Objectives: The aim of the study was to implement a deep-learning tool to produce synthetic double inversion recovery (synthDIR) images and compare their diagnostic performance to conventional sequences in patients with multiple sclerosis (MS). Materials and Methods: For this retrospective analysis, 100 MS patients (65 female, 37 [22–68] years) were randomly selected from a prospective observational cohort between 2014 and 2016. In a subset of 50 patients, an artificial neural network (DiamondGAN) was trained to generate a synthetic DIR (synthDIR) from standard acquisitions (T1, T2, and fluid-attenuated inversion recovery [FLAIR]). With the resulting network, synthDIR was generated for the remaining 50 subjects. These images as well as conventionally acquired DIR (trueDIR) and FLAIR images were assessed for MS lesions by 2 independent readers, blinded to the source of the DIR image. Lesion counts in the different modalities were compared using a Wilcoxon signed-rank test, and interrater analysis was performed. Contrast-to-noise ratios were compared for objective image quality. Results: Utilization of synthDIR allowed to detect significantly more lesions compared with the use of FLAIR images (31.4 ± 20.7 vs 22.8 ± 12.7, P < 0.001). This improvement was mainly attributable to an improved depiction of juxtacortical lesions (12.3 ± 10.8 vs 7.2 ± 5.6, P < 0.001). Interrater reliability was excellent in FLAIR 0.92 (95% confidence interval [CI], 0.85–0.95), synthDIR 0.93 (95% CI, 0.87–0.96), and trueDIR 0.95 (95% CI, 0.85–0.98). Contrast-to-noise ratio in synthDIR exceeded that of FLAIR (22.0 ± 6.4 vs 16.7 ± 3.6, P = 0.009); no significant difference was seen in comparison to trueDIR (22.0 ± 6.4 vs 22.4 ± 7.9, P = 0.87). Conclusions: Computationally generated DIR images improve lesion depiction compared with the use of standard modalities. This method demonstrates how artificial intelligence can help improving imaging in specific pathologies.

Background The benefits of using low-field-strength fetal MRI to evaluate antenatal development include reduced image artifacts, increased comfort, larger bore size, and potentially reduced costs, but studies about fetal low-field-strength MRI are lacking. Purpose To evaluate the reliability and feasibility of low-field-strength fetal MRI to assess anatomic and functional measures in pregnant participants using a commercially available 0.55-T MRI scanner and a comprehensive 20-minute protocol. Materials and Methods This prospective study was performed at a large teaching hospital (St Thomas' Hospital; London, England) from May to November 2022 in healthy pregnant participants and participants with pregnancy-related abnormalities using a commercially available 0.55-T MRI scanner. A 20-minute protocol was acquired including anatomic T2-weighted fast-spin-echo, quantitative T2*, and diffusion sequences. Key measures like biparietal diameter, transcerebellar diameter, lung volume, and cervical length were evaluated by two radiologists and an MRI-experienced obstetrician. Functional organ-specific mean values were given. Comparison was performed with existing published values and higher-field MRI using linear regression, interobserver correlation, and Bland-Altman plots. Results A total of 79 fetal MRI examinations were performed (mean gestational age, 29.4 weeks ± 5.5 [SD] [age range, 17.6-39.3 weeks]; maternal age, 34.4 years ± 5.3 [age range, 18.4-45.5 years]) in 47 healthy pregnant participants (control participants) and in 32 participants with pregnancy-related abnormalities. The key anatomic two-dimensional measures for the 47 healthy participants agreed with large cross-sectional 1.5-T and 3-T control studies. The interobserver correlations for the biparietal diameter in the first 40 consecutive scans were 0.96 (95% CI: 0.7, 0.99; P = .002) for abnormalities and 0.93 (95% CI: 0.86, 0.97; P < .001) for control participants. Functional features, including placental and brain T2* and placental apparent diffusion coefficient values, strongly correlated with gestational age (mean placental T2* in the control participants: 5.2 msec of decay per week; R² = 0.66; mean T2* at 30 weeks, 176.6 msec; P < .001). Conclusion The 20-minute low-field-strength fetal MRI examination protocol was capable of producing reliable structural and functional measures of the fetus and placenta in pregnancy. Clinical trial registration no. REC 21/LO/0742 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Gowland in this issue.

Objective: The number of diagnosed fatal encephalitis cases in humans caused by the classical Borna disease virus (BoDV‐1) has been increasing, ever since it was proved that BoDV‐1 can cause human infections. However, awareness of this entity is low, and a specific imaging pattern has not yet been identified. We therefore provide the first comprehensive description of the morphology of human BoDV‐1 encephalitis, with histopathological verification of imaging abnormalities.Methods: In an institutional review board‐approved multicenter study, we carried out a retrospective analysis of 55 magnetic resonance imaging (MRI) examinations of 19 patients with confirmed BoDV‐1 encephalitis. Fifty brain regions were analyzed systematically (T1w, T2w, T2*w, T1w + Gd, and DWI), in order to discern a specific pattern of inflammation. Histopathological analysis of 25 locations in one patient served as correlation for MRI abnormalities.Results: Baseline imaging, acquired at a mean of 11 ± 10 days after symptom onset, in addition to follow‐up scans of 16 patients, revealed characteristic T2 hyperintensities with a predilection for the head of the caudate nucleus, insula, and cortical spread to the limbic system, whereas the occipital lobes and cerebellar hemispheres were unaffected. This gradient was confirmed by histology. Nine patients (47.4%) developed T1 hyperintensities of the basal ganglia, corresponding to accumulated lipid phagocytes on histology and typical for late‐stage necrosis.Interpretation: BoDV‐1 encephalitis shows a distinct pattern of inflammation in both the early and late stages of the disease. Its appearance can mimic sporadic Creutzfeldt–Jakob disease on MRI and should be considered a differential diagnosis in the case of atypical clinical presentation. ANN NEUROL 2020;88:723–735

Background and purpose: Inflammation has been linked to poor prognoses in cardio- and cerebrovascular conditions. As it is known to increase after ischemia, C-reactive protein (CRP) may serve as a surrogate for systemic inflammation and thus be a hallmark of increased tissue vulnerability. The question arises whether CRP in the acute phase of ischemic stroke, prior to mechanical thrombectomy (MT), might help predict outcomes.Materials and methods: A single-center collective of patients with large-vessel occlusion, who were treated via MT, was analyzed in this observational case–control study. Univariate and multivariate models were designed to test the prognostic value of inflammatory markers (CRP and leukocytosis) in predicting clinical outcomes (modified Rankin score >2) and all-cause mortality 90 days after MT.Results: A total of 676 ischemic stroke patients treated with MT were included. Of these, 313 (46.3%) showed elevated CRP levels (≥5 mg/l) on admission. Poor clinical outcome and mortality at 90 days occurred in 113 (16.7%) and 335 (49.6%) patients and significantly more frequently when initial CRP levels were elevated [213 (64.5%) vs. 122 (42.1%), p < 0.0001, and 79 (25.2%) vs. 34 (9.4%), p < 0.0001, respectively]. CRP levels were highly predictive for impaired outcomes, especially in patients with atrial fibrillation, in both univariate and multivariate models. Interestingly, patients with initially elevated CRP levels also showed more pronounced increases in CRP post-MT.Conclusion: Poor outcome and death occur significantly more often in stroke patients with elevated CRP levels before MT. Our findings suggest that stroke patients with atrial fibrillation and elevated inflammatory markers are of particular risk for poor outcomes.

Generative adversarial networks (GANs) can synthesize high-contrast MRI from lower-contrast input. Targeted translation of parenchymal lesions in multiple sclerosis (MS), as well as visualization of model confidence further augment their utility, provided that the GAN generalizes reliably across different scanners. We here investigate the generalizability of a refined GAN for synthesizing high-contrast double inversion recovery (DIR) images and propose the use of uncertainty maps to further enhance its clinical utility and trustworthiness. A GAN was trained to synthesize DIR from input fluid-attenuated inversion recovery (FLAIR) and T1w of 50 MS patients (training data). In another 50 patients (test data), two blinded readers (R1 and R2) independently quantified lesions in synthetic DIR (synthDIR), acquired DIR (trueDIR) and FLAIR. Of the 50 test patients, 20 were acquired on the same scanner as training data (internal data), while 30 were scanned at different scanners with heterogeneous field strengths and protocols (external data). Lesion-to-Background ratios (LBR) for MS-lesions vs. normal appearing white matter, as well as image quality parameters were calculated. Uncertainty maps were generated to visualize model confidence. Significantly more MS-specific lesions were found in synthDIR compared to FLAIR (R1: 26.7 ± 2.6 vs. 22.5 ± 2.2 p < 0.0001; R2: 22.8 ± 2.2 vs. 19.9 ± 2.0, p = 0.0005). While trueDIR remained superior to synthDIR in R1 [28.6 ± 2.9 vs. 26.7 ± 2.6 (p = 0.0021)], both sequences showed comparable lesion conspicuity in R2 [23.3 ± 2.4 vs. 22.8 ± 2.2 (p = 0.98)]. Importantly, improvements in lesion counts were similar in internal and external data. Measurements of LBR confirmed that lesion-focused GAN training significantly improved lesion conspicuity. The use of uncertainty maps furthermore helped discriminate between MS lesions and artifacts. In conclusion, this multicentric study confirms the external validity of a lesion-focused Deep-Learning tool aimed at MS imaging. When implemented, uncertainty maps are promising to increase the trustworthiness of synthetic MRI.

Aims: To investigate the incremental prognostic value of morphological plaque features beyond clinical risk and coronary stenosis levels. Although associated with the degree of coronary stenosis, most cardiac events occur on the basis of ruptured non-obstructive plaques and consecutive vessel thrombosis. As such, identification of vulnerable plaques is paramount for cardiovascular risk prediction and treatment decisions. Methods and results: A total of 1615 patients with suspected but not previously diagnosed coronary artery disease (CAD) were examined by coronary computed tomography angiography and morphological plaque features were assessed. Mean follow-up was 10.5 (interquartile range 9.2–11.4) years. Cox proportional hazards analysis was used for the composite endpoint of cardiac death and non-fatal myocardial infarction. The study endpoint was reached in 51 patients (36 cardiac deaths, 15 non-fatal myocardial infarctions). In addition to quantitative parameters (presence of any calcified/non-calcified plaque or elevated plaque load), morphologic plaque features such as a spotty or gross calcification pattern and napkin-ring sign (NRS) were predictive for events. However, only spotty calcified plaques and NRS could confer additive prognostic value beyond clinical risk and coronary stenosis level. In a stepwise approach, endpoint prediction beyond clinical risk (Morise score) could be improved by inclusion of CAD severity (χ2 of 27.5, P < 0.001) and further discrimination for spotty calcified plaques (χ2 of 3.89, P = 0.049). Conclusion: Improved cardiovascular risk prediction beyond clinical risk and coronary stenosis levels can be made by discriminating for the presence of spotty calcified plaques. Thus, an intensified prophylactic anti-atherosclerotic treatment appears to be warranted in patients with coronary plaques that show spotty calcifications.