DWDaniel Weiss

Daniel Weiss

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Daniel Weiss has a total of 146 co-authors. Here are the top co-authors by number of shared publications:

  • Bernd Turowski

    Department of Diagnostic and Interventional Radiology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany

    9 shared publications
  • Kai Jannusch

    Department of Diagnostic and Interventional Radiology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany

    7 shared publications
  • Peter Minko

    Department of Diagnostic and Interventional Radiology, University Dusseldorf, Medical Faculty, Dusseldorf, Germany

    7 shared publications
  • Farid Ziayee

    Department of Diagnostic and Interventional Radiology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany

    6 shared publications
  • Gerald Antoch

    Department of Diagnostic and Interventional Radiology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany

    6 shared publications
  • Christian Rubbert

    Department of Diagnostic and Interventional Radiology, Medical Faculty, University Düsseldorf, Düsseldorf, Germany

    5 shared publications

Latest Publications

  • Open Access

    Data-efficient Neuroradiology MRI Imaging Protocol Prediction Using Open-weights Large Language Models

    by Marius Vach, Christian Boschenriedter, Daniel Weiss, Vivien Lorena Ivan, Karl Ludger Radke, Julian Caspers, Christian Rubbert

    Purpose: To evaluate the feasibility of a locally deployable large language model (LLM) system for automated MRI protocol selection addressing data privacy, annotation burden, and scalability limitations. Methods: This retrospective study included 598 German-language MRI order entries from three neuroradiology domains (brain, head/neck, spine) between June 2018 and January 2023. A radiologist labeled entries for 27 protocol classes based on institutional standard operating procedures (SOP). An SOP-grounded AI system using MedGemma 27B was developed to predict the MRI protocol from the order entry. The system was optimized using Stochastic Introspective Mini-Batch Ascent (SIMBA), a self-reflective prompt optimization algorithm, and compared with a hierarchical system that first classified the body region and then the MRI protocol. Data efficiency was evaluated using training subsets of 10–119 examples across 3 optimization runs per subset size. Results: The flat zero-shot model achieved 73.07% accuracy in the three-domain setting on the held-out dataset ( n = 479). In the hierarchical model, prompt optimization yielded 73.90% ± 3.24% with 30 labeled examples and a maximum of 74.46% ± 3.82% but did not outperform the flat approach. Prompt optimization benefited the hierarchical system, whereas the flat model already performed strongly without labeled examples for optimization. Performance on brain and head/neck cases remained broadly stable after expansion from 22 to 27 protocol classes, i.e., including spine. Conclusion: A locally deployable SOP-grounded open-weight LLM can support MRI protocol selection while preserving data privacy and needing minimal labeled data. In this dataset, hierarchical routing and prompt optimization did not improve overall performance over the flat baseline, although they altered optimization behavior and error profile. These findings support prospective evaluation in human-in-the-loop clinical workflows.

  • Open Access

    Diagnostic Potential of Apparent Diffusion Coefficient-Based Lymph Node Classification in Breast Cancer Patients Undergoing [18F]FDG-PET/MRI

    by Helena A. Peters, Marie Scheuer, Daniel Weiss, Matthias Boschheidgen, Vivien Lorena Ivan, Frederic Dietzel, Svjetlana Mohrmann, Eugen Ruckhäberle, Ken Herrmann, Harald H. Quick, et al.

    Background/Objectives: To evaluate the diagnostic potential of apparent diffusion coefficient (ADC) values for classifying lymph nodes as benign or malignant in breast cancer patients undergoing [18F]FDG-PET/MRI staging. Methods: Mean ADC values and short-axis diameters (±standard deviation) of 199 thoracic lymph nodes in 113 newly diagnosed breast cancer patients were retrospectively analyzed. All patients underwent [18F]FDG-PET/MRI staging, between July 2017 and June 2021. A node-by-node comparison was performed with respect to pathological node status. Nodal FDG uptake in whole-body [18F]FDG-PET/MRI served as reference standard for nodal malignancy. Group comparison using Mann–Whitney U test, receiver operating characteristic curve (ROC) analysis and diagnostic performance were calculated. p values below 0.05 were defined as statistically significant. Confidence intervals (CI; 95%) were calculated. Results: Ninety-three lymph nodes were FDG-negative while 106 lymph nodes were FDG-positive. FDG-negative lymph nodes had significantly lower short-axis diameters ((5.1 ± 1.5 mm versus 12.3 ± 5.3 mm); p < 0.01; U: 405.50; Z: −11.24). ADC values were significantly lower in FDG-positive lymph nodes (0.72 ± 0.14 × 10−3 mm2/s) than in FDG-negative lymph nodes ((1.18 ± 0.18 × 10−3 mm2/s); p < 0.01; U: 173.00; Z: −11.80). ROC analysis and Youden’s index revealed an ADC cut-off of 0.95 × 10−3 mm2/s (AUC: 0.98; p < 0.01; 95% CI: 0.96–1.01). According to the calculated cut-off, sensitivity, specificity, and accuracy of ADC values for differentiating FDG-negative from FDG-positive lymph nodes were 98%, 97% and 97%, respectively. Conclusions: ADC values derived from MRI were significantly associated with FDG uptake in this retrospective cohort and may serve as a complementary imaging biomarker for lymph node characterization.


  • Open Access

    RECOAT study - Retrospective evaluation of surface-modified and coated flow diverters for the treatment of unruptured intracranial aneurysms

    by Lukas Goertz, Sophia Hohenstatt, Dominik F. Vollherbst, Charlotte S. Weyland, Omid Nikoubashman, Hanna Styczen, Christian Gronemann, Daniel A. Weiss, Marius Kaschner, Johannes Rueckel, et al.

    Purpose Surface modifications may improve procedural safety of flow diverter (FD) treatment, but larger studies are rare. This multicenter study aims to evaluate the safety and efficacy of unruptured aneurysm treatment with surface-modified and coated FDs. Methods Patients treated with the following FDs at 8 neurovascular centers were retrospectively reviewed: Pipeline Vantage Embolization Device, Pipeline Flex with Shield Technology, FRED X, p48/64 HPC, Derivo Embolization Device (DED), and DED 2heal. Aneurysm characteristics, procedural details, and angiographic results were evaluated in detail. Results A total of 511 patients with 545 aneurysms were included (mean aneurysm size: 8.2 ± 5.5 mm, 12% posterior circulation, 14% non-saccular morphology). Multiple FDs were used in 26/515 (5%) procedures and adjunctive coiling in 76 (15%). Thromboembolic events occurred in 40 (7.8%) cases, including 9 (1.7%) major ischemic events, 20 (3.9%) minor ischemic events and 11 (2.1%) asymptomatic/technical events. Non-saccular aneurysm morphology (p = 0.014) and use of multiple FDs (p = 0.025) favored thromboembolic events. Hemorrhagic events occurred in 5 (1.0%) cases, of which 3 were symptomatic. The combined morbidity and mortality rate was 11/515 (2.1%). Complete and adequate occlusion rates were 66% (196/297) and 80% (238/297) at 6 months, respectively, and 70% (64/92) and 85% (78/92) at 12 months, respectively. Conclusion The present study demonstrates high safety and efficacy for coated or surface-modified FDs. Whether the surface modifications confer a clinical benefit needs to be addressed in comparative studies.


  • Open Access

    Time to standardize: The German Spine Society national clinical guideline on epidural injections

    by Stephan Klessinger, Rezvan Ahmadi, Christopher Büttner, Hans-Raimund Casser, Christian Ewelt, Robert Fessl, Holger Koepp, Heike Norda, Hela-F. Petereit, Philipp Pieroh, et al.


  • Open Access

    Evaluating Retrieval Augmented Generation-enhanced Large Language Models for Question Answering On German Neurovascular Guidelines

    by Marius Vach, Michael Gliem, Daniel Weiss, Vivien Lorena Ivan, Frederik Hauke, Christian Boschenriedter, Christian Rubbert, Julian Caspers

    Purpose: To investigate the feasibility of Retrieval-augmented Generation (RAG)-enhanced Large Language Models (LLMs) in answering questions about two German neurovascular guidelines. Methods: Four LLMs (GPT-4o-mini, Llama 3.1 405B Instruct Turbo, Mixtral 8 × 22B Instruct, and Claude 3.5 Sonnet) with RAG as well as GPT-4o-mini without RAG were evaluated for generating answers about two German neurovascular guidelines (“S3 Guideline for Diagnosis, Treatment, and Follow-up of Extracranial Carotid Stenosis” and “S2e Guideline for Acute Therapy of Ischemic Stroke”). The answers were classified as “correct”, “inaccurate”, or “incorrect” by two neurovascular experts in consensus. Additionally, retrieval performance of five retrieval strategies was analyzed on a synthetic dataset of 384 questions. Results: Claude Sonnet 3.5 achieved the highest answer correctness (70.6% correct, 10.6% wrong), followed by Llama 3.1 (64.7%, 15.3% wrong), GPT-4o-mini with RAG (57.6%, 15.3% wrong), and Mixtral (56.6%, 17.6% wrong). GPT-4o-mini without RAG performed significantly worse (20.0%, 32.9% wrong). Retrieval errors were the primary cause of incorrect answers (80%). For retrieval, BM25 achieved the highest accuracy (82.0%), outperforming vector-based methods like “BAAI/bge-m3” (78.4%). Conclusion: RAG significantly improves LLM accuracy for medical guideline question answering compared to the inherent knowledge of pretrained LLMs alone while still showing significant error rates. Improved accuracy and confidence metrics are needed for safer implementation in clinical routine. Additionally, our results demonstrate the strong performance of general LLMs in medical question answering for non-English languages, such as German, even without specific training.


  • Open Access

    Predictive Power of Baseline [18F]FDG PET/CT for Adverse Events in DLBCL Patients Undergoing CAR-T Cell Therapy

    by Helena A. Peters, Emil Novruzov, Ben-Niklas Bärmann, Daniel Weiss, Matthias Boschheidgen, Vivien Lorena Ivan, Nora Liebers, Johannes Fischer, Eduards Mamlins, Aleksandar Radujkovic, et al.

    Objectives: Evaluation of the predictive potential of pre-CAR-T [18F]FDG PET/CT in Diffuse Large B-Cell Lymphoma (DLBCL) patients concerning Cytokine Release Syndrome (CRS) and Immune Effector Cell-associated Neurotoxicity Syndrome (ICANS). Methods: Eighteen DLBCL patients (mean age: 60 ± 12 years) who underwent pre-therapeutic [18F]FDG-PET/CT and CAR-T cell therapy were retrospectively included. Median follow-up time was ten months (IQR6-16) after CAR-T cell infusion. Age, sex, serum lactate dehydrogenase (LDH), interleukin-6 (IL-6), C-reactive protein (CRP), and modified Endothelial Activation and Stress Index (mEASIX) were obtained. Potential occurrence of CRS/ICANS and the SUVmax were evaluated. Pearson and Spearman correlations, group comparisons (Mann–Whitney U-test) and the odds ratio (OR) were calculated. P values below 0.05 were defined as statistically significant and 95%-confidence intervals (CI) were calculated. Results: Pre-therapeutic SUVmax correlated positively with LDH (r = 0.5; p = 0.02), with the grade of CRS (r = 0.5; p = 0.03) and with the grade of ICANS (r = 0.6; p = 0.01). Appearance of ICANS was significantly correlated with pre-therapeutic SUVmax (p = 0.03; U = 7.0; Z = −2.2). Using ROC analysis and Youden’s index, an SUVmax threshold of 17 (AUC: 0.865; p < 0.01) was defined. Patients exceeding a pre-therapeutic SUVmax of 17 had a significantly higher risk of CRS grade > 1 (OR = 22; CI 2, 314; p = 0.03) and ICANS grade > 1 (OR = 18; CI 1, 271; p = 0.04). Conclusions: Pre-therapeutic SUVmax may be a useful marker for identifying DLBCL patients at risk for CRS and ICANS.

  • Open Access

    Multicenter experience with a novel flow diverter in visceral aneurysms

    by Daniel Weiss, Kai Jannusch, Reinhard Meier, Elmar Spüntrup, De-Hua Chang, Farid Ziayee, Lena Marie Wilms, Bernd Turowski, Gerald Antoch, Peter Minko

    Therapeutic options for the treatment of visceral artery aneurysms (VAA) and pseudoaneurysms (VAPA) include either a surgical or endovascular approach, depending on specific characteristics of both the aneurysm and the individual patient. The therapeutic use of flow diverters (FD) has already yielded excellent results. The purpose of this study is to evaluate initial experiences regarding the use of a novel abdominal FD. In this observational multicenter study, the data of nine consecutive patients with visceral or abdominal aneurysms from four interventional centers, treated with the Derivo peripher FD, between 2022 and 2024 were evaluated regarding the technical feasibility and procedural performance of the device, safety, and treatment outcome. The average size of the treated aneurysms was 18±11 mm, and the affected vessels were the splenic (55.6%), the hepatic (22.2%), the gastroduodenal (11.1%), and the renal artery (11.1%). Complete aneurysm occlusion was observed in every patient after three months, and mid-term follow-up after 12 months was available for eight patients. All FDs remained patent during the follow-up period. No major complications or adverse events arose. The overall mortality rate was 0%. Based on our initial multicenter experience, endovascular treatment of abdominal and visceral aneurysms with an FD was found to be effective and safe. The remarkable results in terms of aneurysm occlusion and low complication rates are promising.


  • Open Access

    Endovascular Fenestration in Aortic Type‐A Dissection With Hepatic Malperfusion Syndrome: A Case Report

    by Daniel Weiss, Kai Jannusch, Lena Wilms, Hannan Dalyanoglu, Tim Ullrich, Gerald Antoch, Peter Minko, Farid Ziayee

    Acute type‐A aortic dissection with malperfusion syndromes represents challenging cases and should always be treated on an interdisciplinary basis, whereby interventional radiologic therapy can be a successful procedure, especially in patients who are not fit for surgery.

  • Endovascular Treatment of Visceral Artery Aneurysms: Data Analysis of the DeGIR/DGNR Quality Management System

    by Daniel Weiss, Kai Jannusch, Lena Marie Wilms, Helena Anne Peters, Jonathan Nadjiri, Gerald Antoch, Farid Ziayee, Peter Minko

    Visceral artery aneurysms (VAA) are rare, and data on their occurrence and treatment are often outdated or heterogeneous. The objective of this study is to provide a comprehensive overview of features and interventional treatment options for VAA. We analyzed demographic, procedural, and clinical data on VAA from 2018 to 2022 based on data derived from the German Society for Interventional Radiology and Minimally Invasive Therapy (DeGIR) registry regarding demographic details, aneurysm characteristics, treatment processes, and outcomes such as technical success, complication rates, and radiation exposure. Group comparisons between aneurysm locations and types were calculated as dependent variables as well as occurrence of incomplete embolization, complication rates, technical success, duration of intervention, and dose-area-product (DAP). In total 647 patients were enrolled. The mean age was 66 years, with 68.5% male. The most affected arteries were splenic (27.5%) and renal (21.3%). True aneurysms were most common (56.7%). Technical success was 91.2%, while primary incomplete embolization occurred in 47.3%. Complications were observed in 4.8%, with a 0.2% mortality. Mean intervention duration was 67 min, mean fluoroscopic time was 36 min and mean DAP was 9560 cGycm2. There were no significant differences in primary incomplete embolization, complication rate, correct placement of embolization material, duration, and DAP based on aneurysm location. Aneurysm types showed a significant difference in incomplete embolization, with dissections having the highest rates and longest durations of intervention and highest DAP. The analysis of DeGIR quality management data reveals consistently high technical success and low complication rates in endovascular treatment of VAA. Dissecting aneurysms exhibit the highest rates in incomplete embolization, requiring the longest interventions and increased radiation. Enhancing the DeGIR registry with follow-up features would benefit assessing VAA occlusion rates. The data provide a comprehensive nationwide overview of how interventional radiology can impact treatment strategy. Viszeralarterienaneurysmen (VAA) sind selten, und die Daten zu ihrem Auftreten und ihrer Behandlung sind oft veraltet oder heterogen. Das Ziel dieser Analyse ist es, einen umfassenden Überblick über interventionelle Behandlungsoptionen und Eigenschaften der VAA zu bieten. Es wurden prozedurale und klinische Daten zu VAA von 2018 bis 2022 analysiert. Die Daten stammen aus dem Deutschen Register für Interventionelle Radiologie und Minimalinvasive Therapie (DeGIR) und umfassen demografische Details, Merkmale der Aneurysmen, Behandlungsprozesse und Ergebnisse wie dem technischen Erfolg, Komplikationsraten und die Strahlenexposition. Gruppenvergleiche zwischen Aneurysmalokalisationen und -typen als abhängige Variablen sowie dem Auftreten von einer primären unvollständigen Embolisation, Komplikationsraten, technischem Erfolg, Interventionsdauer und Dosisflächenprodukt (DFP) wurden berechnet. Es wurden 647 Patienten eingeschlossen. Das Durchschnittsalter betrug 66 Jahre, mit 68,5% Männern. Am häufigsten betroffen waren die Milz- (27,5%) und die Nierenarterien (21,3%). Wahre Aneurysmen waren am häufigsten (56,7%). Der technische Erfolg betrug 91,2%, während eine primäre unvollständige Embolisation bei 47,3% auftraten. Komplikationen traten bei 4,8% auf, mit einer Mortalitätsrate von 0,2%. Die durchschnittliche Interventionsdauer betrug 67 Minuten, die durchschnittliche Fluoroskopiezeit 36 Minuten und das DFP durchschnittlich 9560 cGycm2. Es gab keine signifikanten Unterschiede in primärer unvollständiger Embolisation, Komplikationsrate, korrekter Platzierung von Embolisationsmaterial, Dauer und DFP basierend auf der Aneurysmalokalisation. Aneurysmatypen zeigten einen signifikanten Unterschied im Auftreten einer primären unvollständigen Embolisation, wobei Dissektionen die höchste Rate und die längsten Interventionen sowie das höchste DFP aufwiesen. Die endovaskuläre Behandlung von VAA zeigt hohe technische Erfolgs- und niedrige Komplikationsraten. Dissezierende Aneurysmen zeigten insgesamt die höchste Rate an einer primären unvollständigen Embolisation, erforderten längere Interventionen und dadurch eine erhöhte Strahlenbelastung. Die Erweiterung des DeGIR-Registers um Follow-up-Daten würde dazu beitragen, die Okklusionsraten von VAA besser zu beurteilen. Die Daten zeigen einen umfassenden nationalen Überblick darüber, wie die interventionelle Radiologie die Behandlungsstrategie beeinflussen kann. Received: 10 June 2024 Accepted after revision: 04 January 2025 Article published online: 20 February 2025 © 2025. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany

  • Open Access

    Deep-Learning-based Automated Identification of Ventriculoperitoneal-Shunt Valve Models from Skull X-rays

    by Marius Vach, Daniel Weiss, Vivien Lorena Ivan, Christian Boschenriedter, Luisa Wolf, Thomas Beez, Björn B. Hofmann, Christian Rubbert, Julian Caspers

    Introduction: Ventriculoperitoneal shunts (VPS) are an essential part of the treatment of hydrocephalus, with numerous valve models available with different ways of indicating pressure levels. The model types often need to be identified on X‑rays to assess pressure levels using a matching template. Artificial intelligence (AI), in particular deep learning, is ideally suited to automate repetitive tasks such as identifying different VPS valve models. The aim of this work was to investigate whether AI, in particular deep learning, allows the identification of VPS models in cranial X‑rays. Methods: 959 cranial X‑rays of patients with a VPS were included and reviewed for image quality and complete visualization of VPS valves. The images included four VPS model types: Codman Hakim (n = 774, 81%), Codman Certas Plus (n = 117, 12%), Sophysa Sophy Mini SM8 (n = 35, 4%) and proGAV 2.0 (n = 33, 3%). A Convolutional Neural Network (CNN) was trained using stratified five-fold cross-validation to classify the four VPS model types in the dataset. A finetuned CNN pretrained on the ImageNet dataset as well as a model trained from scratch were compared. The averaged performance and uncertainty metrics were evaluated across the cross-validation splits. Results: The fine-tuned model identified VPS valve models with a mean accuracy of 0.98 ± 0.01, macro-averaged F1 score of 0.93 ± 0.04, a recall of 0.94 ± 0.03 and a precision of 0.95 ± 0.08 across the five cross-validation splits. Conclusion: Automatic classification of VPS valve models in skull X‑rays, using fully automatable preprocessing steps and a CNN, is feasible. This is an encouraging finding to further explore the possibility of automating VPS valve model identification and pressure level reading in skull X‑rays.