding to stronger allergy to the specific allergen. The wheals inspection and allergy diagnosis are performed, visually and subjectively, by the Medical Doctor. This procedure is laborious and somehow unreliable, being subject to variability both intra- and inter-operator because the doctor subjectivity in detecting and measuring the wheals is significant. Registration of the exam result is rarely available in a digital format, useful for data saving, transmission, retrieval and comparative analyses. Many of the above criticalities of the actual Prick Test manual practice are addressed and resolved by the proposed biomedical instrumentation that makes use of digital image-processing and data storage. In this work, we present a prototype of wheal measurement system, designed, developed and characterized to specifically measure geometry and areas of the allergic reaction wheals in Prick Test clinical exams. After software developments from previous version of the instrument, the wheal-meter now allows manual, semi-automatic, automatic, and fully-automatic operating conditions always providing digital exams output.

Field robotics is a fast developing research field, in particular precision agriculture is gaining in popularity due to the high return in productivity and the reduced pollution impact on the environment. The GRAPE project is an ECHORD++ robotic experiment aimed at the use of a mobile robot for automatic pheromone dispenser distribution in vineyards, to reduce pesticide use thanks to pheromone mate disruption. This work describes the autonomous navigation system of such robot. For the specific scenario a real state of the art does not exists, so we adapted techniques that have been designed for different problems, in particular classical methods for navigation and mapping in indoor environments. The vineyard environment is challenging because of many variability factors such as weather, soil and vegetation. These factors hinder the indoor methods introducing noise in the robot perceptions. To solve this problem, we propose a specific navigation system that takes advantage of multiple sensors: wheel encoders, inertial measurement unit (IMU), and GPS to filter the environment noise and accurately estimate the robot odometry. In addition, the system exploits a LIDAR sensor to localize the robot, through the Adaptive Monte Carlo Localization (AMCL) algorithm, and to map the vineyard. We tested the system in simulation with very good results which have been confirmed during field tests in a real vineyard.

Stuck-pipe phenomena can have disastrous effects on drilling performance, with outcomes that can range from time delays to loss of expensive machinery. In this work, we develop three indicators based on mudlog data, which aim to detect three different physical phenomena associated with the insurgence of a sticking. In particular, two indices target respectively the detection of translational and rotational motion issues, while the third index concerns the wellbore pressure. A statistical model that relates these features to documented stuck-pipe events is then developed using advanced machine learning tools. The resulting model takes the form of a depth-based map of the risk of incurring into a stuck-pipe, updated in real-time. Preliminary experimental results on the available dataset indicate that the use of the proposed model and indicators can help mitigate the stuck-pipe issue.

Skeleton-based Human Activity Recognition has achieved great interest in recent years as skeleton data has demonstrated being robust to illumination changes, body scales, dynamic camera views, and complex background. In particular, Spatial-Temporal Graph Convolutional Networks (ST-GCN) demonstrated to be effective in learning both spatial and temporal dependencies on non-Euclidean data such as skeleton graphs. Nevertheless, an effective encoding of the latent information underlying the 3D skeleton is still an open problem, especially when it comes to extracting effective information from joint motion patterns and their correlations. In this work, we propose a novel Spatial-Temporal Transformer network (ST-TR) which models dependencies between joints using the Transformer self-attention operator. In our ST-TR model, a Spatial Self-Attention module (SSA) is used to understand intra-frame interactions between different body parts, and a Temporal Self-Attention module (TSA) to model inter-frame correlations. The two are combined in a two-stream network, whose performance is evaluated on three large-scale datasets, NTU-RGB+D 60, NTU-RGB+D 120, and Kinetics Skeleton 400, consistently improving backbone results. Compared with methods that use the same input data, the proposed ST-TR achieves state-of-the-art performance on all datasets when using joints' coordinates as input, and results on-par with state-of-the-art when adding bones information., Accepted at Computer Vision and Image Understanding (CVIU) 12 pages, 8 figures

This paper presents the development of a MPC architecture for an autonomous wheelchair, that aims at reaching a desired goal while considering physical actuator limitations, obstacle avoidance and comfort constraints. The controller has been designed in such a way that it can be easily applied to an existing commercial wheelchair, without the need for a redesign of the on-board motor controllers. Moreover, thanks to the introduction of an inner feedback linearizing loop, a linear MPC has been considered, allowing for an efficient real-time implementation of the control system. The experimental results show the effectiveness of the proposal, and demonstrate that MPC can be considered a promising control approach to develop autonomous vehicles.

With the advent of Precision Agriculture we are going in the direction of a fourth agricultural revolution. In the future, crop works will be entirely automated by intelligent machines that will carry out all operations, from sowing to harvesting, leading to more efficient use of resources (like water and nutrients) and contamination reduction via a fine-tuned spraying of pesticides. One of the requirements for agricultural machines to accomplish in-field localized operations is to have a digital model of each single crop plant. This framework leads to the idea of Plant Digital Twin, that is, the computerized counterpart of a real plant. With this in mind, we developed a system that can scan crop rows reconstructing and distinguishing individual plants. The final goal is growth monitoring and treatments targeted to the needs of each specific plant., https://youtu.be/r8e6H76TeFk https://youtu.be/Jam0A0BDyN0