Ulrich Grosse

Ulrich Grosse


Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 December 2025 in Biomedicines
Verified
Journal Article
Open access
PDF file
BABassel Almarie
Bassel Almarie
LGLuis Fernando Gonzalez-Gonzalez
Luis Fernando Gonzalez-Gonzalez
Lucas Antônio dos Santos BarbosaLucas Antônio dos Santos Barbosa
Lucas Antônio dos Santos Barbosa
ALAmelie Lutz
Amelie Lutz
UGUlrich Grosse
Ulrich Grosse
Felipe FregniFelipe Fregni
Felipe Fregni
Background: The US Food and Drug Administration (FDA) authorized over 690 machine learning (ML)-enabled medical devices between 1995 and 2023. In 2024, new guidance enabled the inclusion of Predetermined Change Control Plans (PCCPs), raising expectations for transparency, equity, and safety under the Good Machine Learning Practice (GMLP) framework. Objective: The objective was to assess regulatory pathways, predicate lineage, demographic transparency, performance reporting, and PCCP uptake among ML-enabled devices approved by the FDA in 2024. Methods: We conducted a cross-sectional analysis of all FDA-authorized ML-enabled devices in 2024. Data extracted from FDA summaries included regulatory pathway, predicate genealogy, performance metrics, demographic disclosures, PCCPs, and cybersecurity statements. Descriptive and nonparametric statistics were used. Results: The FDA authorized 168 ML-enabled Class II devices in 2024. Most (94.6%) were cleared via 510(k); 5.4% were cleared via De Novo. Radiology dominated (74.4%), followed by cardiovascular (6.5%) and neurology (6.0%). Non-US sponsors accounted for 57.7% of clearances. Among 159 510(k) devices, 97.5% cited an identifiable predicate; the median predicate age was 2.2 years (IQR 1.2–4.1), and 64.5% ML-enabled. Predicate reuse remained uncommon (9.9%). Median review time was 162 days (151 days for 510(k) vs. 372 days De Novo; p < 0.001). A total of 49 devices (29.2%) reported both sensitivity and specificity; 15.5% provided demographic data. PCCPs appeared in 16.7% of summaries, and cybersecurity considerations appeared in 54.2%. Conclusions: While 2024 marked a record year for ML-enabled device approvals and internationalization, uptake of PCCPs and transparent performance and demographic reporting remained limited. Policy efforts to standardize disclosures and strengthen post market oversight are critical for realizing the promises of GMLP.

Verified
Journal Article
Open access
PDF file
HAHaidara Almansour
Haidara Almansour
MMMigdat Mustafi
Migdat Mustafi
MLMario Lescan
Mario Lescan
UGUlrich Grosse
Ulrich Grosse
MAMateja Andic
Mateja Andic
SWSven Stephan Walter
Sven Stephan Walter
Background/Objectives: To assess free-breathing, dynamic radial magnetic resonance angiography (MRA) for detecting endoleaks post-endovascular aortic repair (EVAR) in cases with inconclusive computed tomography angiography (CTA). Methods: This prospective single-center study included 17 participants (mean age, 70 ± 9 years; 13 males) who underwent dynamic radial MRI (Golden-angle RAdial Sparse Parallel-Volumetric Interpolated BrEath-hold, GRASP-VIBE) after inconclusive multiphasic CT for the presence of endoleaks during the follow-up of EVAR-treated abdominal aortic aneurysms. CT and MRI datasets were independently assessed by two radiologists for image quality, diagnostic confidence, and the presence/type of endoleak. Statistical analyses included interrater and intermethod agreement, and diagnostic performance (sensitivity, specificity, area under the curve (AUC)). Results: Subjective image analysis demonstrated good image quality and interrater agreement (k ≥ 0.6) for both modalities, while diagnostic confidence was significantly higher in MRA (p = 0.03). There was significantly improved accuracy for detecting type II endoleaks on MRA (AUC 0.97 [95% CI: 0.87, 1.0]) compared to CTA (AUC 0.66 [95% CI: 0.41, 0.91]; p = 0.03). Although MRA demonstrated higher values for sensitivity, specificity, AUC, and interrater agreement, none of the other types nor the overall detection rate for endoleaks showed differences in the diagnostic performance over CT (p ≥ 0.12). CTA and MRA revealed slight to moderate intermethod concordance in endoleak detection (k = 0.3–0.64). Conclusions: The GRASP-VIBE MRA characterized by high spatial and temporal resolution demonstrates clinical feasibility with good image quality and superior diagnostic confidence. It notably enhances diagnostic performance in detecting and classifying endoleaks, particularly type II, compared to traditional multiphase CTA with inconclusive findings.

24 February 2026 in Oral Radiology
Verified
Journal Article
Open access
PDF file
GBGintare Burian
Gintare Burian
JSJonas Stief
Jonas Stief
TGTobias Greve
Tobias Greve
UGUlrich Grosse
Ulrich Grosse
AHAdib Al-Haj Husain
Adib Al-Haj Husain
EBEgon Burian
Egon Burian
Objectives To compare caries detection using T1-based Free-breathing 3D Stack of Stars GRE (StarVIBE) and T1-weighted pointwise encoding time reduction with radial acquisition (PETRA) MRI sequences with the gold standard consisting of a combination of clinical evaluation and bitewing radiography. Materials and methods Included subjects were evaluated clinically by a dentist for the occurrence of caries. All subjects underwent bitewing radiographs and MRI of the viscerocranium, including 3D StarVIBE and 3D PETRA sequences. Caries lesions were evaluated for penetration depth (D1-D4). Signal intensity (SI) was measured in dental hard tissue, the pulp and caries lesions. Results 80 teeth were analyzed. Among these, 10 showed caries during clinical assessments and bitewing imaging, all of which were also detected on MRI. Furthermore, 5 additional lesions were identified exclusively on MRI. Overall image quality and diagnostic confidence were significantly higher in PETRA than in StarVIBE (4.7 vs. 3.9, p = 0.006). The SI for caries lesions was significantly higher in the PETRA sequence (-0.5 a.u. ± 1.5 a.u. vs. -4.2 a.u. ± 4.0 a.u., p = 0.002). Caries lesion volume was also significantly larger in PETRA (18 mm3± 14.8 mm3) compared to StarVIBE (11 mm3± 14.8 mm3, p = 0.003). Conclusion MRI identified all caries lesions that were detected both clinically and through bitewing imaging, along with additional lesions exclusive to MRI. The PETRA sequence outperformed StarVIBE in quantitative caries visualization.