@inproceedings{JannidisRegerWeimeretal.2015, author = {Jannidis, Fotis and Reger, Isabella and Weimer, Lukas and Krug, Markus and Puppe, Frank}, title = {Automatische Erkennung von Figuren in deutschsprachigen Romanen}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-143332}, pages = {7}, year = {2015}, abstract = {Eine wichtige Grundlage f{\"u}r die quantitative Analyse von Erz{\"a}hltexten, etwa eine Netzwerkanalyse der Figurenkonstellation, ist die automatische Erkennung von Referenzen auf Figuren in Erz{\"a}hltexten, ein Sonderfall des generischen NLP-Problems der Named Entity Recognition. Bestehende, auf Zeitungstexten trainierte Modelle sind f{\"u}r literarische Texte nur eingeschr{\"a}nkt brauchbar, da die Einbeziehung von Appellativen in die Named Entity-Definition und deren h{\"a}ufige Verwendung in Romantexten zu einem schlechten Ergebnis f{\"u}hrt. Dieses Paper stellt eine anhand eines manuell annotierten Korpus auf deutschsprachige Romane des 19. Jahrhunderts angepasste NER-Komponente vor.}, subject = {Digital Humanities}, language = {de} } @article{GehrkeBalbachRauchetal.2019, author = {Gehrke, Alexander and Balbach, Nico and Rauch, Yong-Mi and Degkwitz, Andreas and Puppe, Frank}, title = {Erkennung von handschriftlichen Unterstreichungen in Alten Drucken}, series = {Bibliothek Forschung und Praxis}, volume = {43}, journal = {Bibliothek Forschung und Praxis}, number = {3}, issn = {1865-7648}, doi = {10.1515/bfp-2019-2083}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-193377}, pages = {447 -- 452}, year = {2019}, abstract = {Die Erkennung handschriftlicher Artefakte wie Unterstreichungen in Buchdrucken erm{\"o}glicht R{\"u}ckschl{\"u}sse auf das Rezeptionsverhalten und die Provenienzgeschichte und wird auch f{\"u}r eine OCR ben{\"o}tigt. Dabei soll zwischen handschriftlichen Unterstreichungen und waagerechten Linien im Druck (z. B. Trennlinien usw.) unterschieden werden, da letztere nicht ausgezeichnet werden sollen. Im Beitrag wird ein Ansatz basierend auf einem auf Unterstreichungen trainierten Neuronalen Netz gem{\"a}ß der U-Net Architektur vorgestellt, dessen Ergebnisse in einem zweiten Schritt mit heuristischen Regeln nachbearbeitet werden. Die Evaluationen zeigen, dass Unterstreichungen sehr gut erkannt werden, wenn bei der Binarisierung der Scans nicht zu viele Pixel der Unterstreichung wegen geringem Kontrast verloren gehen. Zuk{\"u}nftig sollen die Worte oberhalb der Unterstreichung mit OCR transkribiert werden und auch andere Artefakte wie handschriftliche Notizen in alten Drucken erkannt werden.}, language = {de} } @article{ToepferCorovicFetteetal.2015, author = {Toepfer, Martin and Corovic, Hamo and Fette, Georg and Kl{\"u}gl, Peter and St{\"o}rk, Stefan and Puppe, Frank}, title = {Fine-grained information extraction from German transthoracic echocardiography reports}, series = {BMC Medical Informatics and Decision Making}, volume = {15}, journal = {BMC Medical Informatics and Decision Making}, number = {91}, doi = {doi:10.1186/s12911-015-0215-x}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-125509}, year = {2015}, abstract = {Background Information extraction techniques that get structured representations out of unstructured data make a large amount of clinically relevant information about patients accessible for semantic applications. These methods typically rely on standardized terminologies that guide this process. Many languages and clinical domains, however, lack appropriate resources and tools, as well as evaluations of their applications, especially if detailed conceptualizations of the domain are required. For instance, German transthoracic echocardiography reports have not been targeted sufficiently before, despite of their importance for clinical trials. This work therefore aimed at development and evaluation of an information extraction component with a fine-grained terminology that enables to recognize almost all relevant information stated in German transthoracic echocardiography reports at the University Hospital of W{\"u}rzburg. Methods A domain expert validated and iteratively refined an automatically inferred base terminology. The terminology was used by an ontology-driven information extraction system that outputs attribute value pairs. The final component has been mapped to the central elements of a standardized terminology, and it has been evaluated according to documents with different layouts. Results The final system achieved state-of-the-art precision (micro average.996) and recall (micro average.961) on 100 test documents that represent more than 90 \% of all reports. In particular, principal aspects as defined in a standardized external terminology were recognized with f 1=.989 (micro average) and f 1=.963 (macro average). As a result of keyword matching and restraint concept extraction, the system obtained high precision also on unstructured or exceptionally short documents, and documents with uncommon layout. Conclusions The developed terminology and the proposed information extraction system allow to extract fine-grained information from German semi-structured transthoracic echocardiography reports with very high precision and high recall on the majority of documents at the University Hospital of W{\"u}rzburg. Extracted results populate a clinical data warehouse which supports clinical research.}, language = {en} } @article{HoernleinMandelIflandetal.2011, author = {H{\"o}rnlein, Alexander and Mandel, Alexander and Ifland, Marianus and L{\"u}neberg, Edeltraud and Deckert, J{\"u}rgen and Puppe, Frank}, title = {Akzeptanz medizinischer Trainingsf{\"a}lle als Erg{\"a}nzung zu Vorlesungen}, series = {GMS Zeitschrift f{\"u}r Medizinische Ausbildung}, volume = {28}, journal = {GMS Zeitschrift f{\"u}r Medizinische Ausbildung}, number = {3}, doi = {10.3205/zma000754}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-133569}, pages = {Doc42}, year = {2011}, abstract = {Introduction: Medical training cases (virtual patients) are in widespread use for student education. Most publications report about development and experiences in one course with training cases. In this paper we compare the acceptance of different training case courses with different usages deployed as supplement to lectures of the medical faculty of Wuerzburg university during a period of three semesters. Methods: The training cases were developed with the authoring tool CaseTrain and are available for students via the Moodle-based eLearning platform WueCampus at Wuerzburg university. Various data about usage and acceptance is automatically collected. Results: From WS (winter semester) 08/09 till WS 09/10 19 courses with about 200 cases were available. In each semester, about 550 different medical students from W{\"u}rzburg and 50 students from other universities processed about 12000 training cases and filled in about 2000 evaluation forms. In different courses, the usage varied between less than 50 and more than 5000 processed cases. Discussion: Although students demand training cases as supplement to all lectures, the data show that the usage does not primarily depend on the quality of the available training cases. Instead, the training cases of nearly all case collections were processed extremely often shortly before the examination. It shows that the degree of usage depends primarily on the perceived relevance of the training cases for the examination."}, language = {de} } @article{MandelHoernleinIflandetal.2011, author = {Mandel, Alexander and H{\"o}rnlein, Alexander and Ifland, Marianus and L{\"u}neburg, Edeltraud and Deckert, J{\"u}rgen and Puppe, Frank}, title = {Aufwandsanalyse f{\"u}r computerunterst{\"u}tzte Multiple-Choice Papierklausuren}, series = {GMS Journal for Medical Education}, volume = {28}, journal = {GMS Journal for Medical Education}, number = {4}, doi = {10.3205/zma000767}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-134386}, pages = {1-15, Doc55}, year = {2011}, abstract = {Introduction: Multiple-choice-examinations are still fundamental for assessment in medical degree programs. In addition to content related research, the optimization of the technical procedure is an important question. Medical examiners face three options: paper-based examinations with or without computer support or completely electronic examinations. Critical aspects are the effort for formatting, the logistic effort during the actual examination, quality, promptness and effort of the correction, the time for making the documents available for inspection by the students, and the statistical analysis of the examination results. Methods: Since three semesters a computer program for input and formatting of MC-questions in medical and other paper-based examinations is used and continuously improved at Wuerzburg University. In the winter semester (WS) 2009/10 eleven, in the summer semester (SS) 2010 twelve and in WS 2010/11 thirteen medical examinations were accomplished with the program and automatically evaluated. For the last two semesters the remaining manual workload was recorded. Results: The cost of the formatting and the subsequent analysis including adjustments of the analysis of an average examination with about 140 participants and about 35 questions was 5-7 hours for exams without complications in the winter semester 2009/2010, about 2 hours in SS 2010 and about 1.5 hours in the winter semester 2010/11. Including exams with complications, the average time was about 3 hours per exam in SS 2010 and 2.67 hours for the WS 10/11. Discussion: For conventional multiple-choice exams the computer-based formatting and evaluation of paper-based exams offers a significant time reduction for lecturers in comparison with the manual correction of paper-based exams and compared to purely electronically conducted exams it needs a much simpler technological infrastructure and fewer staff during the exam."}, language = {de} } @article{KrenzerMakowskiHekaloetal.2022, author = {Krenzer, Adrian and Makowski, Kevin and Hekalo, Amar and Fitting, Daniel and Troya, Joel and Zoller, Wolfram G. and Hann, Alexander and Puppe, Frank}, title = {Fast machine learning annotation in the medical domain: a semi-automated video annotation tool for gastroenterologists}, series = {BioMedical Engineering OnLine}, volume = {21}, journal = {BioMedical Engineering OnLine}, number = {1}, doi = {10.1186/s12938-022-01001-x}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-300231}, year = {2022}, abstract = {Background Machine learning, especially deep learning, is becoming more and more relevant in research and development in the medical domain. For all the supervised deep learning applications, data is the most critical factor in securing successful implementation and sustaining the progress of the machine learning model. Especially gastroenterological data, which often involves endoscopic videos, are cumbersome to annotate. Domain experts are needed to interpret and annotate the videos. To support those domain experts, we generated a framework. With this framework, instead of annotating every frame in the video sequence, experts are just performing key annotations at the beginning and the end of sequences with pathologies, e.g., visible polyps. Subsequently, non-expert annotators supported by machine learning add the missing annotations for the frames in-between. Methods In our framework, an expert reviews the video and annotates a few video frames to verify the object's annotations for the non-expert. In a second step, a non-expert has visual confirmation of the given object and can annotate all following and preceding frames with AI assistance. After the expert has finished, relevant frames will be selected and passed on to an AI model. This information allows the AI model to detect and mark the desired object on all following and preceding frames with an annotation. Therefore, the non-expert can adjust and modify the AI predictions and export the results, which can then be used to train the AI model. Results Using this framework, we were able to reduce workload of domain experts on average by a factor of 20 on our data. This is primarily due to the structure of the framework, which is designed to minimize the workload of the domain expert. Pairing this framework with a state-of-the-art semi-automated AI model enhances the annotation speed further. Through a prospective study with 10 participants, we show that semi-automated annotation using our tool doubles the annotation speed of non-expert annotators compared to a well-known state-of-the-art annotation tool. Conclusion In summary, we introduce a framework for fast expert annotation for gastroenterologists, which reduces the workload of the domain expert considerably while maintaining a very high annotation quality. The framework incorporates a semi-automated annotation system utilizing trained object detection models. The software and framework are open-source.}, language = {en} } @article{Puppe2022, author = {Puppe, Frank}, title = {Gesellschaftliche Perspektiven einer fachspezifischen KI f{\"u}r automatisierte Entscheidungen}, series = {Informatik Spektrum}, volume = {45}, journal = {Informatik Spektrum}, number = {2}, issn = {0170-6012}, doi = {10.1007/s00287-022-01443-6}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-324197}, pages = {88-95}, year = {2022}, abstract = {Die k{\"u}nstliche Intelligenz (KI) entwickelt sich rasant und hat bereits eindrucksvolle Erfolge zu verzeichnen, darunter {\"u}bermenschliche Kompetenz in den meisten Spielen und vielen Quizshows, intelligente Suchmaschinen, individualisierte Werbung, Spracherkennung, -ausgabe und -{\"u}bersetzung auf sehr hohem Niveau und hervorragende Leistungen bei der Bildverarbeitung, u. a. in der Medizin, der optischen Zeichenerkennung, beim autonomen Fahren, aber auch beim Erkennen von Menschen auf Bildern und Videos oder bei Deep Fakes f{\"u}r Fotos und Videos. Es ist zu erwarten, dass die KI auch in der Entscheidungsfindung Menschen {\"u}bertreffen wird; ein alter Traum der Expertensysteme, der durch Lernverfahren, Big Data und Zugang zu dem gesammelten Wissen im Web in greifbare N{\"a}he r{\"u}ckt. Gegenstand dieses Beitrags sind aber weniger die technischen Entwicklungen, sondern m{\"o}gliche gesellschaftliche Auswirkungen einer spezialisierten, kompetenten KI f{\"u}r verschiedene Bereiche der autonomen, d. h. nicht nur unterst{\"u}tzenden Entscheidungsfindung: als Fußballschiedsrichter, in der Medizin, f{\"u}r richterliche Entscheidungen und sehr spekulativ auch im politischen Bereich. Dabei werden Vor- und Nachteile dieser Szenarien aus gesellschaftlicher Sicht diskutiert.}, subject = {K{\"u}nstliche Intelligenz}, language = {de} } @article{DakroubVermaFuehringAgorastouetal.2022, author = {Dakroub, Mohamad and Verma-Fuehring, Raoul and Agorastou, Vaia and Sch{\"o}n, Julian and Hillenkamp, Jost and Puppe, Frank and Loewen, Nils A.}, title = {Inter-eye correlation analysis of 24-h IOPs and glaucoma progression}, series = {Graefe's Archive for Clinical and Experimental Ophthalmology}, volume = {260}, journal = {Graefe's Archive for Clinical and Experimental Ophthalmology}, number = {10}, doi = {10.1007/s00417-022-05651-4}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-323831}, pages = {3349-3356}, year = {2022}, abstract = {Purpose To determine whether 24-h IOP monitoring can be a predictor for glaucoma progression and to analyze the inter-eye relationship of IOP, perfusion, and progression parameters. Methods We extracted data from manually drawn IOP curves with HIOP-Reader, a software suite we developed. The relationship between measured IOPs and mean ocular perfusion pressures (MOPP) to retinal nerve fiber layer (RNFL) thickness was analyzed. We determined the ROC curves for peak IOP (T\(_{max}\)), average IOP(T\(_{avg}\)), IOP variation (IOP\(_{var}\)), and historical IOP cut-off levels to detect glaucoma progression (rate of RNFL loss). Bivariate analysis was also conducted to check for various inter-eye relationships. Results Two hundred seventeen eyes were included. The average IOP was 14.8 ± 3.5 mmHg, with a 24-h variation of 5.2 ± 2.9 mmHg. A total of 52\% of eyes with RNFL progression data showed disease progression. There was no significant difference in T\(_{max}\), T\(_{avg}\), and IOP\(_{var}\) between progressors and non-progressors (all p > 0.05). Except for T\(_{avg}\) and the temporal RNFL, there was no correlation between disease progression in any quadrant and T\(_{max}\), T\(_{avg}\), and IOP\(_{var}\). Twenty-four-hour and outpatient IOP variables had poor sensitivities and specificities in detecting disease progression. The correlation of inter-eye parameters was moderate; correlation with disease progression was weak. Conclusion In line with our previous study, IOP data obtained during a single visit (outpatient or inpatient monitoring) make for a poor diagnostic tool, no matter the method deployed. Glaucoma progression and perfusion pressure in left and right eyes correlated weakly to moderately with each other. Key messages What is known: ● Our prior study showed that manually obtained 24-hour inpatient IOP measurements in right eyes are poor predictors for glaucoma progression. The inter-eye relationship of 24-hour IOP parameters and disease progression on optical coherence tomography (OCT) has not been examined. What we found: ● 24-hour IOP profiles of left eyes from the same study were a poor diagnostic tool to detect worsening glaucoma. ● Significant inter-eye correlations of various strengths were found for all tested parameters}, language = {en} } @article{KempfKrugPuppe2023, author = {Kempf, Sebastian and Krug, Markus and Puppe, Frank}, title = {KIETA: Key-insight extraction from scientific tables}, series = {Applied Intelligence}, volume = {53}, journal = {Applied Intelligence}, number = {8}, issn = {0924-669X}, doi = {10.1007/s10489-022-03957-8}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-324180}, pages = {9513-9530}, year = {2023}, abstract = {An important but very time consuming part of the research process is literature review. An already large and nevertheless growing ground set of publications as well as a steadily increasing publication rate continue to worsen the situation. Consequently, automating this task as far as possible is desirable. Experimental results of systems are key-insights of high importance during literature review and usually represented in form of tables. Our pipeline KIETA exploits these tables to contribute to the endeavor of automation by extracting them and their contained knowledge from scientific publications. The pipeline is split into multiple steps to guarantee modularity as well as analyzability, and agnosticim regarding the specific scientific domain up until the knowledge extraction step, which is based upon an ontology. Additionally, a dataset of corresponding articles has been manually annotated with information regarding table and knowledge extraction. Experiments show promising results that signal the possibility of an automated system, while also indicating limits of extracting knowledge from tables without any context.}, language = {en} } @article{LuxBanckSassmannshausenetal.2022, author = {Lux, Thomas J. and Banck, Michael and Saßmannshausen, Zita and Troya, Joel and Krenzer, Adrian and Fitting, Daniel and Sudarevic, Boban and Zoller, Wolfram G. and Puppe, Frank and Meining, Alexander and Hann, Alexander}, title = {Pilot study of a new freely available computer-aided polyp detection system in clinical practice}, series = {International Journal of Colorectal Disease}, volume = {37}, journal = {International Journal of Colorectal Disease}, number = {6}, doi = {10.1007/s00384-022-04178-8}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-324459}, pages = {1349-1354}, year = {2022}, abstract = {Purpose Computer-aided polyp detection (CADe) systems for colonoscopy are already presented to increase adenoma detection rate (ADR) in randomized clinical trials. Those commercially available closed systems often do not allow for data collection and algorithm optimization, for example regarding the usage of different endoscopy processors. Here, we present the first clinical experiences of a, for research purposes publicly available, CADe system. Methods We developed an end-to-end data acquisition and polyp detection system named EndoMind. Examiners of four centers utilizing four different endoscopy processors used EndoMind during their clinical routine. Detected polyps, ADR, time to first detection of a polyp (TFD), and system usability were evaluated (NCT05006092). Results During 41 colonoscopies, EndoMind detected 29 of 29 adenomas in 66 of 66 polyps resulting in an ADR of 41.5\%. Median TFD was 130 ms (95\%-CI, 80-200 ms) while maintaining a median false positive rate of 2.2\% (95\%-CI, 1.7-2.8\%). The four participating centers rated the system using the System Usability Scale with a median of 96.3 (95\%-CI, 70-100). Conclusion EndoMind's ability to acquire data, detect polyps in real-time, and high usability score indicate substantial practical value for research and clinical practice. Still, clinical benefit, measured by ADR, has to be determined in a prospective randomized controlled trial.}, language = {en} }