Background: Breast clip marker movement after ultrasound-guided biopsy can negatively affect lesion re-localisation rates and surgical outcomes, underscoring the need for improved understanding of the factors influencing clip displacement. Thus, this study aimed to compare four different breast clip markers and identify risk factors for clip migration and dislocation after ultrasound-guided placement. Methods: This retrospective study included 350 patients who underwent ultrasound-guided biopsy of a newly diagnosed breast lesion with placement of one of four types of breast clips (UltraClip Dual Trigger Biodur 108 Coil Marker [UC], TUMARK Professional [TP], TUMARK Vision [TV] and HydroMARK Breast Biopsy Site Marker [HM]). Clip migration and dislocation were assessed immediately after placement and during follow-up imaging for at least 3 months. A binary logistic regression analysis was performed to identify predictors of clip dislocation including lesional, perilesional and procedural parameters. Results: Clip migration rates were 26.0%, 18.0%, 10.0% and 25.0% and clip dislocation rates were 14.0%, 20.0%, 9.0% and 38.0% for UC, TP, TV and HM, respectively. Features significantly associated with clip dislocation included predominantly fatty surrounding tissue (p = 0.046) with low perilesional shear wave velocities (p = 0.054), smooth lesion contours (p = 0.041), soft lesion strain elastography (p =0.001), low clip-to-lesion-surface distance (p = 0.002) and the use of an HM breast clip (p = 0.032). Conclusions: The type of breast clip-marker, as well as perilesional and lesional characteristics, influence the likelihood of clip dislocation. Notably, the hydrogel-coated clip (HM) exhibited the highest rate of dislocation.

This study evaluates the feasibility and technical success rate of real-time virtual sonography (RVS) for prone contrast-enhanced breast MRI sequences during second-look examinations. Usually, additional supine MRI sequences are acquired for coregistration. This single-center retrospective study was performed in a cohort of female patients who underwent contrast-enhanced prone breast MRI followed by second-look ultrasound for MRI-detected incidental lesions. RVS was used to coregister supine ultrasound and prone MRI data without requiring additional supine MRI studies. Lesion localization success, as well as lesion visibility, fusion quality, and histopathological correlation through ultrasound-guided biopsy, were assessed. A covariate analysis of factors affecting lesion localization was performed. A total of 103 female patients (mean age 48.3 ± 11.0 years) with 125 MRI-detected breast lesions were included. Of the lesions, 91.2% were successfully localized using RVS, including a high proportion of non-mass enhancements (41.6%). Ultrasound-guided biopsy was performed in 57.6% of cases, confirming malignancy in 31.9% of those. Covariate analysis identified higher breast volume as the only factor significantly associated with reduced RVS coregistration success (odds ratio 0.993, p = 0.035). RVS represents an advanced imaging approach in breast diagnostics, offering a promising solution to overcome the limitations of standalone modalities and potentially enhance diagnostic accuracy. We showed that prone MRI studies may be sufficient for RVS-based coregistration of breast lesions, potentially rendering additional supine MRI acquisitions unnecessary. This study demonstrates that real-time virtual sonography with contrast-enhanced MRI in the prone position is a feasible and effective method for localizing MRI-detected breast lesions on second-look ultrasound without the need for additional supine MRI. This approach can optimize diagnostic workflows and reduce imaging burden while maintaining high localization rates. RVS localizes MRI-detected breast lesions in the prone position without requiring additional supine MRI for co-registration. Successful localization was achieved in 91.2%, including 41.6% non-mass enhancements, with higher breast volume as the only detrimental factor. RVS enables accurate lesion localization without supine MRI, streamlining workflows, lowering imaging costs, and improving biopsy access. RVS localizes MRI-detected breast lesions in the prone position without requiring additional supine MRI for co-registration. Successful localization was achieved in 91.2%, including 41.6% non-mass enhancements, with higher breast volume as the only detrimental factor. RVS enables accurate lesion localization without supine MRI, streamlining workflows, lowering imaging costs, and improving biopsy access.

Background/Objectives: Coronary artery disease (CAD) remains the leading cause of death worldwide. Traditional cardiovascular risk assessment is based on chronological age and other clinical factors, with inherent limitations and poor accuracy. Objective was to estimate the artificial intelligence (AI)-enhanced biological cardiovascular age calculation derived from coronary computed tomography angiography (CTA) reports using a large language model (LLM), in predicting major adverse cardiovascular events (MACE). Methods: Coronary CTA reports were analyzed using a LLM (ChatGPT-4.0v, OpenAI), from symptomatic patients with suspected CAD who underwent coronary CTA for clinical indications. Patients in which the LLM successfully analyzed the key metrics (1) coronary artery calcium (CAC) score and (2) coronary CTA reports (coronary stenosis severity (CAD-RADS), high-risk anatomy, non-calcified plaque, cardiac function (LVEF and others) were included. Results: 386 CTA reports were uploaded, and 346 (89.6%) included. The mean biological age (bioAGE) was 57.2 ± 10.9 and the chronological 58.5 ± 10.8 years. 137 (39.6%) were women. The intra-individual deviation in bioAGE was high (median: 8.8; IQR 9.98). BioAGE exceeded chronological age in 45.4% patient and was lower or equal in 54.6%) MACE rate was 8.7% comprising 2 deaths, 5 myocardial infarctions, and 22 late revascularizations. The accuracy for prediction of MACE was higher for bioAGE (c = 0.768; 95% CI: 0.681–0.855, p < 0.001) compared to chronological age (c = 0.590; 95% CI: 0.492–0.689, p = 0.102) Conclusions: Biological age calculation from coronary CTA reports using LLM is feasible, yet intra-individual deviations are high. The accuracy for prediction of MACE is improved by bioAGE compared to chronological.

Artificial intelligence (AI) could facilitate and objectify quality assessment in the daily routine. The purpose was to explore the extent to which an AI prototype algorithm is able to replicate the perfect-good-moderate-inadequate (PGMI) system (perfect, good, moderate, inadequate). From a multicentre case collection, 200 standard mammograms (800 images) were selected. A deep learning-based prototype software was used to rate the images in analogy to the PGMI system. The AI results were compared with a reference standard obtained through consensus reading by three expert radiographers and one expert radiologist, using quadratically weighted Cohen’s kappa with confidence intervals (CI) and context-based interpretation. Frequency and reasons for disagreement were evaluated for challenging cases with a discrepancy of two or more grades and a discrepancy in assigning an inadequate. For overall PGMI per image, slight agreement between human consensus and AI was observed for CC views (κ = 0.14) and fair agreement for MLO views (κ = 0.25). The highest agreement was observed for the CC category “M. Pectoralis visibility” (substantial, κ = 0.75). Best category in MLO was “Pectoralis angle” (moderate, κ = 0.49). For other categories, fair, slight or poor agreement was observed. The work-up of disagreement gave insight into misinterpretations of anatomical landmarks and causality issues in the categorization. Transforming the PGMI system into a fully automated AI algorithm is challenging and may differ substantially between subcategories. Further research in computer science and quality assessment methodology is needed to pave the way for AI-based objective quality management in mammography. Profound evaluation of AI algorithms and their ability to replicate human interpretation, scoring, and classification are the basis and scientific framework toward AI-based objective quality management in mammography. AI has huge potential for automated assessment of diagnostic image quality. Compared with human reading agreement, substantial disagreement may also be found. Direct transformation of perfect-good-moderate-inadequate scoring into an AI algorithm is challenging. AI has huge potential for automated assessment of diagnostic image quality. Compared with human reading agreement, substantial disagreement may also be found. Direct transformation of perfect-good-moderate-inadequate scoring into an AI algorithm is challenging.

Background: Vascular calcification is a frequent consequence of ageing and is associated with an increased risk of cardiovascular disease. This study aimed to compare two rapid scoring systems for quantifying calcification of the distal abdominal aorta and iliac arteries and to investigate correlations with increasing age. Methods: Patients aged ≥65 years who sustained pelvic trauma between 2003 and 2023 and underwent computed tomography (CT) were included in this retrospective study. Patients were categorised into three age groups (65–74, 75–84, ≥85). The abdominal aorta calcification score (AACS) and the common, external, and total iliac artery calcification scores (CIACS, EIACS, TIACS) were assessed on cross-sectional images and classified into three severity grades (mild, moderate, severe). Results: A total of 224 patients (mean age 78.8 ± 8.5 years; 62% female) were included. Significant differences between age groups were identified for hypertension (p < 0.001), osteoporosis (p < 0.001), atrial fibrillation (p = 0.015), chronic heart failure (p = 0.004), chronic kidney disease (p < 0.001), neurocognitive disorders (p < 0.001), and anticoagulant therapy (p = 0.002). Calcification severity increased with age across all vascular territories (EIACS p = 0.006; others p < 0.001). In multivariable linear regression, age remained the strongest adjusted predictor of calcification across all vascular regions (β = 0.323–0.376, all p < 0.001). Significant positive correlations were found between aortic and iliac calcifications (all p < 0.001), strongest between AACS and CIACS (ρ = 0.78, CI 0.719–0.835) and TIACS (ρ = 0.745, CI 0.676–0.807). Corresponding categorical associations were most pronounced between AACS and CIACS. Conclusions: The evaluated calcification scores were strongly correlated and demonstrated clear age-dependent trends. Given their simplicity and applicability to routine CT imaging, these methods may provide practical tools for assessing vascular ageing.

Medical image registration is crucial for various clinical and research applications including disease diagnosis or treatment planning which require alignment of images from different modalities, time points, or subjects. Traditional registration techniques often struggle with challenges such as contrast differences, spatial distortions, and modality-specific variations. To address these limitations, we propose a method that integrates learnable edge kernels with learning-based rigid and non-rigid registration techniques. Unlike conventional layers that learn all features without specific bias, our approach begins with a predefined edge detection kernel, which is then perturbed with random noise. These kernels are learned during training to extract optimal edge features tailored to the task. This adaptive edge detection enhances the registration process by capturing diverse structural features critical in medical imaging. To provide clearer insight into the contribution of each component in our design, we introduce four variant models for rigid registration and four variant models for non-rigid registration. We evaluated our approach using a dataset provided by the Medical University across three setups: rigid registration without skull removal, with skull removal, and non-rigid registration. Additionally, we assessed performance on two publicly available datasets. Across all experiments, our method consistently outperformed state-of-the-art techniques, demonstrating its potential to improve multi-modal image alignment and anatomical structure analysis.

Deformable medical image registration is a fundamental task in medical image analysis. While deep learning-based methods have demonstrated superior accuracy and efficiency, they often overlook the critical role of regularization in ensuring robustness and anatomical plausibility. We propose DARE (Deformable Adaptive Regularization Estimator), a novel registration framework that dynamically adjusts elastic regularization based on the gradient norm of the deformation field. The method integrates strain and shear energy terms, adaptively modulated to balance deformation stability and flexibility, and includes a folding-prevention mechanism that penalizes regions with negative deformation Jacobian to reduce folding artifacts and encourage topologically plausible mappings. Extensive experiments on the IXI, OASIS, and MUI-P datasets demonstrate that DARE consistently improves registration accuracy while preserving anatomical plausibility. Compared with classical regularizers and state-of-the-art learning-based methods, DARE achieves higher Dice scores and reduces folding artifacts, while maintaining low strain energy and realistic volume changes in key brain structures. These results confirm that the proposed adaptive regularization mechanism delivers robust, accurate, and physiologically consistent deformation fields across clinical scenarios related to brain registration.

Purpose: Accurate target volume delineation is critical for effective stereotactic radiotherapy (SRT) of brain metastases. This study systematically investigates how MRI sequence selection and the time elapsed after contrast agent (CA) administration affect the apparent metastases volumes, with the goal of optimizing MRI protocols for radiation therapy planning. Materials and Methods: A total of 49 patients with 414 brain metastases were included and randomized into 6 groups with varying imaging sequences (MPRAGE, SPACE, and VIBE) and timepoints after CA administration. Lesions smaller than 0.03 cm3 were excluded due to resolution limitations. Lesion volumes were independently assessed by radiology and radiation oncology specialists, and mean values were analyzed. The effects of MRI sequence and time delay on lesion volume were evaluated using t tests, ANOVA, and multiple linear regression. Results: Both MRI sequence and CA timing significantly influenced measured volumes. On average, SPACE volumes were 20% larger than MPRAGE, and VIBE volumes were 10% larger than SPACE, independent of timing. Lesion volumes increased progressively with time after CA administration at rates of 0.63%, 0.58%, and 0.36% per minute for MPRAGE, SPACE, and VIBE, respectively. Smaller lesions (<1 cm3) showed greater relative intersequence differences, primarily due to variations in visible lesion borders. Conclusions: Both MRI sequence choice and imaging time after CA administration significantly affect the apparent volume of brain metastases in SRT planning. Although SPACE and VIBE sequences enhance small lesion detection, they may also increase border blurring and inter-rater variability. Standardizing protocols to account for these factors is essential for improving delineation accuracy, reducing toxicity risk, and optimizing SRT outcomes.

Occlusive cervical artery dissection (CeAD) is associated with worse patient outcome. The net clinical benefit of acute revascularization measures has to be weighed against the likelihood of spontaneous recanalization. Our aim was to assess the hitherto un-addressed impact of spontaneous recanalization on stroke risk in patients with occlusive CeAD. MRI verified CeAD patients with initially occlusive CeAD within cohort study that did not undergo acute revascularization measures were assessed. Follow-up data derived from clinical routine and study specific assessments. Outcomes of interest were occurrence of (i) recanalization and (ii) ischemic stroke upstream of CeAD-related occlusion. Adjusted logistic regression analysis addressed the impact of recanalization on said outcomes. 97/328 (29.6%) patients had occlusive CeAD and did not undergo acute revascularization treatment. Upon follow-up, 56/97 (57.7%) showed spontaneous recanalization of initially occlusive CeAD. Female sex (OR 0.41[0.18, 0.97]; P = 0.043) and internal carotid artery dissection (OR 0.33[0.14, 0.78]; P = 0.012) were the only factors independently associated with recanalization. Within a median follow-up of 8.2 (1.58, 12.8) years, a total of 18/97 (18.6%) patients suffered ischemic stroke upstream of the initially CeAD-affected vessel. After adjusting for confounders, spontaneous recanalization was independently associated with lower rates of cerebral ischemia upon follow-up (OR 0.28[0.09, 0.90]; P = 0.032), most notably also independent of type of antithrombotic treatment. Spontaneous recanalization in occlusive CeAD is associated with lower rates of stroke upon follow-up. These results indicate that persistent CeAD-related occlusion remains a risk-factor for recurrent ischemic events, thus calling for future trials addressing optimal medical treatment. N/A. Lukas Mayer-Suess.

Background: Comorbid personality disorders (PDs) in patients with anorexia nervosa (AN) are associated with increased psychopathology, higher suicide risk, and poorer treatment response and outcomes. This study aimed to examine associations between gray matter (GM) volume and PDs in female adolescents with AN before and after short-term psychotherapeutic and nutritional therapy. Methods: Eighteen female adolescents with acute AN, mean age 15.9 years, underwent 3T magnetic resonance imaging before and after weight restoration. The average interval between scans was 2.6 months. Structural brain changes were analyzed using voxel-based morphometry. PDs were assessed using the Structured Clinical Interview for DSM-IV Axis II Disorders (SCID II) and the Assessment of Identity Development Questionnaire. Results: SCID-II total scores showed significant positive associations with GM volume in the mid-cingulate cortex at both time points and in the left superior parietal–occipital lobule at baseline. The histrionic subscale correlated with GM volume in the thalamus bilaterally and the left superior parietal–occipital lobule in both assessments, as well as with the mid-cingulate cortex at follow-up. Borderline and antisocial subscales were associated with GM volume in the thalamus bilaterally at baseline and in the right mid-cingulate cortex at follow-up. Conclusions: PDs in female adolescent patients with AN may be specifically related to GM alterations in the thalamus, cingulate, and parieto-occipital regions, which are present during acute illness and persist after weight restoration therapy.