Centre de Recherche en Acquisition et Traitement de l'Image pour la Santé, French National Centre for Scientific Research

Background Patients with ARDS have heterogeneous lungs which exposes them to the risk of lung injury exacerbation by mechanical ventilation. Functional lung CT imaging gives a comprehensive description of regional lung mechanical behaviour. Here, we investigated whether CT registration-based regional lung function parameters are associated with survival in patients with COVID-ARDS. Methods We conducted a two-centre prospective observational study of adult COVID-ARDS patients with an indication for CT within 72 h of onset. Dual volume CT images were aligned using image-registration. Regional lung functional parameters, and their spatial distributions, were analysed by univariable Cox proportional hazard models with survival as the main outcome. Selected variables based on the univariable analysis were included in a stepwise Cox model adjusted for age, sex, body mass index and SAPSII. Results 94 patients were included in the study. Recruitment was associated with a higher (HR = 1.45, p = 0.023) hazard of death, while apical (sΔVz) and central (sΔVx) displacement of specific volume change centre-of-mass were associated with a lower hazard of death (HR = 0.72, p = 0.041; HR = 0.68, p = 0.031, respectively). Conclusions Our data show that in addition to recruitment, the spatial distribution of specific volume change, a surrogate measure of regional lung ventilation, is associated with the risk of death in mechanically ventilated COVID-19 ARDS patients. Our findings suggest that CT image-registration based functional biomarkers may have prognostic value in COVID-ARDS patients. Trial registration This study was retrospectively registered in Clinical Trials under NCT06113276 (https://clinicaltrials.gov/study/NCT06113276) on 27/10/2023.

BACKGROUND Diagnosing cerebral amyloid angiopathy (CAA) after spontaneous lobar intracerebral hemorrhage has significant clinical implications. A previous postmortem study found that the presence of both subarachnoid hemorrhage (SAH) and hematoma finger-like projections (FLP) on acute-stage computed tomography strongly rules in CAA. In the present study, we assessed the diagnostic value of these imaging markers against histopathologic diagnosis in less severe presentations. METHODS This retrospective (2002–2022) multicenter study included patients aged ≥45 years with lobar intracerebral hemorrhage of unknown cause, for whom acute-stage computed tomography or magnetic resonance imaging and neuropathological samples from hematoma evacuation or biopsy were available. Centralized consensus reading (2 raters) of imaging and neuropathological data (including Aβ immunohistochemistry) were performed. Analysis was restricted to samples containing at least 10 vessels. The diagnostic performance was evaluated against the neuropathological reference, that is, CAA/no CAA. RESULTS We analyzed data from 66 patients (age, 65±9 years; men, 33%; hematoma volume, 45±26 mL; death within 1 year, 14%) from 6 French centers. Neuropathological material included samples from hematoma evacuation (n=48) and biopsy (n=18). CAA was present in 38 patients (58%), and FLP and SAH were observed in 29 (44%) and 50 (76%) patients, respectively. FLP had a sensitivity of 0.58 (95% CI, 0.41–0.74) and a specificity of 0.74 (95% CI, 0.54–0.89) for the diagnosis of CAA. SAH demonstrated a high sensitivity of 0.92 (95% CI, 0.78–0.98; negative predictive value=0.80 [0.52–0.96]) but moderate specificity of 0.43 (95% CI, 0.24–0.63). The combined presence of FLP and SAH had a specificity of 0.54 (95% CI, 0.37–0.71) and a sensitivity of 0.79 (95% CI, 0.59–0.92). CONCLUSIONS This study is the first to evaluate the diagnostic performance of FLP and SAH with histopathologic reference in nonautopsied patients. The results suggest these markers have lower diagnostic performance than previously reported in severe hematomas leading to early death. However, the high sensitivity of SAH suggests its potential clinical utility in ruling out CAA when absent.

We investigated whether baseline levels of biomarkers related to endotheliopathy, thromboinflammation, and fibrosis were associated with clinical outcomes in hospitalized COVID-19 patients. We analyzed the associations between baseline levels of 21 biomarkers and time to hospital discharge and change in NEWS-2 score in patients from DisCoVeRy trial. We fitted multivariate models adjusted for baseline ISARIC 4C score, disease severity, D-dimer values, and treatment regimen. Between March 22 and June 29, 2020, 603 participants were randomized; 454 had a sample collected at baseline and analyzed. The backward selection of multivariate models showed that higher baseline levels of soluble suppressor of tumorigenicity 2 (sST2) and nucleosomes were statistically associated with a lower chance of hospital discharge before day 29 (sST2: aHR 0.24, 95% CI [0.15–0.38], p < 10⁻⁹; nucleosomes: aHR 0.62, 95% CI [0.48–0.81], p < 10⁻³). Likewise, higher levels of baseline sST2 were statistically associated with lower changes in the NEWS-2 score between baseline and day 15 (adjusted beta 4.47, 95% CI [2.65–6.28], p < 10⁻⁵). Moreover, we evaluated sST2 involvement in a confirmation cohort (SARCODO study, 103 patients) and found that elevated baseline sST2 levels were significantly associated with lower rates of hospital discharge before day 29 and a higher model performance (AUC at day 29 of 92%) compared to models without sST2. sST2 emerged as an independent predictor of clinical outcomes in two large cohort of hospitalized COVID-19 patients, warranting further investigation to elucidate its role in disease progression and potential as a therapeutic target.

Autoantibodies neutralizing Type I interferons increase the risk of severe viral diseases and are linked to autoimmune conditions. The automated VIDAS assay is suitable for anti‐IFN‐α2 IgGs quantification, offering a swift, reliable, user‐friendly, single test for clinical management. image

Background The intravoxel incoherent motion (IVIM) parameter estimation is affected by noise, while existing CNN‐based fitting methods utilize neighborhood spatial features around voxels to obtain more robust parameters. However, due to the heterogeneity of tissue, neighborhood features with low similarity can lead to excessively smooth parameter maps and even loss of tissue details. Purpose To propose a novel neural network fitting approach, IVIM‐CNNsimilar, which utilizes similar neighborhood information of voxels to assist in the estimation of IVIM parameters in diffusion‐weighted imaging (DWI). Methods The proposed fitting model is based on convolutional neural network (CNN), which first identifies the similar neighborhoods of voxels through cluster analysis and then uses CNN to learn the spatial features of similar neighborhoods to reduce the impact of noise on the parameter estimation of the voxel. To evaluate the performance of the proposed method, comparisons were conducted with the least squares (LSQ), Bayesian, PI‐DNN, and IVIM‐CNNunet algorithms on both simulated and in vivo brains, including 23 healthy brains and three brain tumors, in terms of root mean square error (RMSE) of IVIM parameters and the parameter contrast ratio between the tumor and normal regions. Results The CNN‐based methods, such as IVIM‐CNNsimilar and IVIM‐CNNunet, yield smoother parameter maps compared to voxel‐based methods like nonlinear least squares, segmented nonlinear least squares, Bayesian, and PI‐DNN. Additionally, the IVIM‐CNNsimilar retains more local tissue details while maintaining smoothness of parameter maps compared to the IVIM‐CNNunet. In simulated experiments, IVIM‐CNNsimilar outperforms IVIM‐CNNunet in terms of parameter estimation accuracy (SNR = 30; RMSE [DD] = 0.0168 vs. 0.0253; RMSE (FF) = 0.0001 vs. 0.0002; RMSE [D∗$D^{*}$] = 0.0266 vs. 0.0416). In addition, compared with other methods, the proposed IVIM‐CNNsimilar is more robust to noise, which is reflected in the lower RMSE of each parameter at different SNRs. For in vivo brains, compared to other methods, IVIM‐CNNsimilar achieved the highest PCR for most parameters when comparing the normal and tumor regions. Conclusions The IVIM‐CNNsimilar method uses similar neighborhood information to assist IVIM parameter fitting by reducing the impact of noise on voxel parameter estimation, thereby improving the accuracy of parameter estimation and increasing the potential for IVIM clinical application.

Background Fasting shows promise for public health, but concerns about muscle loss hinder its acceptance, particularly among the elderly. We explored the impact of long‐term fasting (12 days, 250 kcal/day) on muscle structure, metabolism and performance. Methods We prospectively assessed muscle volume, composition, relaxometry data and lipid metabolism in 32 subjects (16 men; 50% over 50 years old) before fasting, at the end of fasting and 1 month post‐fasting. Techniques included high‐resolution 3D Dixon MR imaging, multiecho CSE and single‐voxel MR spectroscopy. Dynamic ³¹P‐MRS, quantitative MRI, maximal voluntary contraction (MVC) measurements and exercise testing (VO2peak) were repeated throughout the protocol. Results Although the average body weight loss was 5.9 kg (7.4%, p < 0.001), the skeletal muscle volume change measured on the right calf muscle was 271 mL (5.4%, p < 0.001). This closely aligns with expected losses of glycogen (1%–2%) and bound water (3%–4%), estimated to total 404–505 mL. MVC (anaerobic lactic metabolism) remained preserved in both thighs and calf muscles, regardless of sex or age. Unchanged T2 showed that fasting did not induce structural or inflammatory changes. MRI/MRS revealed fat redistribution among tissues, with subcutaneous fat decrease (by 417.2 cm³, p < 0.01) and total fat fraction increase (by 0.2%, p < 0.05) in muscle. The intramyocellular lipid pool increased by 2.2 times (p < 0.05), whereas the extracellular lipid pool decreased to 1.4 times (p < 0.05), revealing rapid lipid trafficking and adaptation. During fasting, the T2* value increased by 1.2 ms (p < 0.001), likely because of changes in the configuration of intracellular lipid droplets, with an increased proportion of lipid droplets of smaller size, optimizing accessibility of lipid fuels and mitochondrial FA. Exercise testing (VO2peak) showed no change in maximal oxygen uptake, but fat oxidation improved with a 10% decrease in the exercise respiratory exchange ratio (p < 0.001). Mitochondrial oxidative capacity and PCr resynthesis rates in muscle were maintained. Females improved their mitochondrial function by D + 12, with τPCr decreasing to 29.61 s (p < 0.01), surpassing males and demonstrating better fat oxidation capabilities. Conclusions Long‐term fasting did not alter muscle metabolism or performance, nor induced structural or inflammatory changes. The decrease in muscle volume is minor when accounting for glycogen and water depletion during fasting. Fat is relocated to the intracellular compartment of myocytes. Both anaerobic and aerobic metabolic pathways remain unchanged after 12 days of fasting in both sexes and older subjects. This suggests that human muscles, like those in animals, have evolved to withstand seasonal food shortages and endure long fasting periods.

Background Emergency coronary angiogram after a cardiac arrest without ST-segment elevation myocardial infarction (STEMI) is still a matter of debate. To better select patients who may benefit from this procedure, we tested a visual coronary artery calcification (VCAC) score available in chest CT to predict significant coronary artery stenosis and/or culprit lesion or ad hoc or delayed percutaneous coronary intervention (PCI). Results A total of 113 patients with cardiac arrest and without STEMI who had a coronary angiogram and chest CT (January 2013 to March 2023, Croix-Rousse Hospital, Lyon, France) were retrospectively included. VCAC was scored from 0 (no calcification) to 3 (diffuse calcification) for each 4 four main arteries (left main, left anterior descending, circumflex, and right coronary artery). At baseline the median [interquartile range] age was 65.8 years [53.4–75.7], 61.9% were male, and 59.3% presented with ventricular fibrillation. Coronary angiogram identified at least one significant coronary artery stenosis in 32.7%, and ad hoc and delayed PCI were performed in 12.4% and 6.2% of the patients, respectively. VCAC score was an excellent predictor of significant coronary artery stenosis with an area under the ROC curve (AUC) of 0.95 (95%CI [0.90-1.00]) and the optimal threshold was ≥ 4 (specificity 94.7%, sensitivity 91.9%). For the detection of culprit coronary artery stenosis, the AUC was at 0.90 (95%CI [0.85–0.96]) and the optimal threshold was ≥ 5 (specificity 83.5%, sensitivity 87.5%). The AUC was 0.886 [0.823–0.948] (specificity 81.8%, sensitivity 85.7%) for ad hoc PCI and 0.921 [0.872–0.972] (specificity 85.3%, sensitivity 88.9%) for both delayed and ad hoc PCI with a same optimal threshold of VCAC ≥ 5. A VCAC score ≥ 4 had a sensitivity at 100% to predict a significant or culprit coronary artery stenosis and ad hoc or delayed PCI. Conclusions The present study found that a non-dedicated CT thorax may be useful to measure VCAC and if this is scored ≥ 4 it allows physicians to better select patients resuscitated from cardiac arrest with non-STEMI and without history of coronary artery disease who may benefit from an emergency coronary angiogram to detect a significant or culprit coronary artery stenosis and had PCI if appropriate.

The development of nanosystems with enhanced photothermal and photoacoustic properties is crucial for advancing theranostic applications in cancer therapy. This study explores polymeric nanoparticles constituted by a biocompatible poly(ethylene glycol)‐block‐poly(benzyl malate) copolymer and loaded with metal‐bis(dithiolene) complexes (M = Ni, Pd, Pt). These nanoparticles, prepared via a robust nanoprecipitation method, demonstrate uniform morphology, efficient encapsulation (~70%), and tailored near‐infrared (NIR) optical absorption properties. Photothermal and photoacoustic evaluations revealed superior performance of Palladium‐loaded nanoparticles, offering high contrast for imaging and significant temperature increases under NIR laser irradiation. Cytotoxicity assays confirmed their non‐toxicity without laser exposure, while effective cancer cell eradication was achieved upon irradiation at power densities ≥2 W/cm2. In vivo experiments on zebrafish embryos bearing human cancer xenografts showed significant tumor size reduction (20%) post‐treatment with Palladium‐loaded nanoparticles under 880 nm laser irradiation. These findings underscore that metal‐bis(dithiolene)‐loaded nanoparticles can be versatile agents for combined diagnostics and photothermal therapy, paving the way for further optimization and clinical translation.

Background Assessing vitamin A (VA) status using retinol and retinol-binding protein (RBP) in the presence of infection and inflammation remains challenging, as both markers prove to be unreliable during such physiological disturbances. Objective This study aimed to assess the association between common infections and inflammation and VA status of children in rural Burkina Faso. Methods Two community-based cross-sectional studies were conducted in the villages of Sourkoudougou and Banakeledaga, in Southwestern Burkina Faso, one during the dry season (November 2016– January 2017) and the second during the rainy season (August– September 2017). In total, 115 children, 36–59 months of age, were included. The ¹³C-retinol isotope dilution test (RID) was used to assess total body VA stores (TBS) and VA total liver reserves (TLR). Malaria infection and intestinal parasites were evaluated at enrollment. Serum C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP) were measured. Univariable and multivariable linear regressions were used to test the associations between VA status and infection and inflammation status. Results No VA deficiency (TLR ≤ 0.1 µmol/g liver) was detected using RID method. Geometric means (95% confidence interval) of TBS and TLR were respectively 473 (412; 543) µmol and 0.86 (0.75; 0.99) µmol/g liver. One-third of study participants were found to have hypervitaminosis A (TLR > 1.0 µmol/g liver). Elevated CRP (> 5.0 mg/L) and AGP (> 1.0 g/L) were respectively detected in 1.9% and 28.6% of children. Positive malaria was diagnosed in 5 children. Intestinal parasites were found in one out of two (47.6%) participants, and other morbidities were detected in 2 participants. In a multivariable adjusted regression, significant positive weak associations were found between Log TLR and CRP concentrations (N = 79, β = 0.058, p = 0.004) and between Log TBS and CRP concentrations (N = 79, β = 0.054, p = 0.005). Conclusion Study children were apparently healthy with high prevalence of asymptomatic intestinal parasites and chronic inflammation. TLR and TBS were positively associated with the acute phase protein CRP warranting further investigation. Trial registration The study was registered retrospectively (22 March 2018) as a clinical trial with the Pan African Clinical Trials Registry (Cochrane South Africa; PACTR201803002999356).

Low signal to noise ratio (SNR) remains one of the limitations of diffusion weighted (DW) imaging. How to suppress the influence of noise on the subsequent analysis about the tissue microstructure is still challenging. This work proposed a novel self-supervised learning model, Replace2Self, to effectively reduce spatial correlated noise in DW images. Specifically, a voxel replacement strategy based on similar block matching in Q-space was proposed to destroy the correlations of noise in DW image along one diffusion gradient direction. To alleviate the signal gap caused by the voxel replacement, an image mixing strategy based on complementary mask was designed to generate two different noisy DW images. After that, these two noisy DW images were taken as input, and the non-correlated noisy DWimage after voxel replacement was taken as learning target, a denoising network was trained for denoising. To promote the denoising performance, a complementary mask mixing consistency loss and an inverse replacement regularization loss were also proposed. Through the comparisons against several existing DW image denoising methods on extensive simulation data with different noise distributions, noise levels and b-values, as well as the acquisition datasets and the ablation experiments, we verified the effectiveness of the proposed method. Regardless of the noise distribution and noise level, the proposed method achieved the highest PSNR, which was at least 1.9% higher than the suboptimal method when the noise level reaches 10%. Furthermore, our method has superior generalization ability due to the use of the proposed strategies.

Nucleus accurate segmentation is a crucial task in biomedical image analysis. While convolutional neural networks (CNNs) have achieved notable progress in this field, challenges remain due to the complexity and heterogeneity of cell images, especially in overlapping regions of nuclei. To address the limitations of current methods, we propose a mechanism of multiple differential convolution and local-variation attention in CNNs, leading to the so-called multiple differential convolution and local-variation attention U-Net (MDLA-UNet). The multiple differential convolution employs multiple differential operators to capture gradient and direction information, improving the network’s capability to detect edges. The local-variation attention utilizes Haar discrete wavelet transforms for level-1 decomposition to obtain approximate features, and then derives high-frequency features to enhance the global context and local detail variation of the feature maps. The results on the MoNuSeg, TNBC, and CryoNuSeg datasets demonstrated superior segmentation performance of the proposed method for cells having complex boundaries and details with respect to existing methods. The proposed MDLA-UNet presents the ability of capturing fine edges and details in feature maps and thus improves the segmentation of nuclei with blurred boundaries and overlapping regions.