Despite their remarkable performance, deep neural networks remain unadopted in clinical practice, which is considered to be partially due to their lack of explainability. In this work, we apply explainable attribution methods to a pre-trained deep neural network for abnormality classification in 12-lead electrocardiography to open this "black box" and understand the relationship between model prediction and learned features. We classify data from two public databases (CPSC 2018, PTB-XL) and the attribution methods assign a "relevance score" to each sample of the classified signals. This allows analyzing what the network learned during training, for which we propose quantitative methods: average relevance scores over a) classes, b) leads, and c) average beats. The analyses of relevance scores for atrial fibrillation and left bundle branch block compared to healthy controls show that their mean values a) increase with higher classification probability and correspond to false classifications when around zero, and b) correspond to clinical recommendations regarding which lead to consider. Furthermore, c) visible P-waves and concordant T-waves result in clearly negative relevance scores in atrial fibrillation and left bundle branch block classification, respectively. Results are similar across both databases despite differences in study population and hardware. In summary, our analysis suggests that the DNN learned features similar to cardiology textbook knowledge.

Summary As cloud adoption increases, so do the number of available cloud service providers. Moving complex applications between clouds can be beneficial—or other times necessary—but achieving this so‐called cloud portability is rarely straightforward. This article presents the adoption of OASIS TOSCA, a standard in the declarative description of cloud applications, to encourage and facilitate cloud portability in MiCADO, an application‐level multi‐cloud orchestration and auto‐scaling framework. The interface to MiCADO is an Application Description Template, which draws from the TOSCA specification to describe an application in MiCADO. The generic design of these templates is presented and their applicability for achieving portability between different container and cloud environments is analysed and evaluated. A proof‐of‐concept where MiCADO serves as the deployment and execution engine for a Science Gateway in Sleep Healthcare is then described. In this proof‐of‐concept, MiCADO facilitates the deployment of a complex healthcare application, which is then moved from one cloud service provider to another with only minimal changes to the template which originally described it. This TOSCA‐based approach to templates in MiCADO encourages movement between clouds by making cloud portability more approachable.

Cloud computing has many benefits relevant to the healthcare industry. Although the adoption of cloud services for healthcare systems is increasing, employment of cloud services raises many security and privacy concerns for patients and healthcare providers. We still lack a clear set of requirements consented by the different stakeholders; here in particular IT and healthcare professionals. In this study, we examine whether user perspectives on requirements for e-health on the cloud are consistent with best practice guidelines and regulatory requirements. This work contributes to the requirements engineering phase for a secure e-health cloud framework developed in a European project (ASCLEPIOS, https://www.asclepios-project.eu/). We used qualitative analysis, based on in-depth interviews, to describe and characterize clinicians' perspectives on the requirements of cloud services for healthcare data security and privacy. We examined whether these user perspectives were in harmony with the regulatory framework of the General Data Protection Regulation (GDPR), and best practice guidelines of a relevant standard, ISO 18308:2011. Ten clinicians were identified and interviewed at six healthcare organizations in Norway, the Netherlands and Germany. While user perspectives were largely consistent with both GDPR and ISO, some concerning differences in access control were noted between large and small healthcare institutions.

Convolutional Neural Networks (CNN) are used for automatic cancer detection in pathological images. These data-driven experiments are difficult to reproduce, because the CNNs may require CUDA-enabled Nvidia GPUs for acceleration and training is often performed on a large dataset stored on a researcher's computer, inaccessible to others. We introduce the RED file format for reproducible experiment description, where executable programs are packaged and referenced as Docker container images. Data inputs and outputs are described as network resources using standard transmission and authentication protocols instead of local file paths. Following the FAIR guiding principles, the RED format is based on and compatible with the established Common Workflow Language specification. RED files are interpreted by the accompanying Curious Containers (CC) software. Arbitrarily large datasets are mounted inside containers via FUSE network filesystems like SSHFS. SSHFS is compared to NFS and a local SSD in artificial benchmarks and in the context of a CNN training scenario, where SSHFS introduces a performance decrease by a factor of 1.8. We are convinced that RED can greatly improve the reproducibility of deep learning workloads and data-driven experiments. This is in particular important in clinical scenarios where the result of an analysis may contribute to a patient's treatment.

During the panel discussion, participants observed that the workshop topics represent the wide range of issues that are currently investigated in this field.On the one hand, typical methods of Big Data environments, such as the programming language Python and the programming framework

Summary Applications for biomedical data processing often integrate external libraries and frameworks for common algorithmic tasks. It typically reduces development time and increases overall code quality. With the introduction of lightweight container‐based virtualization, the bundling of applications and their required dependencies has become feasible, and containers can be transferred and executed in distributed environments. However, the incorporation of unreviewed code poses a security threat as it might contain malicious components. In this paper, measures to minimize risks of untrusted application execution are presented. Based on the system calls issued during sample execution of the application, both the container itself and the container runtime configuration are restricted to the set of actions the application requires. It is shown that the employed security measures are suited to counteract different attacks while application runtime is not affected.

Applications for data analysis of biomedical data are complex programs and often consist of multiple components. Re-usage of existing solutions from external code repositories or program libraries is common in algorithm development. To ease reproducibility as well as transfer of algorithms and required components into distributed infrastructures Linux containers are increasingly used in those environments, that are at least partly connected to the internet. However concerns about the untrusted application remain and are of high interest when medical data is processed. Additionally, the portability of the containers needs to be ensured by using only security technologies, that do not require additional kernel modules. In this paper we describe measures and a solution to secure the execution of an example biomedical application for normalization of multidimensional biosignal recordings. This application, the required runtime environment and the security mechanisms are installed in a Docker-based container. A fine-grained restricted environment (sandbox) for the execution of the application and the prevention of unwanted behaviour is created inside the container. The sandbox is based on the filtering of system calls, as they are required to interact with the operating system to access potentially restricted resources e.g. the filesystem or network. Due to the low-level character of system calls, the creation of an adequate rule set for the sandbox is challenging. Therefore the presented solution includes a monitoring component to collect required data for defining the rules for the application sandbox. Performance evaluation of the application execution shows no significant impact of the resulting sandbox, while detailed monitoring may increase runtime up to over 420%.

Clinical Sleep Research is an inherent multidisciplinary field, as many health issues may affect a person's sleep conditions and sleep disorders may cause several health problems. Many patients with chronic sleep disorders suffer from different further medical conditions - called multimorbidity. Due to the high variety of the reasons and the courses of sleep disorders, individual cases are difficult to compare. Therefore there is a high demand for sleep researchers to collaborate with each other to reach necessary participant numbers and multidisciplinary expertise. To date, inter-institutional sleep research is poorly supported by IT systems. In particular the heterogeneity and the quality variations within the acquired bio signal data - caused by different bio signal recorders or different measurement procedures - are impeding common bio signal data processing. In this manuscript we introduce a virtual research platform supporting inter-institutional data sharing and processing. The infrastructure is based on XNAT - a free and open-source neuroimaging research platform - a loosely coupled service oriented architecture and scalable virtualization in the backend. The system is capable of local pseudonymization of bio signal data, mapping to a standardized set of parameters and automatic quality assessment. Terms and quality measures are derived from the "Manual for the Scoring of Sleep and Associated Events" of the American Academy of Sleep Medicine, the de-facto standard for diagnostic bio signal analysis in sleep medicine.

Neurosciencific research is increasingly based on image analysis methods. Large sets of imaging data are processed using complex image analysis tools. While today magnetic resonance imaging (MRI) is widely used for both functional and anatomical analysis of the human brain, new imaging modalities are beginning to prove their capabilities for neurological research. Among them, optical coherence tomography (OCT) allows for noninvasive visualization of anatomical structures on a micrometer scale. Becoming a standard diagnostic tool in ophthalmology, it is of rising interest for neurological research. Crucial to all data analysis methods is the quality of the input data. The platform presented in this paper is designed for automatic quality assessment of retinal OCTs. It extends the image management platform XNAT by services to calculate and store quality measures. It is also extensible regarding new quality measure algorithms, allowing the developer to upload Matlab code, compile it for the infrastructure's hardware architecture and test it in the system. The image processing tools to calculate the quality measures are provided as a cloud-based service employing Open Stack as underlying IT infrastructure. The prototype implementation encompassing security and performance aspects are presented.

The non-invasive method to track fibers of the human brain by analyzing diffusion weighted magnetic resonance images improves research of human brain structures and becomes therefore increasingly important. With fiber tracking, a connectivity map which depicts the degree of connectivity of the single voxels can be generated and used to improve knowledge about the human brain. Several tools exist to produce connectivity maps. One of them is part of the FMRIB Software Library (FSL) and free for non-commercial purposes. Due to long and therefore impracticable computing time on small computer cluster solutions, a GUI and the necessary software were implemented for the German MediGRID infrastructure. This was achieved by using wrapper scripts and a workflow for the Generic Workflow Execution Service (GWES). The solution is about 15 times faster than a small local cluster installation, depending on the number of employed MediGRID resources. This enables processing of connectivity maps for practical use in biomedical research. By using the D-Grid infrastructure, this solution is also suitable for small institutes without compute center capacities. For usability reasons, the GUI ConBrain was developed.

Analysis of diffusion weighted magnetic resonance images serves increasingly for non-invasive tracking of nerve fibers in the human brain, both in clinical diagnosis and basic research. Diffusion-tensor imaging (DTI) enables in-vivo research on the internal structure of the central nervous system, an estimation of the interconnection of functional areas and diagnosis of brain tumors and de-myelinating diseases. But modeling the local diffusion parameters is computationally expensive and on standard desktop computers runtimes of up to days are common. A workflow based grid implementation of the algorithm with slice-based parallelization has shown significant speedup. However, in production use, the implementation frequently delayed and even failed, discouraging the medical collaborators to take up the management of the data processing themselves. Therefore a comprehensive analysis of possible sources for errors and delays as well as their real impact in the respective infrastructure is vital to enable clinical researchers to fully exploit the benefits of the Healthgrid application. In this manuscript, we tested different implementations of the DTI analysis with respect to robustness and runtime. Based on the results, concrete application improvements as well as general suggestions for the layout and maintenance of Healthgrids are concluded.

The analysis of biosignals, such as the electroencephalogram EEG or the electrocardiogram (ECG), is essential for diagnosis in many medical areas, in particular sleep medicine and sleep research. A standard method in this field is the polysomnography, a multidimensional biosignal recording during the whole bedtime phase. Within the SIESTA project, a European multicenter study, comprehensive clinical and polysommnographic records from over 300 persons has been collected. To make the data available for researchers as reference for clinical research and development of new analysis tools, the SIESTA database is implemented into a grid infrastructure. To date, the complete data is stored into the grid and different algorithms for automated ECG analysis are implemented. The database can be queried and the matching data can be analysed on record level and collection level. The application is modelled as a workflow and integrated into the grid using a workflow manager. A graphical user interface is implemented as a grid portlet. It allows the initialization of new computation tasks as well as the monitoring and result-retrieval from already launched analyses.

Medical grid networks typically deal with extremely sensitive information and therefore require a special diligence in terms of security and reliability. This holds especially true in Medical Imaging, which is why the medical community long established DICOM (Digital Imaging and Communication in Medicine), a world-wide imaging and communication standard for secure and reliable data interchange. Most healthgrid projects today use DICOM with a combination of GridFTP and the Reliable-File-Transfer (RFT) webservice. Due to the multiple protocols and services involved, this solution is not ideally suited in terms of reliability or fault-tolerance. The proposed solution in this paper replaces the GridFTP combination by an end- to-end Grid-enhanced DICOM implementation and models the GridDICOM transfers as complex, Petri-Nets-based workflows. Based on these workflows, a respective workflow engine can autonomously and reliably control transfers including complex fault recovery and optimized routing strategies. A first prototype of the service, the components, and the respective workflows have been developed and successfully tested in MediGRID, the German grid network for life-sciences.

Grid computing, the collaboration of distributed resources across institutional borders, is an emerging technology to meet the rising demand on computing power and storage capacity in fields such as high-energy physics, climate modeling, or more recently, life sciences. A secure, reliable, and highly efficient data transport plays an integral role in such grid environments and even more so in medical grids. Unfortunately, many grid middleware distributions, such as the well-known Globus Toolkit, lack the integration of the world-wide medical image communication standard Digital Imaging and Communication in Medicine (DICOM). Currently, the DICOM protocol first needs to be converted to the file transfer protocol (FTP) that is offered by the grid middleware. This effectively reduces most of the advantages and security an integrated network of DICOM devices offers. In this paper, a solution is proposed that adapts the DICOM protocol to the Globus grid security infrastructure and utilizes routers to transparently route traffic to and from DICOM systems. Thus, all legacy DICOM devices can be seamlessly integrated into the grid without modifications. A prototype of the grid routers with the most important DICOM functionality has been developed and successfully tested in the MediGRID test bed, the German grid project for life sciences.