The aim of the study was to investigate whether the ventriculoperitoneal (VP) shunt valve setting can be reliably assessed using maximum intensity projection (MIP) reconstructions from non-contrast, full-dose head CT scans, and how this method performs in comparison to conventional lateral skull radiographs. This retrospective study included 41 adult patients (mean age 59 ± 25 years) with Codman Certas programmable VP shunt valves who underwent lateral skull X‑ray and a same-day, non-contrast head CT scan between January and July 2024. From the CT data, MIP reconstructions of the valve region were generated. Three neuroradiologists, blinded to each other’s assessments, independently rated valve settings and image quality using a 5-point Likert scale. Mean reconstruction time was recorded. Radiation dose data were extracted from institutional dose-monitoring software. Valve settings were identifiable in all 44 CT/X-ray image pairs, with 95

Abstract Bone mineral density (BMD) is a biomarker for frailty, and CT-derived radiodensity can be extracted fully automatically as a surrogate. Because these measurements might be affected by image noise, which varies substantially in the clinical routine, this systematic large-animal study investigates the consistency of CT-based BMD measurements under different radiation dose settings. Twenty Göttingen minipigs underwent six non-contrast CT examinations with five dose levels (CTDIvol: 0.53–10.01 mGy; 5%, 10%, 20%, 40%, 100%; 600 scans). BMD was assessed using CT-derived radiodensity (Hounsfield units, HU) by segmenting the complete ninth thoracic vertebra, and by placing a region of interest (ROI) in the trabecular bone. RM-ANOVA was used to assess statistical significance. Data are presented as mean with standard deviation. The BMD measurement remained consistent between the control and the different dose settings. Even the lowest dose setting (5%: complete = 761 [± 56] HU, ROI = 749 [± 72] HU) showed no significant differences compared to the control (complete = 756 [± 55] HU, ROI = 738 [± 68] HU). Finally, CT-based BMD measurements remained consistent and are therefore robust to substantial dose reduction, indicating the technical feasibility of comparing CT examinations with different dose protocols, relevant for opportunistic screening.

Abstract Accurate field-of-view (FoV) prescription in oblique coronal and axial planes is essential for high-quality prostate MRI but remains operator-dependent and variable. We developed and evaluated a ResNet-based deep learning framework for automated FoV planning. In this retrospective multicenter study, FoV prescriptions were annotated on PI-CAI dataset. Three readers assessed intra- and inter-rater variability to establish reference consistency. Three neural network variants were trained on 1,474 examinations from PI-CAI dataset (2012–2021), and the optimal model was selected by internal validation. Generalizability and clinical utility were tested on three external cohorts totaling 530 examinations (2021–2024) using a non-inferiority design. The selected model achieved non-inferior performance for slice positioning, with differences ranging from 0.16 ± 0.99 to 0.37 ± 0.48. Across sites, FoV overlaps ranged from 82.4 ± 4.1% to 88.7 ± 6.0%, and the angle differences between predicted and reference planes were 4.66 ± 4.89° (Site I), 3.46 ± 2.80° (Site II), and 2.99 ± 2.90° (Site III). Clinical utility was high at all sites, with acceptability rates of 97.9%, 97.7%,98.8%, 98.1% and 98.1% for Site I (Raters 1–5), 95.7%, 97.8%, 100%, 95.7% and 97.8% for Site II (Raters 1–5), and 100% for all raters at Site III. These findings demonstrate the feasibility of automated FoV positioning for prostate MRI and indicate excellent clinical utility.

BACKGROUND AND OBJECTIVE:Children with severe neurological disorders are at risk of secondary respiratory morbidity due to impaired airway clearance and dysphagia, but systematic data on structural lung changes remain scarce. METHODS:We retrospectively analyzed all clinically indicated chest CT examinations at a tertiary care center (2015-2025) in 34 children with severe neurological disorders (median age 10 years), excluding those with primary lung disease. Exploratory hierarchical clustering identified clinical phenotypes based on disease etiology, respiratory support, dysphagia, and mobility. RESULTS:Structural abnormalities were common. Lobar consolidation (76%) and bronchial wall thickening (62%) were the most frequent CT findings. Cluster analysis identified three phenotypes: a stable neuromuscular phenotype without bronchiectasis or Pseudomonas aeruginosa colonization, an advanced neuromuscular-dysphagic phenotype with the highest bronchiectasis prevalence (64%, p = 0.005), and a neurologic-dysphagic phenotype characterized by frequent consolidations (86%) and ground-glass opacities (36%). Although most scans were elective (76%), CT prompted management changes in 91% of patients, mainly intensified airway clearance, antibiotic treatment, and escalated respiratory support. Among CT abnormalities associated with subsequent management change, 60% were not clearly identifiable on retrospectively reviewed preceding chest radiographs. CONCLUSION:In this selective cohort of children with severe neurological disorders undergoing clinically indicated chest CT, structural lung abnormalities were common despite the absence of a primary pulmonary diagnosis. Exploratory clinical phenotyping suggested distinct risk patterns linked to dysphagia, impaired airway clearance, and microbial colonization, and may help identify patients in whom CT provides clinically relevant additional information. Prospective studies are warranted before broader CT-based strategies can be recommended.

The role of balloon guide catheters (BGC) in endovascular thrombectomy for acute ischemic stroke remains controversial, with conflicting evidence regarding clinical outcomes. We evaluated the impact of BGC use on periprocedural metrics and early neurological improvement (ENI). In this retrospective study, 541 patients with anterior circulation large vessel occlusion (intracranial ICA/M1/M2) undergoing endovascular thrombectomy at a tertiary stroke center (2018–2025) were 1:1 propensity-score-matched based on baseline characteristics (BGC+ [n = 91] vs. BGC- [n = 91]). Primary outcome was ENI (NIHSS reduction ≥ 8 points at 24 h or NIHSS < 2 at discharge). Secondary outcomes included reperfusion efficacy (groin-to-reperfusion time, first-pass effect [FPE], reperfusion grade), safety outcomes (any intracranial hemorrhage), in-hospital mortality, and length of hospital stay. BGC use was associated with higher FPE (52

PurposeThe standard modality for the diagnosis of ventriculoperitoneal (VP) shunt failure is the radiographic shunt series (RSS). However, ultra-low dose computed tomography (ULD-CT) may replace RSS. The aim of this study was to compare the radiation doses of RSS and ULD-CT in the diagnosis of mechanical shunt failure in human phantoms including pediatric phantoms and to further reduce the total CT dose by reducing the topogram radiation.MethodsMechanical VP shunt complications were placed on human phantoms representing ages of 1, 5, 10 and 30 years. RSS and ULD-CT were performed on each phantom with different radiation doses of the topogram with varying tube currents (10, 20, 30, 40 and 50 mAs). Effective doses of RSS and ULD-CT were estimated by using conversion factors.ResultsULD-CT demonstrated lower effective doses than RSS in phantoms representing ages 5, 10 and 30 years, while successfully depicting all mechanical shunt complications. However, higher effective doses were assessed for ULD-CT scans of the 1-year phantom than RSS. The effective doses for RSS and ULD-CT (utilizing 10 mAs topograms), respectively, were estimated as follows: 1-year: 0.056 vs. 0.133 mSv; 5-year: 0.186 vs. 0.123 mSv; 10-year: 0.240 vs. 0.107 mSv; 30-year: 0.641 vs. 0.076 mSv.ConclusionThis study demonstrated ULD-CT as an alternative to RSS for the detection of mechanical VP shunt complications, reducing radiation doses in phantoms equivalent to 5 years of age and older while demonstrating the complications equally to RSS. This may be of clinical interest, especially in children due to the reduction of radiation risks.

This study evaluated the relationship between image quality level (IQL), radiation dose, image noise, and signal-to-noise ratio (SNR) in pediatric head CT, comparing energy-integrating detector CT (EID-CT) and photon-counting CT (PCCT) across age-specific phantoms. Head CT scans of 1-, 5-, and 10-year-old anthropomorphic phantoms were performed across a range of IQLs on EID-CT and PCCT. For each IQL, radiation dose, image noise, and SNR were recorded and compared between the two systems. PCCT required lower CTDIvol across all phantoms (p < 0.05). In the clinically relevant IQL-range (IQL 200–300) the 10-year-old phantom showed the greatest relative radiation dose reduction using PCCT (9

Although the radiation dose impact of planning scans prior to the actual computed tomography (CT) scan has been addressed in previous studies, their systematic assessment across paediatric age groups and their relevance for establishing local diagnostic reference levels (DRLs) have not been thoroughly investigated and are not currently included in guideline recommendations. To investigate the influence of radiation dose from topogram, pre-monitoring, and monitoring scans by age on total radiation exposure and whether their doses need to be considered in DRLs and to establish local DRLs for these scan components. 102 chest CT scans with contrast material monitoring from 101 children (median age 12 years, interquartile range (IQR) 7-16 years) and 4498 chest CT scans from 4476 adults (⩾18 years) between January 2021 and May 2023 were retrospectively analysed. Paediatric scans were grouped into 5 age groups and compared with adult scans. The dose-length product (DLP) of topogram, pre-monitoring, and monitoring scans were evaluated and their effect on total radiation dose was analysed. Local DRLs were established for 5 different age groups. Radiation dose proportions of planning scans to total DLP in chest CT increased with decreasing age group (mainlyp< 0.05), with the monitoring scan having the greatest impact. The median topogram proportion increased from 0.3% (⩾18 years) to 1.6% (<1 year), the pre-monitoring proportion from 0.5% (⩾18 years) to 3.5% (<1 year), and the monitoring proportion from 2.2% (⩾18 years) to 26.9% (<1 year). In children under 10 years, monitoring scans in chest CT have a large impact on radiation dose, contributing a median of 13%-27% of the total radiation dose compared to adults (median: 2.2%). Therefore, these planning scans should be monitored and included in radiation dose optimization efforts such as DRLs.

Intracranial artery dissection (IADS) is an uncommon cause of acute ischemic stroke (AIS). In hyperacute settings, stenting is constrained by the need for intensive antiplatelet therapy. This study evaluated the antithrombogenic hydrophilic polymer–coated pEGASUS-HPC self-expanding stent for IADS-related AIS. This multicenter retrospective cohort study across seven stroke centers included consecutive patients with AIS and occlusion or severe flow-limiting stenosis caused by symptomatic IADS, who were treated with the pEGASUS-HPC. The outcomes were technical and angiographic success (device opening; mTICI score ≥2b when reperfusion was targeted), peri-procedural complications, mRS score, and follow-up patency. In 36 treated patients (mean age 62.0 ± 19.6 years; 63.9

INTRODUCTION:Accurate prescription of oblique coronal and oblique sagittal field of views (FOV) is essential for diagnostic shoulder MRI. Manual planning is radiographer-dependent, time-consuming, and subject to inter- and intra-operator variability, leading to inconsistent image quality and incomplete coverage. Although deep learning (DL) has advanced automated scan planning in non-oblique planes, oblique shoulder prescriptions remain underexplored; an automated DL approach could standardize FOV prescription, reduce operator dependence, and improve reproducibility and workflow without compromising diagnostic quality. METHODS:In this retrospective multicenter study, 575 shoulder MRI examinations (2019-2025) from four sites were included. Sites A (n=151) and B (n=220) were used for training; testing was performed on sites C (n=61), and D (n=143). A two-stage pipeline was implemented using five oriented bounding box (OBB) variants of YOLOv11 (n, s, m, l, x): Stage 1 performed slice selection; Stage 2 performed FOV prescription. Performance was evaluated against radiographers' prescriptions using mean absolute slice difference (MASD, slices), intersection over union (IoU), and mean absolute angle difference (MAAD, degrees). Clinical utility was assessed by three raters. RESULTS:The YOLOv11-OBB-l model achieved the lowest MASD for Stage 1 (1.016±0.153 slices). For Stage 2, YOLOv11-OBB-x performed best (coronal IoU, 0.847±0.003; sagittal IoU, 0.852±0.007; MAAD, 3.259±0.190°). During testing across each site, MASD ranged from 0.700±0.837 to 1.192±2.550 slices; MAAD from 2.811±2.348 to 4.396±7.158°; coronal IoU from 0.800±0.092 to 0.872±0.065; and sagittal IoU from 0.824±0.111 to 0.887±0.047. Mean clinical utility was 97.2%. Performance was noninferior to interrater variability across all sites and metrics. CONCLUSION:DL-based automated FOV prescription for shoulder MRI achieves performance comparable to radiographers, generalizes across institutions, and demonstrates high clinical utility.

PURPOSE:To determine the incidence and characteristics of severe cardiorespiratory events during intra-arterial chemotherapy (IAC) for retinoblastoma and to evaluate a potential association with the COVID-19 pandemic. METHODS:This single-center retrospective observational study included 108 children who underwent 280 IAC procedures between January 2017 and May 2024. Severe cardiorespiratory events were defined by acute oxygen desaturation accompanied by increased peak airway pressure, reduced tidal volume, bradycardia, or hypotension. Patient demographics, procedural details, and periprocedural outcomes were extracted from anesthesia and radiology records. A multivariable generalized linear mixed model with a logit link was used to assess associations with severe cardiorespiratory events while accounting for repeated procedures within patients. RESULTS:Severe cardiorespiratory events occurred in 18 procedures (6.4%) among 15 patients, increasing from 3 procedures (2.2%) before the COVID-19 pandemic to 15 procedures (10.5%) during or after the COVID-19 pandemic (P=0.011). Procedures performed during or after the pandemic were associated with higher odds of severe cardiorespiratory events (OR 7.1, 95% CI 1.6 to 31.8). Most events occurred during the second or subsequent IAC sessions. Management included manual ventilation, increased oxygen delivery, propofol bolus, and deepening of anesthesia. No procedures were interrupted due to severe cardiorespiratory events. CONCLUSION:Severe cardiorespiratory events during IAC were more frequent during and after the COVID-19 pandemic, often presenting as bronchospasm-like events. All events were managed successfully without procedural interruption. Given the retrospective design and the lack of individual SARS-CoV-2 infection data, these findings should be considered hypothesis-generating and require confirmation in larger prospective multicenter studies.

BACKGROUND:Dual antiplatelet therapy (DAPT) is commonly used for flow diversion but may increase hemorrhagic complications. The surface-modified flow diverter (FD) p48 MW HPC may allow treatment with prasugrel-based single antiplatelet therapy (SAPT). Our study compared the safety and effectiveness of prasugrel-based SAPT vs DAPT in patients with unruptured saccular aneurysms arising from distal parent vessels treated with p48 MW HPC. METHODS:We conducted a multicenter, retrospective cohort study of adults with unruptured saccular aneurysms arising from distal parent vessels treated with the p48 MW HPC FD. The primary effectiveness outcome was complete aneurysm occlusion within 12 months. The primary safety outcome was MRI-detected silent ischemic lesions within 48 hours post-procedure, as well as periprocedural complications. Secondary outcomes included hemorrhagic complications, ischemic events, and all-cause mortality within 180 days after treatment. RESULTS:Overall, 101 patients were treated (mean age, 60.7 years; 69.3% women). SAPT and DAPT were used in 49 (48.5%) and 52 (51.5%) patients, respectively. Complete occlusion within 12 months was achieved in 79.6% (SAPT) vs 70.6% (DAPT) (P=0.36). Rates of ischemic and hemorrhagic events did not differ significantly (2.0% vs 7.7%, P=0.191; 0% vs 5.8%, P=0.088, respectively). In multivariable analysis, diabetes mellitus was an independent negative predictor of complete aneurysm occlusion (OR 0.26, 95% CI 0.07 to 0.96; P=0.043). Antiplatelet regimen (SAPT vs DAPT; OR 3.39, 95% CI 0.38 to 30.47; P=0.276), aneurysm neck size (P=0.104), aneurysm location (P=0.510), stent diameter (P=0.570), and proximal and distal parent-vessel diameters (P=0.398 and P=0.314, respectively) were not independently associated with complete occlusion. CONCLUSION:In this multicenter retrospective cohort of distal small-vessel flow diversion with the p48 MW HPC, prasugrel-based SAPT was not associated with a significant increase in thromboembolic/ischemic events compared with DAPT; angiographic occlusion and hemorrhage signals were exploratory, with numerically fewer intracranial hemorrhages under SAPT without statistical significance.

This study provides a Rapid Analysis and Processing of Image Data (RAPID) framework that combines deep learning-based CT topogram analysis with DICOM spatial geometry to enable reliable anatomical labelling of CT series independent of inconsistent textual metadata. In this single-centre retrospective study, three YOLOv8-based models comprising the RAPID framework were trained on CT topograms to perform global anatomical classification, body-region detection, and landmark detection. Classification used 83207 topograms (20,802 test), while landmark and body region detection models were trained on 2000 (500 test) and 1926 (481 test) topograms, respectively, collected between 2003 and 2022. Model performance was evaluated using the F1 score and mAP50, with additional external validation on the external cohort. Furthermore, three radiologists independently reviewed 150 randomly selected predictions for detection models using a Likert-scale-based clinical assessment with inter-rater agreement. Across a total of 65,250 patients (median age, 62 years; interquartile range, 23; 44

Background To develop and evaluate a deep learning–based method for delineating scan areas on CT localizers to reduce eye lens exposure during brain CT. Methods In this retrospective study, 2,175 adult brain CT scans were collected from internal and external cohorts. Ground-truth scan areas aligned to the supraorbitomeatal line were annotated in consensus by a radiographer and a radiologist. YOLOv11 rotated object detection models (N,S,M,L,X) were trained with five-fold cross-validation. Performance was evaluated using absolute angular difference (AAD) and the Dice similarity index. The best-performing model was applied to independent test cohorts using an ensemble. Predicted scan areas were used to reconstruct CT volumes, which were assessed for eye lens inclusion and cranial coverage and compared against clinical routine acquisitions. Statistical significance was tested with one-sided test for paired proportions, evaluating superiority in reducing eye lens inclusion and non-inferiority with respect to cranial coverage. Results During cross-validation, the YOLOv11-M model achieved near-perfect alignment with the ground truth (AAD, 1.11°±1.03°; Dice-index, 0.974±0.012). On independent test cohorts, predicted scan areas maintained high accuracy (AAD, 1.12–1.55°; Dice-index, 0.959–0.972). Reconstructed CT volumes demonstrated significant reductions in eye lens inclusion by 76.5% in the internal cohort and by 95.4% and 87.7% in the external cohorts. Cranial coverage was preserved, with rates comparable to routine practice, indicating non-inferior brain coverage. Conclusion Automated delineation of brain CT localizers using a deep learning model substantially reduced unnecessary eye lens exposure while maintaining cranial coverage. This approach may improve patient safety when integrated into clinical workflows.

Purpose:Computed tomography (CT) plays a central role in oncologic imaging, yet repeated examinations contribute substantially to cumulative radiation exposure. This study aimed to evaluate inter- and intra-individual radiation dose variability in chest and abdominal CT and the impact of CT device model and protocol standardization. Materials and Methods:In this retrospective single-center study, 42441 CT scans from 4986 adult oncologic patients were analyzed. Dose metrics (CTDIvol, DLP, SSDE, effective dose) were extracted using automated dose monitoring. Inter- and intra-individual radiation dose variability was assessed across four CT device models and various protocol subtypes. Intra-individual radiation dose variability was calculated relative to the lowest dose per patient and compared across CT devices and protocol subtypes. Results:Radiation dose varied substantially between devices, with CTDIvol differences of up to 2.4-fold in chest CT (2.98-7.26 mGy) and 1.7-fold in abdominal CT (5.26-8.77 mGy). The median intra-individual radiation dose variability was 93.7% (IQR 19.8-142.0%) in non-contrast chest CT, 66.3% (31.1-105.4%) in contrast-enhanced chest CT, 19.8% (11.8-32.3%) in non-contrast abdominal CT, and 28.2% (16.6-41.0%) in contrast-enhanced abdominal CT. When consecutive scans were performed on the same scanner, intra-individual radiation dose variability decreased to 14.7% (IQR 8.1-33.1%), 18.1% (9.5-31.3%), 11.7% (7.8-19.8%), and 15.3% (8.3-24.7%), respectively, indicating substantial device-specific effects. Conclusion:Significant radiation dose variability persists in oncologic CT, both between and within patients, despite the use of standardized protocols. Device-adapted dose management and consistent device use may improve dose consistency, support optimization in oncologic imaging, and reduce radiation exposure. Key Points:· Substantial radiation dose variability persists across CT devices despite protocol standardization.. · Intra-individual radiation dose variability is significant and highest in non-contrast chest CT.. · Consistent use of the same scanner reduces intra-individual radiation dose variability significantly.. · Internal diagnostic reference levels may improve radiation dose consistency and minimize exposure.. Citation Format:· Moradians AF, Rosok D, Serger R et al. Inter- and intra-individual radiation dose variability in oncologic chest and abdominal computed tomography. Rofo 2026; DOI 10.1055/a-2786-2534.

Medical speech is a central interface for numerous tasks in modern healthcare workflows, including documentation, report generation, and structured clinical communication. Although documentation is a cornerstone of clinical care, it imposes a substantial administrative burden on health care providers. Automatic Speech Recognition (ASR) offers a promising solution for real-time documentation. However, general-purpose models such as OpenAI’s Whisper struggle with the linguistic density, domain-specific terminology, and structural conventions characteristic of clinical dictation. Moreover, the development of high-performance medical ASR is constrained by a persistent privacy–utility bottleneck, as access to large-scale, annotated clinical audio data is limited by stringent data protection regulations. This study aims to develop and evaluate a privacy-preserving framework for fine-tuning ASR models on synthetic clinical speech data, eliminating the need for sensitive patient recordings. We investigate whether LLM-generated, terminology-grounded audio can bridge the performance gap between general-purpose ASR and the demands of real-world clinical dictation across medical specialties. We present MumbleMED, a privacy-by-design framework for scalable and configurable fine-tuning of ASR models. Realistic clinical narratives were synthesized using Large Language Models (LLMs) based on standardized coding systems (ICD-10, OPS, and RadLex) and converted into high-quality audio via a German Text-to-Speech (TTS) system cloning institutional voices. The resulting pipeline was used to fine-tune Whisper model variants (Tiny to Large v2) on 33,898 synthetic samples (32 hours of audio). Performance was evaluated against a gold-standard benchmark comprising 175 professionally recorded clinical reports from five medical specialties, including radiology, pathology, discharge letters, progress notes and surgical reports. Fine-tuning substantially reduced transcription error rates across all model sizes. On the Combined TTS test dataset, the best-performing model, Combined MumbleMED based on Whisper Large v2, reduced WER from 36.71% (baseline Whisper Large v2) to 8.20% and CER from 21.07% to 4.20%. The baseline Whisper Large v2 model achieved a Word Error Rate (WER) of 68.22% on authentic clinical dictation, which decreased to 28.90% after MumbleMED fine-tuning, representing an absolute reduction of nearly 40%. Stratified analyses revealed the lowest error rates in surgical and radiology reports (WER ~19%), whereas discharge letters remained more challenging. In a focused evaluation of 6,576 medical terms, term-specific WER decreased from 38.45% to 13.31%. Qualitative assessment showed that models learned to correctly map verbalized punctuation to symbols and normalize complex clinical units. MumbleMED demonstrates that synthetic clinical speech can effectively bridge the domain gap in ASR performance while remaining compliant with regulatory requirements. By enabling local deployment of high-accuracy ASR within secure clinical infrastructures, the framework provides a practical and privacy-preserving pathway for adapting transcription models to clinical needs.

Purpose:Photon-counting computed tomography (PCCT) offers new possibilities for optimizing image quality while reducing radiation dose. However, the influence of acquisition parameters such as tube voltage and spectral shaping with tin filtration on chest imaging remains insufficiently understood. Materials and Methods:Adult male and female anthropomorphic phantoms were scanned using a dual-source PCCT system across a broad range of image quality levels (IQL 1-100). Radiation dose and signal-to-noise ratio (SNR) were evaluated for protocols using 120 kVp and 140 kVp without tin filtration, and 100 kVp with tin filtration. Results:The male phantom required a significantly higher tube current and radiation dose than the female, reflecting sex-based attenuation differences. Increasing the tube voltage to 140 kVp raised the radiation dose with minimal SNR improvement. Tin filtration substantially reduced the dose despite a higher mAs, with the radiation dose and tube current plateauing near IQL 80. In the low-dose range (IQL 1-20), tin filtration enabled dose reduction without saturation. Slight SNR trade-offs between bone and mediastinum highlight the need for tissue-specific considerations when optimizing PCCT chest protocols. Conclusion:Sex- and tissue-specific differences with tin filtration and varying tube voltages underscore baseline differences in radiation dose and image quality behavior between representative adult body geometries. These baseline relationships provide a technical foundation for future patient-specific protocol optimization. Key Points:· A higher tube voltage increased the radiation dose with limited image quality benefit.. · Tin filtration enables substantial radiation dose reduction in low-dose adult chest PCCT.. · Tin filtration causes image quality and dose plateaus at image quality levels around 80.. · Baseline radiation dose and image quality relationships support future patient-specific chest PCCT optimization.. Citation Format:· Klüner LV, Zensen S, aus der Wiesche H et al. Influence of Tube Voltage, Tin Filtration, and Sex on Radiation Dose and Image Quality in Adult Chest Photon-Counting CT: A Phantom-Based Signal-to-Noise-Based Analysis. Rofo 2026; DOI 10.1055/a-2829-0195.

INTRODUCTION:Accurate malignancy risk prediction of pulmonary lesions is essential in the context of lung cancer screening and early diagnosis. Widely used risk prediction tools, such as the Brock and Herder models, have shown limited performance in high-risk cohorts referred for bronchoscopic evaluation. The LIONS PREY (lung lesion score predicts malignancy) was specifically developed to provide a simple and reliable estimate of malignancy risk in this clinical setting. This study aimed to externally validate the LIONS PREY model and to compare its predictive performance with the Brock and Herder models in patients undergoing index navigational bronchoscopy for pulmonary lesion evaluation. METHODS:We retrospectively analysed patients who underwent index navigational bronchoscopy for peripheral pulmonary lesions between December 2019 and March 2024 at a tertiary academic centre. Malignancy risk was estimated using the LIONS PREY, Brock, and Herder models. Model performance was assessed via receiver operating characteristic (ROC) analysis and classification accuracy. RESULTS:Among 193 evaluable lesions, 134 (69.4%) were histologically confirmed malignant. Malignant lesions were associated with increased growth dynamics (2.7 ± 4.1 mm vs. 1.0 ± 4.3 mm, p < 0.001) and spiculation (56.0% vs. 23.7%, p < 0.001). The LIONS PREY achieved an area under the ROC curve (AUC) of 0.94, outperforming Herder (AUC 0.72; p < 0.001) and Brock (AUC 0.62; p < 0.001) models. Correct classification rates were highest for LIONS PREY (85.0%) versus Herder (70.3%) and Brock (69.1%) (all p < 0.001). CONCLUSION:The LIONS PREY demonstrated superior predictive accuracy over Brock and Herder in high-risk patients undergoing navigational bronchoscopy, supporting its clinical utility for malignancy risk stratification.

The Contour Neurovascular System (CNS) is an intrasaccular flow-disrupting device for treating intracranial aneurysms. Conventional follow-up imaging of the CNS with MR angiography or conventional CT angiography often suffers from significant artifacts. Photon-counting CT angiography (PCCTA) is a novel technique offering superior spatial resolution and reduced beam-hardening artifacts. We report one of the first cases of a 48-year-old patient with an anterior communicating artery aneurysm treated with a CNS who underwent follow-up imaging with PCCTA. The examination provided markedly artifact-reduced images, enabling clear visualization of the CNS, complete aneurysm occlusion, and adjacent vessel patency. Compared with MR angiography, PCCTA offered substantially improved depiction of the device itself, highlighting its unique value for precise, non-invasive follow-up of intracranial neurovascular implants and its potential as an alternative for detailed post-procedural assessment of the CNS.

While large language models (LLMs) have shown promise in medical text analysis, their application in automated medical billing code extraction remains underexplored, particularly for the German medical fee schedule system (GOÄ). Therefore, an LLM was fine-tuned to perform multi-label classification of GOÄ codes from radiology reports automatically, and its performance was compared with state-of-the-art commercial and open-source LLMs. Following ethics committee approval, we analyzed 499,601 radiology reports from 124,497 patients, containing 1,799,971 manually identified GOÄ codes as ground truth. The MediPhi-Instruct 4B model was fine-tuned using five-fold cross-validation. Performance was evaluated on the hold-out test set and compared against GPT-5, GPT-4.1, GPT-oss, Kimi-K2, Deepseek-R1, Deepseek-V3, Gemini 2.5, Llama-70B, and Qwen-3 LLMs on a subset of 500 anonymized and 350 cleaned reports using zero-shot and few-shot prompting techniques. The fine-tuned model achieved an accuracy of 77.15