Refine
Year of publication
Document Type
- conference proceeding (presentation, abstract) (133) (remove)
Language
- English (133) (remove)
Is part of the Bibliography
- no (133)
Keywords
- Artificial Intelligence (4)
- Bildgebendes Verfahren (4)
- Deep Learning (4)
- Künstliche Intelligenz (4)
- Kernspintomografie (3)
- Speiseröhrenkrankheit (3)
- CFD (2)
- Diagnose (2)
- Endoscopy (2)
- Gehirn (2)
Institute
- Fakultät Informatik und Mathematik (61)
- Fakultät Maschinenbau (38)
- Regensburg Center of Biomedical Engineering - RCBE (26)
- Regensburg Medical Image Computing (ReMIC) (24)
- Fakultät Elektro- und Informationstechnik (19)
- Regensburg Center of Health Sciences and Technology - RCHST (17)
- Labor für Digitalisierung (LFD) (14)
- Fakultät Angewandte Sozial- und Gesundheitswissenschaften (11)
- Labor Biomechanik (LBM) (11)
- Labor Biofluidmechanik (9)
Begutachtungsstatus
- peer-reviewed (37)
- begutachtet (6)
Aims
Endoscopic retrograde cholangiopancreaticography (ERCP) is the gold standard in the diagnosis as well as treatment of diseases of the pancreatobiliary tract. However, it is technically complex and has a relatively high complication rate. In particular, cannulation of the papillary ostium remains challenging. The aim of this study is to examine whether a deep-learning algorithm can be used to detect the major duodenal papilla and in particular the papillary ostium reliably and could therefore be a valuable tool for inexperienced endoscopists, particularly in training situation.
Methods
We analyzed a total of 654 retrospectively collected images of 85 patients. Both the major duodenal papilla and the ostium were then segmented. Afterwards, a neural network was trained using a deep-learning algorithm. A 5-fold cross-validation was performed. Subsequently, we ran the algorithm on 5 prospectively collected videos of ERCPs.
Results
5-fold cross-validation on the 654 labeled data resulted in an F1 value of 0.8007, a sensitivity of 0.8409 and a specificity of 0.9757 for the class papilla, and an F1 value of 0.5724, a sensitivity of 0.5456 and a specificity of 0.9966 for the class ostium. Regardless of the class, the average F1 value (class papilla and class ostium) was 0.6866, the sensitivity 0.6933 and the specificity 0.9861. In 100% of cases the AI-detected localization of the papillary ostium in the prospectively collected videos corresponded to the localization of the cannulation performed by the endoscopist.
Conclusions
In the present study, the neural network was able to identify the major duodenal papilla with a high sensitivity and high specificity. In detecting the papillary ostium, the sensitivity was notably lower. However, when used on videos, the AI was able to identify the location of the subsequent cannulation with 100% accuracy. In the future, the neural network will be trained with more data. Thus, a suitable tool for ERCP could be established, especially in the training situation.
Purpose
In 1969, Germany founded a new type of tertiary education institution, the so-called “Universities of Applied Sciences” (UAS). In contrast to traditional universities, UAS are supposed to educate their students based on scientific methods but with a high degree of application orientation (Wissenschaftsrat, 2010). The goal was to generate highly employable graduates. Since then the number of UAS in Germany has risen to 216 in 2018/2019 (Statistisches Bundesamt, 2019). ~45% of all German students start their tertiary education at a UAS (Autorengruppe Bildungsberichterstattung, 2018). UAS graduates are well perceived by domestic employers, enjoy high recruiting rates and attractive salary levels.
The educational model has proven so successful that its approach and underlying philosophy has been replicated under terms such as “cooperative education” or “applied learning” in a vast number of different countries and educational systems around the globe (Altbach, Reisberg, & Rumbley, 2011).
This talk critically describes the key aspects that constitute German UAS’ approaches to industry-oriented applied learning. The findings are relevant for academics, policy makers, and industry representatives to generate ideas on how to improve study curricula, policies, or collaboration between academia and industry.
Method
The approach taken consists of a descriptive and structured account of the key aspects that constitute the study curricula of a UAS from an applied learning perspective. By building on their own experiences as teaching professors and part of a German UAS’ managerial team, the research method taken resembles an action research-oriented approach.
Key Findings
German UAS have numerous application-oriented learning elements in their curricula. These elements are performed inside the university like lab courses but many of these learning elements are carried out in cooperation with industry.
Bachelor programs span over seven semesters. Fundamentals are taught in the first two semesters and from the third semester on, more and more application-oriented lectures and labs become part of the curricula. In the fifth semester, a mandatory internship is integrated in the study program. This internship is carried out either at companies in Germany or abroad. It is the goal of the internship that the students gain first practical experience in a real work environment. The work tasks in internship should be similar to the later professional demands and should contain a project. The students themselves are responsible for finding an appropriate job and organize the administrative aspects around the internship, thus their self-organization skills are advanced. The UAS teaching staff supervises the internships from an academic perspective. This mandatory internship has numerous advantages: students experience real industry environment and they build contacts to later employers thus forming a potential pathway to future employment.
Bachelor and master courses each finish with thesis work. The majority of theses is carried out in companies under university supervision. This is a second chance for the students to accomplish a company project, to gain practical experience and to get in touch with potential employers.
In addition to undergraduate and postgraduate education, German UAS do a lot of applied research together with companies. This applied research can also be the basis for the work of Ph.D students.
Industry representatives are also involved when designing new study programs as consultants and give advice for accreditation and reaccreditation of programs.
All these activities of German universities of applied sciences lead to a high practical relevance of education programs and to a high degree of employability of UAS graduates.
IOT Backdoors in Cars
(2019)
Connecting cheap IoT devices to the safety-critical network of a car can be an extremely bad idea, but at least it allows us to hack together our own automotive gadget. This talk explains the complete procedure involved in transforming a cheap OBD GSM dongle designed for fleet management into a open source automotive hacking tool. First, the hardware reverse engineering is demonstrated, showing how each component is interconnected and working together. With this knowledge, it was possible to capture the communication of the GSM module and understand the OTA protocol used by this dongle, which can be used to extract the firmware. A quick reverse engineering of the software will show that no cryptographic authentication is used for the OTA updates, and therefore a pirate GSM BTS can be used to obtain remote code execution. After that, a new open source firmware is written for the device, which can easily be extended and controlled remotely with the LUA scripting language. Examples on how hacking this dongle remotely can affect the safety of the driver will be also given.
This talk will provide a general overview on how Scapy can be used for automotive penetration testing. All present features of Scapy for automotive penetration will be introduced and explained. Also an overview of higher level automotive protocols will be given.
As automotive penetration testing becomes more important, the lack of free tools for automotive network penetration testing led us to integrate new features in Scapy. Scapy is a well established framework for packet manipulation. The flexibility of Scapy allowed us to implement automotive interfaces (CAN) and automotive protocols (ISOTP, GMLAN, UDS, DoIP, OBD-II).
This talk explains the basics of these automotive protocols, the workflow with Scapy for automotive network penetration testing. A live demonstration with some embedded hardware will be given.