Options
Fraunhofer-Institut für Angewandte Informationstechnik FIT
Now showing
1 - 3 of 3
-
PublicationDeepKneeExplainer: Explainable knee osteoarthritis diagnosis from radiographs and magnetic resonance imaging( 2021)
;Karim, Rezaul ;Cochez, Michael ;Beyan, Oya ;Rebholz-Schuhmann, DietrichOsteoarthritis (OA) is a degenerative joint disease, which significantly affects middle-aged and elderly people. Although primarily identified via hyaline cartilage change based on medical images, technical bottlenecks like noise, artifacts, and modality impose an enormous challenge on high-precision, objective, and efficient early quantification of OA. Owing to recent advancements, approaches based on neural networks (DNNs) have shown outstanding success in this application domain. However, due to nested non-linear and complex structures, DNNs are mostly opaque and perceived as black-box methods, which raises numerous legal and ethical concerns. Moreover, these approaches do not have the ability to provide the reasoning behind diagnosis decisions in the way humans would do, which poses an additional risk in the clinical setting. In this paper, we propose a novel explainable method for knee OA diagnosis based on radiographs and magnetic resonance imaging (MRI), which we called DeepKneeExplainer. First, we comprehensively preprocess MRIs and radiographs through the deep-stacked transformation technique against possible noises and artifacts that could contain unseen images for domain generalization. Then, we extract the region of interests (ROIs) by employing U-Net architecture with ResNet backbone. To classify the cohorts, we train DenseNet and VGG architectures on the extracted ROIs. Finally, we highlight class-discriminating regions using gradient-guided class activation maps (Grad-CAM++) and layer-wise relevance propagation (LRP), followed by providing human-interpretable explanations of the predictions. Comprehensive experiments based on the multicenter osteoarthritis study (MOST) cohorts, our approach yields up to 91% classification accuracy, outperforming comparable state-of-the-art approaches. We hope that our results will encourage medical researchers and developers to adopt explainable methods and DNN-based analytic pipelines towards an increasing acceptance and adoption of AI-assisted applications in the clinical practice for improved knee OA diagnoses. -
PublicationFinding and analysing energy research funding data: The EnArgus system( 2021)
;Heiwolt, Karoline ;Lange, ChristophThis paper presents the concept, a system-overview, and the evaluation of EnArgus, the central information system for energy research funding in Germany. Initiated by the German Federal Ministry for Economic Affairs and Energy (BMWi), EnArgus establishes a one-stop information system about all recent and ongoing energy research funding projects in Germany. Participants ranging from laypersons to experts were surveyed in three workshops to evaluate both the public and expert interfaces of the EnArgus system in comparison to peer systems. The results showed that the EnArgus system was predominantly evaluated positively by the various participants. It contributes to making the energy sector more transparent and offers clear advantages for professional use compared to similar systems. The system's semantic processing enables more precise hits and better coverage by including semantically related terms in search results; its intelligence makes it fail-safe, rendering it suitable for areas where poor results can have dire consequences. Reporting on an actual real-world system, the paper also provides a roadmap-view of how electronic filing of administrative project data can be semantically enhanced and opened-up to provide the basis for new ways into the data that are key for future breakthrough AI interfaces. -
PublicationEnergiekompendium. Ein Nachschlagewerk für Grundbegriffe, Konzepte und TechnologienDas Gelingen der Energiewende erfordert einen weitreichenden Wandel des derzeitigen Energiesystems. Dazu müssen alle Bereiche des Systems enger miteinander verzahnt und in ihrer Gesamtheit umgestaltet werden. Eine wichtige Säule dessen ist die Energieforschung. Dieses Nachschlagewerk richtet sich an alle, die angesichts der vielfältigen Themen einen einfachen und grundlegenden Einstieg in die Begrifflichkeiten der Energieforschung suchen. Die Beiträge sollen dem Leser durch eine Fokussierung auf das Wesentliche helfen, sich schnell mit vielfältigen Themen der Energiewende vertraut zu machen und sich daran zu orientieren. Dazu decken über 3800 Stichwörter bzw. rund 2200 Beiträge sowohl die Grundbegriffe aus Forschung und Technik, damit in Verbindung stehende Konzepte als auch zahlreiche energietechnologische Begriffe ab, die sich über die gesamte Energiewandlungskette hinweg erstrecken.