Maintaining cognitive function despite the presence of Alzheimer disease (AD) pathology is the foundation of cognitive reserve. Although the theory of cognitive reserve is strongly supported by empirical research, the field lacks standardized, validated methods for quantifying cognitive and brain reserve. We tested whether associations between AD pathology and cognitive function were modified by proxy measures of cognitive reserve (years of education, socioeconomic status; SES) and brain reserve (brain-predicted age difference, and a volumetric AD signature). We hypothesized that greater structural brain integrity, higher education, and higher SES would attenuate the association between greater AD pathology and poorer cognitive performance. This cross-sectional study analyzed baseline data from a multisite randomized clinical trial, which was conducted at 3 US universities and enrolled cognitively unimpaired, physically inactive, community-dwelling adults. AD pathology was measured via plasma assays for phosphorylated tau (p-tau)-217 in the whole cohort, and PET for β-amyloid (Aβ) in a subset of participants as a secondary analysis. The primary outcome of cognitive function was evaluated by a comprehensive cognitive assessment. SES was measured via the MacArthur Socioeconomic Status Index, and magnetic resonance imaging was used to calculate brain-predicted age difference (brain-PAD) and a volumetric AD signature. Data were analyzed using linear regression models with interaction terms for moderation analyses. A total of 621 participants (aged 69.9 ± 3.8, 71% female) had available data for the main analyses and 355 had PET Centiloid data available. Brain-PAD moderated the association between AD pathology (measured by p-tau217) and multiple cognitive domains, including episodic memory (β = -0.09 [-0.16 to -0.02]), processing speed (β = -0.08 [-0.15 to -0.01]), working memory (β = -0.10 [-0.18 to -0.03]), and executive function/attentional control (β = -0.08 [-0.15 to -0.01]). Specifically, the negative association of greater AD pathology with poorer cognition was weakest in individuals with younger appearing brains. A latent SES score also moderated the relationship between p-tau217 and episodic memory (β = 0.08 [0.01-0.16]), but this did not survive correction for multiple comparisons. Neither years of education nor the volumetric AD signature moderated pathology-cognition associations. These results support the hypothesis that higher cognitive and brain reserve may help buffer the cognitive consequences of AD pathology. Strategies to increase both cognitive and brain reserve could help to boost resilience against emerging AD pathology; however, longitudinal studies are needed to confirm these conclusions.

Cerebral small vessel disease (CSVD) is a major cause of dementia and stroke, typically identified by lesions such as white matter hyperintensity (WMH), lacunes, cerebral microbleeds (CMBs), enlarged perivascular spaces, and brain atrophy. Biomarker-based biological age (BA), derived from routinely measured clinical indicators and largely reflecting vascular and metabolic physiologic burden, may provide complementary information beyond chronological age. The aim of this study was to investigate the associations between BA residual and both the presence and progression of CSVD and its neuroimaging manifestations. In the population-based Polyvascular Evaluation for Cognitive Impairment and Vascular Events cohort, participants underwent brain magnetic resonance imaging (MRI) at baseline (2017-2019) and at follow-up (2022-2024). BA was estimated using 3 established methods: PhenoAge, Klemera-Doubal method, and homeostatic dysregulation. Progression of CSVD and its imaging markers, including new lacunes, new CMBs, progression of enlarged perivascular spaces in the basal ganglia (BG-EPVS), and progression of WMHs, was assessed. Associations between BA residual and progression of CSVD were analyzed using logistic or ordinal logistic regression models, as appropriate. A total of 3,050 participants were included at baseline (mean age 61.22 ± 6.67 years; 53.51% female); 2,662 completed follow-up MRI over a median of 4.7 years, and 388 were lost to follow-up. Higher BA residual, particularly KDMAge and PhenoAge residual, was associated with greater CSVD burden at baseline and follow-up. Moreover, KDMAge residual was positively associated with the progression of total CSVD burden (odds ratio [OR] = 1.18, 95% CI 1.08-1.30; p = 0.001), new lacunes (OR = 1.31, 95% CI 1.13-1.51, p < 0.001), and new CMBs (OR = 1.13, 95% CI 1.01-1.25; p = 0.028), but not with progression of BG-EPVS or WMHs at follow-up. PhenoAge residual showed results consistent with those of KDMAge residual, whereas HDAge residual showed no significant association with progression of CSVD or its imaging markers. This community-based cohort study demonstrated that BA residuals, particularly KDMAge and PhenoAge, were associated with higher odds of CSVD progression, especially with development of new lacunes and new CMBs. Biomarker-based BA residual may help identify individuals at higher risk of clinically relevant CSVD progression. Future studies with longer follow-up periods and more diverse cohorts are warranted to validate these findings.

Background: Accurate brain tumor detection is critical for early diagnosis and effective treatment planning in neuro-oncology. Magnetic resonance imaging (MRI) is a cornerstone for identifying and localizing brain tumors, guiding clinical interventions, and enhancing patient outcome. Advanced deep learning models, such as YOLOv8, offer promising solutions for automated and precise tumor detection in MRI. Objectives: This study aimed to develop and evaluate a YOLOv8-based model for the accurate identification and localization of brain tumors, including pituitary tumors, meningiomas, and gliomas, using a publicly available Kaggle dataset of annotated MRI images. Materials and methods: The YOLOv8 pretrained model was employed for brain tumor detection on a Kaggle dataset comprising annotated MRI images, including cases with pituitary, meningioma, and glioma, and no tumor. The dataset was pre-processed and split into training and validation sets for further analysis. The YOLOv8 model was fine-tuned to optimize the tumor detection and localization. Performance metrics, including precision, recall, F-1 score, and mean average precision (mAP), were calculated, and loss values were analyzed to evaluate the model’s learning efficiency. Results: The YOLOv8 model achieved a precision of 98.9%, recall of 98.9%, and accuracy of 99.5%. The mean average precision (mAP) reached 97.6%, indicating a high accuracy in detecting and localizing brain tumors. Loss value analysis demonstrated stable convergence during training, reflecting a robust model performance. Conclusion: The YOLOv8-based approach provides a highly accurate and reliable method for detecting and localizing brain tumors in MRI. With exceptional precision, recall, and mAP, this model demonstrates significant potential for clinical applications, enabling faster and more precise neurooncological diagnosis and treatment planning.

Estimated glomerular filtration rate (GFR, eGFR) from serum creatinine outlines global kidney function, subject to biases in clinical scenarios from muscle wasting and inflammation (prevalent in kidney cancer) and failure to estimate split renal function (SRF, crucial in operative planning). Measured GFR (mGFR) (or true mGFR [mGFRt] without body surface area normalization) from diethylene-triamine-pentaacetate (99mTc-DTPA) tracer clearance is the gold standard for bilateral kidney function, involving extended clearance times and radioactivity. Imaging-derived total kidney volumes are functional proxies but do not probe tissue quality. We employed advanced quantitative diffusion-weighted (DW) magnetic resonance (MR) imaging (MRI) at 3.0 T in addition to kidney volume measurements in a cohort of 27 patients with renal mass (26 and 18 underwent eGFR and mGFR tests, respectively). Cardiac-gated diffusion tensor imaging (DTI) and intravoxel incoherent motion (IVIM) parameters were derived. Individual MR metrics were evaluated for correlation with eGFR and mGFR with Pearson correlations and mixed-model analysis, respectively; LASSO-penalized multivariable regression was employed for mGFR prediction. The metrics were compared between proteinuria groups using 2-sample t tests. Kidney volume correlated with renal function (split volume vs. split mGFR r = 0.54; split volume vs. split mGFRt r = 0.69). MR metrics correlated with individual kidney mGFR and mGFRt (r = 0.76 and 0.81, respectively). Mixed-effects LASSO multiple regression analysis predicted mGFR and mGFRt (R2 = 0.880 and 0.700, respectively). In addition, MR metrics differentiated proteinuria status. Advanced DW MRI metrics may provide surrogates of mGFR and proteinuria. Parameters from bipolar encoding in diastole (emphasizing tubular flow) and flow compensation in systole (emphasizing vascular flow) were often informative.

Innovation in magnetic resonance imaging (MRI) is often focused on achieving higher magnetic field strengths, allowing for increased sensitivity and image resolution. Although many clinical settings have access to 1.5 T or 3.0 T MRI systems, 7.0 T systems are becoming more common and human research magnets as high as 11.7 T are in use (Boulant et al. Nature Methods, 2024, 21, 2013–2016). However, higher field is not always better. For example, higher magnetic fields experience more significant variations (B0 and B1 field inhomogeneities) that can introduce image artefacts. High magnetic fields also come at a high price, with the rule of thumb that the price scales directly with the field strength. To increase accessibility and have a broader health impact, the use of low-field MRI systems is being investigated in clinical populations ranging from stroke (Bhat et al. Journal of Magnetic Resonance Imaging, 2021, 54, 372–390) to pregnancy (Aviles Verdera et al. Radiology, 2023, 309, e223050). In this issue of BJOG, Bansal et al. used a 0.55 T MRI to image the cervix in late gestation in 97 low-risk pregnancies as part of the MiBirth study. This prospective cohort study, run by a multidisciplinary team and a Patient and Public Involvement group, aimed to use a combination of imaging modalities (ultrasound, MRI) and relevant anatomy (uterus, cervix, pelvis, placenta and foetus) to predict the mode of birth. In this work, the authors assessed the feasibility of using low-field MRI to provide measurements of cervical remodelling. They found their image reconstruction and automated segmentation protocols were of good quality, with high inter-rater reliability for cervical length, volume, and internal and external os diameters. A larger cervical stroma volume during late gestation, suggesting failure to remodel, was associated with an increased risk of caesarean section. Although knowledge about the possible mode of delivery is important in preparation for the birth experience, there is still much work to be done to accurately predict adverse birth outcomes (e.g., preterm birth) using MRI. The potential benefits of this study are twofold. First, compared with ultrasound, MRI may be able to detect more subtle changes in cervical morphology and provide details of cervical microstructure and hydration (Oláh, BJOG, 1994, 101, 255–257). Second, low-field MRI has several advantages including reduced image artefacts, the potential for a larger bore size to accommodate pregnant individuals and a lower cost. With the rising price and uncertainty of cryogen availability (liquid nitrogen and liquid helium), another advantage of low-field MRI systems is that they do not require superconducting magnets and therefore can be cryogen-free. They may also be designed to be portable, allowing the system to be brought to the bedside and increasing accessibility to this advanced medical imaging modality. Future areas of research should include a longitudinal study design in a cohort at high risk for preterm birth. Almost 70% of the participants were White and imaged at one hospital. Future studies should focus on using low-field MRI in more diverse populations and settings. In tandem with clinical studies using low-field MRI systems, biomedical engineers and imaging physicists must continue to develop software and hardware solutions (e.g., denoising methods and low-noise electronics) to improve the image quality at low fields. This may usher in a new and exciting era of initial screening and monitoring during pregnancy using low-field MRI. The author takes full responsibility for this article. The author has nothing to report. The author declares no conflicts of interest. The author has nothing to report.

Exposure to high acoustic noise levels generated during 7 Tesla (T) magnetic resonance imaging (MRI) may affect auditory function. To better understand these effects, this study examines how such noise exposure influences outer hair cell function in healthy adults, with a focus on the efficiency of a dual hearing protection method. In this within-subjects longitudinal study, outer hair cell function was assessed in 39 participants (aged 18-41 years) using distortion product otoacoustic emissions (DPOAEs). Participants completed two 7 T MRI sessions on the same day while wearing dual hearing protection (foam earplugs in combination with sound-absorbing impression paste). DPOAEs were measured before the first MRI session and immediately after both MRI sessions, with follow-up assessments on a separate day to detect any enduring alterations in auditory function. Among the 39 participants, DPOAEs demonstrated no significant variations in outer hair cell function across the assessed time points, suggesting that cochlear function remained stable. This study suggests that, with adequate dual hearing protection, repeated exposure to high acoustic noise during 7 T MRI does not adversely affect outer hair cell function in healthy adults. Our findings highlight the importance of rigorous hearing protection in MRI, particularly in ultra-high-field settings. Dual hearing protection, combining foam earplugs with sound-absorbing impression paste, is shown to effectively prevent measurable cochlear deterioration during repeated 7 T MRI sessions. These results support adopting multiple protective barriers as a practical standard and contribute to auditory safety in diagnostic imaging.

Radiographers' perspectives on interactions with patients exhibiting cognitive impairment and dementia during magnetic resonance imaging examinations.

Magnetic resonance imaging (MRI) is a valuable clinical and research tool for patients managed using deep brain stimulation (DBS). Unfortunately, MRI under these conditions is associated with substantial risks, necessitating stringent regulations that limit its clinical and research utility. In addition, magnetic susceptibility differences between DBS hardware and surrounding tissues significantly compromise spatial encoding mechanisms of conventional MRI sequences, resulting in image signal loss and geometric distortions. The impact on gradient-recalled echo echo-planar imaging sequences commonly utilised for functional MRI in DBS settings is particularly severe. This review presents a range of mitigation strategies aimed at both safety and enhanced image quality, spanning innovations in DBS hardware design to advanced MRI sequences capable of addressing issues inherent to the presence of DBS hardware, such as electrode heating and susceptibility artefacts. Additionally, we highlight approaches incorporating the discussed novel postprocessing techniques and functional MRI acquisition protocols, along with their limitations and associated challenges to enable their wider dissemination, with the overarching objective of improving the quality of life of DBS patients.

Hypoxic-ischemic brain injury (HIBI) is a well-described sequela of pediatric cardiac arrest, but the epidemiology and clinical implications of hypoxic-ischemic spinal cord injury (HISCI) remain poorly understood. Only isolated reports describe HISCI following cardiopulmonary resuscitation (CPR). We aimed to describe the incidence, imaging characteristics, and clinical context of HISCI inpediatriccardiac arrest patients undergoing clinically indicated MRI. We conducted a single-center retrospective descriptive case series of consecutively identified pediatric cardiac arrest patients who underwent spinal magnetic resonance imaging (MRI) within two weeks of resuscitation (2018-2023). Cases were identified from an institutional cardiac arrest database. MRI scans were independently reviewed by a pediatric neuroradiologist for evidence of HISCI. Of 717 cardiac arrest patients, 36 (5%) underwent spinal MRI within two weeks of arrest, primarily for trauma evaluation (72%). Four patients (11%) had MRI evidence of HISCI. All four experienced out-of-hospital cardiac arrest with CPR durations ranging from 8 to 90min and initial serum lactate>4 mmol/L. Two arrests were traumatic. All four patients had concomitant HIBI, and two met criteria for death by neurologic criteria. Among the 32 patients without HISCI, 9 (28%) had HIBI and 19 (59%) had traumatic arrest. HISCI was identified in 11% of pediatric cardiac arrest patients who underwent post-arrest spinal MRI for clinical indications. Recognition of HISCI has potential implications for neuroprognostication, rehabilitation planning, and determination of brain death by neurologic criteria. Larger prospective studies are needed to define the incidence, risk factors, and outcomes of HISCI following pediatric cardiac arrest.

Anti-GluK2 antibody-associated autoimmune encephalomyelitis with delayed MRI abnormalities: case report.

The pathophysiological processes that drive disability progression in multiple sclerosis (MS) are not adequately captured by clinically available imaging biomarkers. Slowly expanding lesions (SELs) are a novel magnetic resonance imaging (MRI) biomarker proposed to reflect histopathologically defined chronic active lesions, which are associated with disease progression. SELs are identified through complex MRI analysis pipelines that measure volume changes in chronic MS lesions across serial MRI scans. This review provides a comprehensive overview of SELs, covering their proposed histopathological correlates, associations with clinical outcomes and response to disease-modifying therapies (DMTs). A distinguishing aim of this review is to provide the reader with a clear understanding of different SEL analysis techniques, particularly the two most established automated approaches, including their respective strengths and limitations. Each technique is compatible with conventional clinical MRI sequences, making clinical translation feasible. Current DMTs have modest effects on SELs, paralleling their limited efficacy in progressive MS. SELs may become a key outcome measure in clinical trials of DMTs for progressive MS. However, further research is needed to ascertain the optimal parameters for identifying clinically meaningful chronic lesion expansion with each analysis technique, and to clarify which SEL measures are most strongly associated with disability progression.

Evaluation of condyle histogram data based on magnetic resonance imaging, condylar morphology, and disc position in patients with jaw deformities.

The purpose of this study is to perform an independent assessment of three state-of-the-art tools for the detection of focal cortical dysplasia (FCD) from Magnetic Resonance images (MRI). These tools include DeepFCD, the Multi-center Epilepsy Lesion Detection (MELD) Classifier, and MELDGraph. T1-weighted and fluid-attenuated inversion recovery MRIs from 101 epilepsy patients with FCD and 101 epilepsy patients without FCD were retrospectively included. Classifiers were evaluated at patient-level by their ability to correctly identify the presence of any FCD lesions, and at lesion-level by their capacity to identify lesions within regions delineated by neuroradiologists in MRI reports. A calibrated threshold for DeepFCD prediction probabilities was empirically determined to improve classifier specificity. Classifier test-retest consistency was measured using the Dice coefficient on repeated MRI scans of 21 individuals. At patient-level, MELDClassifier achieved 52% accuracy (sensitivity=91%, specificity=14%), MELDGraph reached 61% accuracy (sensitivity=76%, specificity=47%) and DeepFCD performed with 56% accuracy (sensitivity=62%, specificity=50%) at an empirically determined threshold of 0.90. At lesion-level, MELDClassifier performed with a sensitivity of 70% and a positive predictive value (PPV) of 13%. MELDGraph reached 53% sensitivity and PPV of 36%, whereas the DeepFCD performed with 30% sensitivity and PPV of 19%. Test-retest reliability was low, with an average [min, max] Dice coefficient of 0.28 [0.0, 1.0] for MELDClassifier, 0.38 [0.0, 1.0] for MELDGraph, and 0.35 [0.05, 0.54] for DeepFCD. This study highlights the current limitations of using deep learning models in FCD diagnosis and emphasizes the need to enhance the tools' accuracy, reliability, and interpretability to improve clinical utility.

Pontocerebellar hypoplasia type 2A (PCH2A) is a rare autosomal recessive neurodegenerative disease caused by a specific pathogenic variant in the TSEN54 gene (p.A307S). Affected children show early but initially unspecific symptoms, diagnosed primarily through postnatal magnetic resonance imaging (MRI), with confirmation by genetic testing. This study examines the diagnostic process and key considerations for accurate diagnosis. We retrospectively collected data from 65 children (33 girls, 32 boys) with genetically confirmed PCH2A as part of a Natural History Study. Data were gathered via parental questionnaires, interviews, and medical reports. The cohort was divided into two groups based on year of birth: children born before (n = 30) and after (n = 35) the identification of the pathogenic variant in 2008. Prenatally, in 4 of 21 cases with specialized ultrasound (gestational weeks 12-32), only unspecific cerebellar abnormalities were reported. One fetal MRI (week 31) revealed clear cerebellar hypoplasia, in two others (week 21 and 31), slight cerebellar abnormalities were reported. Postnatal neurosonography often indicated disease features (26/54), later confirmed by MRI (62/63). Clinical symptoms appeared at a median age of 0 months (range 0-6 months), often initially suggesting acute rather than congenital issues. In the group born after 2008, median time from first symptoms to genetic confirmation was 5 months. PCH2A presents early with nonspecific symptoms. Prenatal and postnatal ultrasound imaging can fail to detect the condition, with MRI being the gold standard for diagnosis. Over time, the diagnostic process, including genetic confirmation, has become faster.

This study aims to evaluate the diagnostic performance of a ResNet50-based convolutional neural network (CNN) in detecting osteochondral lesions of the talus (OLTs) on magnetic resonance imaging (MRI) and to compare its efficacy between T1- and T2- weighted sequences. A total of 219 ankle MRI scans were reviewed retrospectively, including 60 with confirmed OLTs and 159 without lesions. From each study, coronal and sagittal T1- and T2-weighted images were extracted and standardized to 224 × 224 pixels. Augmentation techniques were applied to strengthen model training. Data were divided into training, validation, and test sets in a 60:20:20 split. A ResNet50 model initialized with ImageNet weights was fine-tuned using crossentropy loss with class weighting. Diagnostic performance was summarized with accuracy, precision, recall, and F1-scores. The model performed better on T1 sequences, achieving an accuracy of 94.1% (95% confidence interval [CI] 88.3-97.1%) and an area under the curve [AUC] of 0.93 (95% CI 0.87-0.97), with patient cases classified at 0.92 precision and 0.82 recall. Healthy controls in the T1 group were recognized with 0.95 precision and 0.98 recall. In contrast, T2 sequences were less reliable, showing an accuracy of 87.2% (95% CI 80.5-91.9%) and an AUC of 0.91 (95% CI 0.85-0.95). Precision for patient cases in the T2 group was notably lower (0.65) despite a recall of 0.81. Misclassifications were more frequent in the T2 dataset, as evidenced by the confusion matrices. Even with a relatively modest dataset, the ResNet50 model delivered strong results for T1-weighted MRI. While T2 images proved more challenging, suggesting that deep learning can add value to routine assessment of OLTs.

Fractal analysis and magnetic resonance imaging (MRI) semantic features to identify intracranial solitary fibrous tumours and atypical meningiomas.

To investigate dose heterogeneity in liver tumors and nontumor target liver (NTTL) following transarterial radioembolization (TARE) by developing an experimental magnetic resonance (MR) imaging-compatible ex vivo perfusion model based on human tumor-bearing livers and to validate the observed heterogeneity patterns against in vivo data. Fractionated TARE was performed under MR imaging in 4 machine-perfused human tumor-bearing liver explants using fluorescent holmium microspheres. Dose heterogeneity was quantified by calculating the homogeneity index (HI) from MR imaging-based dose maps (voxel size, <2.5 mm). These results were validated against HI values from 2 TARE-treated patients. Fluorescence microscopy was used to assess the microscopic distribution of 4 distinct microsphere fractions. MR imaging-based dose maps revealed lower heterogeneity in liver tumors (mean HI, 2.41; range, 0.72-4.43) than in nontumor target liver (mean HI, 2.95; range, 1.58-5.94); however, this difference was not significant (P = .06) and was primarily driven by higher microsphere concentrations in tumors, which were associated with reduced heterogeneity (ρ = -0.88, P < .001). Microspheres administered in consecutive fractions decreased the HI while mostly preserving the spatial distribution pattern of earlier fractions, as confirmed by fluorescence microscopy. TARE induces heterogeneous dose distributions in both liver tumors and nontumor target liver at a scale below the resolution of nuclear imaging. Although these findings provide insight into microsphere distribution and dose heterogeneity, the clinical significance of fine-scale dose heterogeneity and its potential impact on treatment outcomes remains uncertain and warrants further investigation.

Segmentation of uterus and myoma from MRI image based on deep learning.

The study aimed to evaluate the utility of 3D transrectal ultrasonography (3D-TRUS) for perianal Crohn's disease (pCD) activity stratification. Retrospectively enrolled Crohn's disease patients underwent both 3D-TRUS and magnetic resonance imaging (MRI) within 4 weeks between January 2021 and June 2023, excluding those with recent perianal surgery (≤ 4 weeks), suboptimal image quality, or lesions beyond detection range. 3D-TRUS examined parameters as follows: the number of fistulas, the length, diameter, echogenicity, color Doppler and internal openings of the main fistula, rectal wall thickness, gas in the fistula and collections. MRI was used as the reference standard. The least absolute shrinkage and selection operator regression was used to identify key ultrasound features for nomogram construction. Model performance in terms of accuracy, discrimination and clinical utility was evaluated by receiver operating characteristic curves, calibration curves and decision curve analysis, respectively. Our study included a total of 105 patients, with 49 classified as high activity (median age, 27 years [13-65 years]) and 56 as low activity (median age, 30 years [13-79 years]). A nomogram for identifying perianal CD activity was constructed, incorporating ultrasound characteristics including the number of fistulas, echogenicity and length of main fistula, collections, and gas in the fistula. The area under the curve of this model was 0.934 (95% CI: 0.883-0.988), and it exhibited good internal reliability and clinical net benefit. 3D-TRUS serves as a clinically valuable imaging method for assessing pCD activity, and the nomogram developed based on this modality demonstrates good diagnostic efficacy. Question The study explores the role of 3D transrectal ultrasonography(3D-TRUS) in evaluating perianal Crohn's disease (pCD) activity. Findings The nomogram model based on 3D-TRUS has good differentiation ability for pCD activity. Clinical relevance 3D-TRUS is a reliable tool for distinguishing disease activity of pCD and serves as a clinically valuable imaging modality, facilitating clinical management and follow-up.

Reduced Rate of Complications With Transperineal Prostate Biopsy: Real-world Data From a National Cohort Derived From the Danish Prostate Cancer Registry.