Search

Zusammenfassung Hintergrund Im Rahmen der gesundheitsökonomischen Analyse wurde in einem Piggyback-Ansatz die Kosteneffektivität des am Point of Care Apotheke erbrachten Präventionsprogrammes der GLICEMIA 2.0-Studie evaluiert, das versuchte, die Teilnehmenden in der Interventionsgruppe zu einer verbesserten glykämischen Kontrolle bei Typ-2-Diabetes mit nachhaltiger Incentivierung von gesundheits-bewusstem Verhalten, therapeutischer Compliance und Adhärenz heranzuführen. Die Kontrollgruppe erhielt ein passives Medikationsmanagement und ein Monitoring zum Diabetes-Status. Methode Primärer Endpunkt der GLICEMIA 2.0-Studie war die Stabilisierung des HbA1c-Wertes. Zur gesundheitsökonomischen Auswertung wurden inkrementale Unterschiede in Output-Veränderungen untersucht, definiert als Differenz der Häufigkeitsverteilung der HbA1c-Werte zwischen den Gruppen im zeitlichen Verlauf. Als Kostenparameter wurden direkte Programmkosten und antizipierte indirekte Kosten der Inanspruchnahme ärztlicher Leistungen erfasst. Zur Validierung von Kosteneffektivitätsschwellen wurde über die Bildung von ICER-Werten ein Net-Monetary-Benefit-Ansatz herangezogen. Ergebnis Die Interventionsgruppe erreichte signifikant höhere Verbesserungen der HbA1c-Werte. Durch die vorgenommene Risikostratifizierung über den Anfangsmedian der HbA1c-Werte können Kosteneffektivitätspotenziale bei hohen HbA1c-Ausgangswertengezeigt werden. Aufgrund des Untersuchungszeitraums sind keine langfristigen Unterschiede in Inanspruchnahmen ärztlicher Leistungen ersichtlich. Schlussfolgerung Das GLICEMIA 2.0-Programm indiziert deutliche Effektivitätspotenziale besonders bei höheren Risikograden. Der Wirkeffekt scheint die Adhärenz der Interventions- im Vergleich zur Kontrollgruppe befördert zu haben. Eine Ermittlung der unmittelbaren Wirkungshebel der Gruppenintervention würde aufgrund des Charakters einer komplexen Intervention einen längeren Zeitraum benötigen. Aufgrund eines fehlenden Follow-up können über längerfristige Effekte nur eingeschränkte Aussagen getroffen werden.

Tuning of model-based boosting algorithms relies mainly on the number of iterations, while the step-length is fixed at a predefined value. For complex models with several predictors such as Generalized additive models for location, scale and shape (GAMLSS), imbalanced updates of predictors, where some distribution parameters are updated more frequently than others, can be a problem that prevents some submodels to be appropriately fitted within a limited number of boosting iterations. We propose an approach using adaptive step-length (ASL) determination within a non-cyclical boosting algorithm for Gaussian location and scale models, as an important special case of the wider class of GAMLSS, to prevent such imbalance. Moreover, we discuss properties of the ASL and derive a semi-analytical form of the ASL that avoids manual selection of the search interval and numerical optimization to find the optimal step-length, and consequently improves computational efficiency. We show competitive behavior of the proposed approaches compared to penalized maximum likelihood and boosting with a fixed step-length for Gaussian location and scale models in two simulations and two applications, in particular for cases of large variance and/or more variables than observations. In addition, the underlying concept of the ASL is also applicable to the whole GAMLSS framework and to other models with more than one predictor like zero-inflated count models, and brings up insights into the choice of the reasonable defaults for the step-length in the simpler special case of (Gaussian) additive models.

(1) Background: Since 2013, weekly screening for multidrug-resistant Gram-negative (MDRGN) bacteria has been performed in German neonatal intensive care units (NICU). National guidelines recommend considering these colonization analyses for antibiotic treatment regimens. Our retrospective single center study provides insight into the clinical dichotomy of bacterial colonization and infection rates in neonates. (2) Methods: We analyzed microbiological data of neonates admitted to our tertiary level NICU over nine years. Colonization with MDRGN/Serratia marcescens (SERMA) was compared to microbiological findings in sepsis and pneumonia. (3) Results: We analyzed 917 blood and 1799 tracheal aspirate samples. After applying criteria from the Nosocomial Infection Surveillance for Neonates (NEO-KISS), we included 52 and 55 cases of sepsis and pneumonia, respectively; 19.2% of sepsis patients and 34.5% of pneumonia patients had a prior colonization with MDRGN bacteria or SERMA. In these patients, sepsis was not attributable to MDRGN bacteria yet one SERMA, while in pneumonias, ten MDRGN bacteria and one SERMA were identified. We identified late-onset pneumonia and cesarean section as risk factors for MDRGN/SERMA acquisition. (4) Conclusions: Colonization screening is a useful tool for hygiene surveillance. However, our data suggest that consideration of colonization with MDRGN/SERMA might promote extensive use of last resort antibiotics in neonates.

Mixture Density Networks (MDN) belong to a class of models that can be applied to data which cannot be sufficiently described by a single distribution since it originates from different components of the main unit and therefore needs to be described by a mixture of densities. In some situations, however, MDNs seem to have problems with the proper identification of the latent components. While these identification issues can to some extent be contained by using custom initialization strategies for the network weights, this solution is still less than ideal since it involves subjective opinions. We therefore suggest replacing the hidden layers between the model input and the output parameter vector of MDNs and estimating the respective distributional parameters with penalized cubic regression splines. Applying this approach to data from Gaussian mixture distributions as well gamma mixture distributions proved to be successful with the identification issues not playing a role anymore and the splines reliably converging to the true parameter values.

Zusammenfassung

A piggyback approach was used to evaluate the cost-effectiveness of the prevention program delivered at the point of care pharmacy in the GLICEMIA 2.0 study that sought to guide participants in the intervention group to improved glycemic control in type 2 diabetes with sustained incentivization of lifestyle changes, therapeutic compliance, and adherence. The control group received passive medication management and monitoring.Primary endpoint of the GLICEMIA 2.0 study was the stabilization of HbA1c values. For health economic evaluation, incremental differences in output changes were examined, defined as the difference in the distribution of the HbA1c values between both groups over time. Direct program costs and anticipated indirect costs of service utilization were used as cost parameters. A net monetary benefit approach was used to validate cost-effectiveness thresholds via the formation of ICER values.The intervention group had significantly higher reductions in HbAThe GLICEMIA program indicates considerable effectiveness potentials, especially for high-risk patients. The study design seems to have assisted the intervention group's adherence in contrast to the control group. Detailed impacts within the complex intervention could not be validated due to restrictions of the study design as a complex intervention. Overall, statements about effect sustainability and further utilization rates progressions are limited due to a lack of follow-up.Im Rahmen der gesundheitsökonomischen Analyse wurde in einem Piggyback-Ansatz die Kosteneffektivität des am Point of Care Apotheke erbrachten Präventionsprogrammes der GLICEMIA 2.0-Studie evaluiert, das versuchte, die Teilnehmenden in der Interventionsgruppe zu einer verbesserten glykämischen Kontrolle bei Typ-2-Diabetes mit nachhaltiger Incentivierung von gesundheits-bewusstem Verhalten, therapeutischer Compliance und Adhärenz heranzuführen. Die Kontrollgruppe erhielt ein passives Medikationsmanagement und ein Monitoring zum Diabetes-Status.Primärer Endpunkt der GLICEMIA 2.0-Studie war die Stabilisierung des HbADie Interventionsgruppe erreichte signifikant höhere Verbesserungen der HbADas GLICEMIA 2.0-Programm indiziert deutliche Effektivitätspotenziale besonders bei höheren Risikograden. Der Wirkeffekt scheint die Adhärenz der Interventions- im Vergleich zur Kontrollgruppe befördert zu haben. Eine Ermittlung der unmittelbaren Wirkungshebel der Gruppenintervention würde aufgrund des Charakters einer komplexen Intervention einen längeren Zeitraum benötigen. Aufgrund eines fehlenden Follow-up können über längerfristige Effekte nur eingeschränkte Aussagen getroffen werden.

Besides tremendous treatment success in advanced melanoma patients, the rapid development of oncologic treatment options comes with increasingly high costs and can cause severe life-threatening side effects. For this purpose, predictive baseline biomarkers are becoming increasingly important for risk stratification and personalized treatment planning. Thus, the aim of this pilot study was the development of a prognostic tool for the risk stratification of the treatment response and mortality based on PET/MRI and PET/CT, including a convolutional neural network (CNN) for metastasized-melanoma patients before systemic-treatment initiation. The evaluation was based on 37 patients (19 f, 62 ± 13 y/o) with unresectable metastasized melanomas who underwent whole-body 18F-FDG PET/MRI and PET/CT scans on the same day before the initiation of therapy with checkpoint inhibitors and/or BRAF/MEK inhibitors. The overall survival (OS), therapy response, metastatically involved organs, number of lesions, total lesion glycolysis, total metabolic tumor volume (TMTV), peak standardized uptake value (SULpeak), diameter (Dmlesion) and mean apparent diffusion coefficient (ADCmean) were assessed. For each marker, a Kaplan–Meier analysis and the statistical significance (Wilcoxon test, paired t-test and Bonferroni correction) were assessed. Patients were divided into high- and low-risk groups depending on the OS and treatment response. The CNN segmentation and prediction utilized multimodality imaging data for a complementary in-depth risk analysis per patient. The following parameters correlated with longer OS: a TMTV < 50 mL; no metastases in the brain, bone, liver, spleen or pleura; ≤4 affected organ regions; no metastases; a Dmlesion > 37 mm or SULpeak < 1.3; a range of the ADCmean < 600 mm2/s. However, none of the parameters correlated significantly with the stratification of the patients into the high- or low-risk groups. For the CNN, the sensitivity, specificity, PPV and accuracy were 92%, 96%, 92% and 95%, respectively. Imaging biomarkers such as the metastatic involvement of specific organs, a high tumor burden, the presence of at least one large lesion or a high range of intermetastatic diffusivity were negative predictors for the OS, but the identification of high-risk patients was not feasible with the handcrafted parameters. In contrast, the proposed CNN supplied risk stratification with high specificity and sensitivity.

The aim of this study was to assess the diagnostic value of ADC distribution curves for differentiation between benign and malignant parotid gland tumors and to compare with mean ADC values. 73 patients with parotid gland tumors underwent head-and-neck MRI on a 1.5 Tesla scanner prior to surgery and histograms of ADC values were extracted. Histopathological results served as a reference standard for further analysis. ADC histograms were evaluated by comparing their similarity to a reference distribution using Chi2-test-statistics. The assumed reference distribution for benign and malignant parotid gland lesions was calculated after pooling the entire ADC data. In addition, mean ADC values were determined. For both methods, we calculated and compared the sensitivity and specificity between benign and malignant parotid gland tumors and three subgroups (pleomorphic adenoma, Warthin tumor, and malignant lesions), respectively. Moreover, we performed cross-validation (CV) techniques to estimate the predictive performance between ADC distributions and mean values. Histopathological results revealed 30 pleomorphic adenomas, 22 Warthin tumors, and 21 malignant tumors. ADC histogram distribution yielded a better specificity for detection of benign parotid gland lesions (ADChistogram: 75.0% vs. ADCmean: 71.2%), but mean ADC values provided a higher sensitivity (ADCmean: 71.4% vs. ADChistogram: 61.9%). The discrepancies are most pronounced in the differentiation between malignant and Warthin tumors (sensitivity ADCmean: 76.2% vs. ADChistogram: 61.9%; specificity ADChistogram: 81.8% vs. ADCmean: 68.2%). Using CV techniques, ADC distribution revealed consistently better accuracy to differentiate benign from malignant lesions (“leave-one-out CV” accuracy ADChistogram: 71.2% vs. ADCmean: 67.1%). ADC histogram analysis using full distribution curves is a promising new approach for differentiation between primary benign and malignant parotid gland tumors, especially with respect to the advantage in predictive performance based on CV techniques.

We describe a publicly available dataset of annotated Positron Emission Tomography/Computed Tomography (PET/CT) studies. 1014 whole body Fluorodeoxyglucose (FDG)-PET/CT datasets (501 studies of patients with malignant lymphoma, melanoma and non small cell lung cancer (NSCLC) and 513 studies without PET-positive malignant lesions (negative controls)) acquired between 2014 and 2018 were included. All examinations were acquired on a single, state-of-the-art PET/CT scanner. The imaging protocol consisted of a whole-body FDG-PET acquisition and a corresponding diagnostic CT scan. All FDG-avid lesions identified as malignant based on the clinical PET/CT report were manually segmented on PET images in a slice-per-slice (3D) manner. We provide the anonymized original DICOM files of all studies as well as the corresponding DICOM segmentation masks. In addition, we provide scripts for image processing and conversion to different file formats (NIfTI, mha, hdf5). Primary diagnosis, age and sex are provided as non-imaging information. We demonstrate how this dataset can be used for deep learning-based automated analysis of PET/CT data and provide the trained deep learning model.

Posttraumatic stress disorder (PTSD) is a severe mental disorder that can develop after a traumatic event. PTSD has been reported to be associated with activation of the innate immune system, as measured by increased levels of pro-inflammatory cytokines. While it is well known that PTSD patients display increased levels of interleukin 6 (IL-6) when compared with healthy controls, the relationship between cytokine secretion and treatment outcome has been hardly investigated yet. The aim of this study was to assess the potential association of inflammatory activation and therapy outcome in PTSD. Before therapeutic intervention, we applied the Trier Social Stress Test (TSST) as a method to elicit psychosocial stress and an acute inflammatory response. IL-6 levels were measured in blood plasma of PTSD patients at different time points before and after the TSST. Severity of depressive, trauma-related, and somatic symptoms was assessed before and 8 weeks after trauma-focused treatment in a multimodal day clinic setting. We showed that high reactivity of IL-6 to psychosocial stress at the beginning of the therapy was associated with a negative therapy outcome in PTSD, especially regarding depressive symptoms. This study suggests plasma IL-6 reactivity as a potential molecular marker to predict treatment outcome in PTSD.

Background: Due to the establishment of a nationwide certification system for cancer centers in Germany, the availability of psycho-oncological services for cancer patients has increased substantially. However, little is known about the specific intervention techniques that are applied during sessions in an acute care hospital, since a standardized taxonomy is lacking. With this study, we aimed at the investigation of psycho-oncological intervention techniques and the development of a comprehensive and structured taxonomy thereof. Methods: In a stepwise procedure, a team of psycho-oncologists generated a data pool of interventions and definitions that were tested in clinical practice during a pilot phase. After an adaptation of intervention techniques, interrater reliability (IRR) was attained by rating 10 previously recorded psycho-oncological sessions. A classification of interventions into superordinate categories was performed, supported by cluster analysis. Results: Between April and June 2017, 980 psycho-oncological sessions took place. The experts agreed on a total number of 22 intervention techniques. An IRR of 89% for 2 independent psycho-oncological raters was reached. The 22 techniques were classified into 5 superordinate categories. Discussion/Conclusion: We developed a comprehensive and structured taxonomy of psycho-oncological intervention techniques in an acute care hospital that provides a standardized basis for systematic research and applied care. We expect our work to be continuously subjected to further development: future research should evaluate and expand our taxonomy to other contexts and care settings.

Differences in brain signal intensity ratios (SIRs) of deep brain nuclei in T1-weighted (T1w) magnetic resonance images were reported as an indicator of gadolinium brain retention. Variable methods of image reconstruction and inhomogeneity correction for T1w images exist, which might affect the accuracy of SIRs. The aim of our prospective study was to investigate the effect of flow artifact compensation (FAC) and intensity inhomogeneity correction (IIC) on the dentate nucleus-to-pons and globus pallidus-to-thalamus SIRs in study participants who had previously received multiple doses of gadobutrol.This study included 76 participants who received five or more gadobutrol-enhanced scans between 2007 and 2017. A control group of 25 participants without gadolinium-based contrast agent application in their patient history was included for comparison. Unenhanced brain magnetic resonance imaging including two T1w spin-echo sequences with and without FAC was performed in all participants. Both sequences were reconstructed with and without IIC. Images were assessed for flow artifacts and SIRs were calculated.Using FAC, a lower proportion of participants had to be excluded from the final analysis of dentate nucleus-to-pons SIR due to flow artifacts (15% versus 46%, p0.001). Without IIC, a difference was found between the study and the control group for the dentate nucleus-to-pons ratio (p = 0.004), but not for the same sequence reconstructed with IIC (p = 0.29). For the globus pallidus-to-thalamus ratio, no difference was found between the study and control group.The application of an IIC algorithm has significant impact on brain nuclei SIRs for the assessment of gadolinium brain retention.

Background Attenuation correction (AC) of PET data is usually performed using a second imaging for the generation of attenuation maps. In certain situations however—when CT- or MR-derived attenuation maps are corrupted or CT acquisition solely for the purpose of AC shall be avoided—it would be of value to have the possibility of obtaining attenuation maps only based on PET information. The purpose of this study was to thus develop, implement, and evaluate a deep learning-based method for whole body [18F]FDG-PET AC which is independent of other imaging modalities for acquiring the attenuation map. Methods The proposed method is investigated on whole body [18F]FDG-PET data using a Generative Adversarial Networks (GAN) deep learning framework. It is trained to generate pseudo CT images (CTGAN) based on paired training data of non-attenuation corrected PET data (PETNAC) and corresponding CT data. Generated pseudo CTs are then used for subsequent PET AC. One hundred data sets of whole body PETNAC and corresponding CT were used for training. Twenty-five PET/CT examinations were used as test data sets (not included in training). On these test data sets, AC of PET was performed using the acquired CT as well as CTGAN resulting in the corresponding PET data sets PETAC and PETGAN. CTGAN and PETGAN were evaluated qualitatively by visual inspection and by visual analysis of color-coded difference maps. Quantitative analysis was performed by comparison of organ and lesion SUVs between PETAC and PETGAN. Results Qualitative analysis revealed no major SUV deviations on PETGAN for most anatomic regions; visually detectable deviations were mainly observed along the diaphragm and the lung border. Quantitative analysis revealed mean percent deviations of SUVs on PETGAN of − 0.8 ± 8.6% over all organs (range [− 30.7%, + 27.1%]). Mean lesion SUVs showed a mean deviation of 0.9 ± 9.2% (range [− 19.6%, + 29.2%]). Conclusion Independent AC of whole body [18F]FDG-PET is feasible using the proposed deep learning approach yielding satisfactory PET quantification accuracy. Further clinical validation is necessary prior to implementation in clinical routine applications.

Purpose To assess the role of CT-texture analysis (CTTA) for differentiation of pancreatic ductal adenocarcinoma (PDAC) from pancreatic neuroendocrine neoplasm (PNEN) in the portal-venous phase as compared with visual assessment and tumor-to-pancreas attenuation ratios. Methods 53 patients (66.1 ± 8.6y) with PDAC and 42 patients (65.5 ± 12.2y) with PNEN who underwent contrast-enhanced CT for primary staging were evaluated. Volumes of interests (VOIs) were set in the tumor tissue at the portal-venous phase excluding adjacent structures. Based on pyradiomics library, 92 textural features were extracted including 1st, 2nd, and higher order features, and then compared between PNEN and PDAC. The visual assessment classified tumors into hypo-, iso-, or hyperdense to pancreas parenchyma or into homogeneous/heterogeneous. Additionally, attenuation ratios between the tumors and the non-involved pancreas were calculated. Results 8/92 (8.6%) highly significant (p p Conclusions Our data indicate that CTTA is a feasible tool for differentiation of PNEN from PDAC and also of G1 from G2/3 PNEN in the portal-venous phase. Visual assessment and tumor-to-parenchyma ratios were not useful for discrimination.

Machine learning (ML) is considered an important technology for future data analysis in health care.The inherently technology-driven fields of diagnostic radiology and nuclear medicine will both benefit from ML in terms of image acquisition and reconstruction. Within the next few years, this will lead to accelerated image acquisition, improved image quality, a reduction of motion artifacts and - for PET imaging - reduced radiation exposure and new approaches for attenuation correction. Furthermore, ML has the potential to support decision making by a combined analysis of data derived from different modalities, especially in oncology. In this context, we see great potential for ML in multiparametric hybrid imaging and the development of imaging biomarkers.In this review, we will describe the basics of ML, present approaches in hybrid imaging of MRI, CT, and PET, and discuss the specific challenges associated with it and the steps ahead to make ML a diagnostic and clinical tool in the future.· ML provides a viable clinical solution for the reconstruction, processing, and analysis of hybrid imaging obtained from MRI, CT, and PET..· Küstner T, Hepp T, Seith F. Multiparametric Oncologic Hybrid Imaging: Machine Learning Challenges and Opportunities. Fortschr Röntgenstr 2022; 194: 605 - 612.Maschinelles Lernen (ML) gilt als eine wichtige Technologie für die zukünftige Datenanalyse im Gesundheitswesen.Die inhärent technologiegetriebene diagnostische Radiologie und Nuklearmedizin werden sowohl bei der Bildaufnahme als auch bei der Bildrekonstruktion von ML profitieren. In den nächsten Jahren wird dies zu einer beschleunigten Bildaufnahme, einer verbesserten Bildqualität, einer Reduzierung von Bewegungsartefakten und – für die PET-Bildgebung – zu einer reduzierten Strahlenexposition und neuen Ansätzen zur Schwächungskorrektur führen. Darüber hinaus hat ML das Potenzial, die Entscheidungsfindung durch eine kombinierte Analyse von Daten aus verschiedenen Modalitäten, insbesondere im Bereich der Onkologie, zu unterstützen. In diesem Zusammenhang sehen wir ein großes Potenzial für ML in der multiparametrischen Hybrid-Bildgebung und der Entwicklung von bildgebenden Biomarkern.In diesem Review werden wir die Grundlagen von ML beschreiben, Ansätze in der hybriden Bildgebung von MRT, CT und PET vorstellen und die damit verbundenen spezifischen Herausforderungen und die kommenden Schritte diskutieren, um ML in Zukunft zu einem diagnostischen und klinischen Werkzeug zu machen.· ML bietet eine praktikable klinische Lösung für die Rekonstruktion, Verarbeitung und Analyse von Hybrid-Bildgebung der MRT, CT und PET..

Background Reference intervals represent the expected range of physiological test results in a healthy population and are essential to support medical decision making. Particularly in the context of pediatric reference intervals, where recruitment regulations make prospective studies challenging to conduct, indirect estimation strategies are becoming increasingly important. Established indirect methods enable robust identification of the distribution of “healthy” samples from laboratory databases, which include unlabeled pathologic cases, but are currently severely limited when adjusting for essential patient characteristics such as age. Here, we propose the use of mixture density networks (MDN) to overcome this problem and model all parameters of the mixture distribution in a single step. Results Estimated reference intervals from varying settings with simulated data demonstrate the ability to accurately estimate latent distributions from unlabeled data using different implementations of MDNs. Comparing the performance with alternative estimation approaches further highlights the importance of modeling the mixture component weights as a function of the input in order to avoid biased estimates for all other parameters and the resulting reference intervals. We also provide a strategy to generate partially customized starting weights to improve proper identification of the latent components. Finally, the application on real-world hemoglobin samples provides results in line with current gold standard approaches, but also suggests further investigations with respect to adequate regularization strategies in order to prevent overfitting the data. Conclusions Mixture density networks provide a promising approach capable of extracting the distribution of healthy samples from unlabeled laboratory databases while simultaneously and explicitly estimating all parameters and component weights as non-linear functions of the covariate(s), thereby allowing the estimation of age-dependent reference intervals in a single step. Further studies on model regularization and asymmetric component distributions are warranted to consolidate our findings and expand the scope of applications.

Purpose: We introduce and evaluate deep learning methods for weakly supervised segmentation of tumor lesions in whole-body fluorodeoxyglucose-positron emission tomography (FDG-PET) based solely on binary global labels ("tumor" versus "no tumor"). Approach: We propose a three-step approach based on (i) a deep learning framework for image classification, (ii) subsequent generation of class activation maps (CAMs) using different CAM methods (CAM, GradCAM, GradCAM++, ScoreCAM), and (iii) final tumor segmentation based on the aforementioned CAMs. A VGG-based classification neural network was trained to distinguish between PET image slices with and without FDG-avid tumor lesions. Subsequently, the CAMs of this network were used to identify the tumor regions within images. This proposed framework was applied to FDG-PET/CT data of 453 oncological patients with available manually generated ground-truth segmentations. Quantitative segmentation performance was assessed for the different CAM approaches and compared with the manual ground truth segmentation and with supervised segmentation methods. In addition, further biomarkers (MTV and TLG) were extracted from the segmentation masks. Results: A weakly supervised segmentation of tumor lesions was feasible with satisfactory performance [best median Dice score 0.47, interquartile range (IQR) 0.35] compared with a fully supervised U-Net model (median Dice score 0.72, IQR 0.36) and a simple threshold based segmentation (Dice score 0.29, IQR 0.28). CAM, GradCAM++, and ScoreCAM yielded similar results. However, GradCAM led to inferior results (median Dice score: 0.12, IQR 0.21) and was likely to ignore multiple instances within a given slice. CAM, GradCAM++, and ScoreCAM yielded accurate estimates of metabolic tumor volume (MTV) and tumor lesion glycolysis. Again, worse results were observed for GradCAM. Conclusions: This work demonstrated the feasibility of weakly supervised segmentation of tumor lesions and accurate estimation of derived metrics such as MTV and tumor lesion glycolysis.

Purpose Purpose of this study is to evaluate the diagnostic accuracy of quantitative T2/ADC values in differentiating between PCa and lesions showing non-specific inflammatory infiltrates and atrophy, features of chronic prostatitis, as the most common histologically proven differential diagnosis. Methods In this retrospective, single-center cohort study, we analyzed 55 patients suspected of PCa, who underwent mpMRI (3T) including quantitative T2 maps before robot-assisted mpMRI-TRUS fusion prostate biopsy. All prostate lesions were scored according to PI-RADS v2.1. Regions of interest (ROIs) were annotated in focal lesions and normal prostate tissue. Quantitative mpMRI values from T2 mapping and ADC were compared using two-tailed t tests. Receiver operating characteristic curves (ROCs) and cutoff were calculated to differentiate between PCa and chronic prostatitis. Results Focal lesions showed significantly lower ADC and T2 mapping values than normal prostate tissue (p p r = 0.70; p Conclusion T2 mapping showed high diagnostic accuracy for differentiating between PCa and chronic prostatitis, comparable to the performance of ADC values.

Purpose Prostate biparametric magnetic resonance imaging (bpMRI) including T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI) might be an alternative to multiparametric MRI (mpMRI, including dynamic contrast imaging, DCE) to detect and guide targeted biopsy in patients with suspected prostate cancer (PCa). However, there is no upgrading peripheral zone PI-RADS 3 to PI-RADS 4 without DCE in bpMRI. The aim of this study was to evaluate bpMRI against mpMRI in biopsy-naïve men with elevated prostate-specific antigen (PSA) scheduled for robot-assisted-transperineal fusion-prostate biopsy (RA-TB). Methods Retrospective single-center-study of 563 biopsy-naïve men (from 01/2015 to 09/2018, mean PSA 9.7 ± 6.5 ng/mL) with PI-RADSv2.1 conform mpMRI at 3 T before RA-TB. Clinically significant prostate cancer (csPCa) was defined as ISUP grade ≥ 2 in any core. Two experienced readers independently evaluated images according to PI-RADSv2.1 criteria (separate readings for bpMRI and mpMRI sequences, 6-month interval). Reference standard was histology from RA-TB. Results PI-RADS 2 was scored in 5.1% of cases (3.4% cancer/3.4% csPCa), PI-RADS 3 in 16.9% (32.6%/3.2%), PI-RADS 4 in 57.6% (66.1%/58.3%) and PI-RADS 5 in 20.4% of cases (79.1%/74.8%). For mpMRI/bpMRI test comparison, sensitivity was 99.0%/97.1% (p p p Conclusion In patients planned for RA-TB with elevated PSA and clinical suspicion for PCa, specificity was higher in bpMRI vs. mpMRI, which could solve constrains regarding time and contrast agent.

Background: A transjugular intrahepatic portosystemic shunt (TIPS) is a common treatment for patients with portal hypertension. In these patients, the portal vein can be punctured under the guidance of cone-beam computed tomography (CBCT). Objectives: To compare standard-dose (SD) CBCT with low-dose (LD) CBCT, as three-dimensional (3D) intraprocedural guidance for transhepatic puncture in TIPS placement, in terms of image quality, radiation dose, technical success, and complications. Patients and Methods: A total of 44 patients were retrospectively enrolled in this study. Eighteen patients underwent LD-CBCT, while 26 patients underwent SD-CBCT for guiding the portal vein puncture. A quantitative assessment of image quality was performed by calculating the contrast-to-noise ratio (CNR) of the hepatic portal vein. This analysis was based on a five-point vascular visualization scale (VVS), ranging from optimal (score = 1) to non-diagnostic (score = 5), while a three-point Likert scale was used for motion artifacts (1 = no motion artifacts, 3 = blurred). Image streak artifacts were also rated from one to three, based on the image quality results. Technical success was also investigated, including the number of puncture attempts, time to successful portal vein access, and radiation dose of the TIPS procedure. Results: Based on the results, TIPS could be placed successfully in all cases. Neither VVS (LD-CBCT VVS: 2.78, SD-CBCT VVS: 2.54; P = 0.467), nor the procedure time showed any significant differences between the groups (LD-CBCT: 48.3 min, SD-CBCT: 40.2 min; P = 0.45). Moreover, the objective evaluation of image quality indicated the lower quality of LD-CBCT images; however, the difference was not statistically significant (LD-CBCT CNR: 1.1 ± 0.76, SD-CBCT CNR: 1.3 ± 1.1; P = 0.5). The median number of puncture attempts was the same for SD-CBCT and LD-CBCT (n = 3; range: 1 - 6). Also, the mean dose area product (DAP) was significantly lower in LD-CBCT as compared to SD-CBCT (LD-CBCT: 2733 ± 848 µGm2, SD-CBCT: 6119 ± 1677 µGm2; P < 0.0001). The total DAP was significantly lower using LD-CBCT (LD-CBCT: 14831 ± 9299 µGm2, SD-CBCT: 20985 ± 10127 µGm2; P = 0.047). Conclusion: Both SD-CBCT and LD-CBCT provided successful 3D guidance for portal vein puncture during TIPS creation. Although these methods did not differ significantly in terms of image quality, complications, or number of puncture attempts, LD-CBCT significantly reduced the radiation dose.

Background Patients with hepatic metastatic uveal melanoma still have a poor outcome. Purpose To evaluate overall survival (OS), progression-free survival (PFS), and response predictors in these patients treated with chemosaturation by percutaneous hepatic perfusion with melphalan (CS-PHP). Material and Methods Between June 2015 and March 2020, a total of 29 patients (median age 69.7 years; age range 30–81 years; 60% women; median BMI 25.7 kg/m2; range 18.7–35.3kg/m2; 1–6 procedures per patient) were treated with 53 CS-PHPs. All patients received cross-sectional imaging for initial and follow-up examinations. Baseline tumor load, extrahepatic tumor load, tumor response, PFS, and OS were assessed. Non-parametric statistics were used. Results After the initial CS-PHP, a partial response was observed in 11 patients (41%), stable disease in 12 patients (44%) and progressive disease in 4 patients (15%); two patients died before the response was evaluated. After initial CS-PHP, median OS was 12.9 ± 7.4 months and median PFS was 7.1 ± 7.4 months. OS after one year was 50%. After the second CS-PHP, median PFS was 7.9 ± 5.7 months. Seven patients had a liver tumor burden >25%, associated with a significantly shorter OS (6.0 ± 2.4 vs. 14.1 ± 12.7 months; P = 0.008). At the time of first CS-PHP, 41% (12/29) of the patients had extrahepatic metastases that did not affect OS (11.1 ± 8.4 months vs. 12.9 ± 13.6 months; P = 0.66). Conclusion CS-PHP is a safe and effective treatment for the hepatic metastatic uveal melanoma, especially for patients with a hepatic tumor burden

Major depressive disorder (MDD) is a severe psychiatric condition with key symptoms of low mood and lack of motivation, joy, and pleasure. Recently, the acid sphingomyelinase (ASM)/ceramide system has been implicated in the pathogenesis of MDD. ASM is a lysosomal glycoprotein that catalyzes the hydrolysis of sphingomyelin, an abundant component of membranes, into the bioactive sphingolipid ceramide, which impacts signaling pathways. ASM activity is inhibited by several common antidepressant drugs. Human and murine studies have confirmed that increased ASM activity and ceramide levels are correlated with MDD. To define a molecular marker for treatment monitoring, we investigated the mRNA expression of SMPD1, which encodes ASM, in primary cell culture models, a mouse study, and a human study with untreated MDD patients before and after antidepressive treatment. Our cell culture study showed that a common antidepressant inhibited ASM activity at the enzymatic level and also at the transcriptional level. In a genetically modified mouse line with depressive-like behavior, Smpd1 mRNA expression in dorsal hippocampal tissue was significantly decreased after treatment with a common antidepressant. The large human study showed that SMPD1 mRNA expression in untreated MDD patients decreased significantly after antidepressive treatment. This translational study shows that SMPD1 mRNA expression could serve as a molecular marker for treatment and adherence monitoring of MDD.

A correction to this paper has been published: https://doi.org/10.1007/s00261-021-03084-x

Purpose The aims of this study were to train and evaluate deep learning models for automated segmentation of abdominal organs in whole-body magnetic resonance (MR) images from the UK Biobank (UKBB) and German National Cohort (GNC) MR imaging studies and to make these models available to the scientific community for analysis of these data sets. Methods A total of 200 T1-weighted MR image data sets of healthy volunteers each from UKBB and GNC (400 data sets in total) were available in this study. Liver, spleen, left and right kidney, and pancreas were segmented manually on all 400 data sets, providing labeled ground truth data for training of a previously described U-Net-based deep learning framework for automated medical image segmentation (nnU-Net). The trained models were tested on all data sets using a 4-fold cross-validation scheme. Qualitative analysis of automated segmentation results was performed visually; performance metrics between automated and manual segmentation results were computed for quantitative analysis. In addition, interobserver segmentation variability between 2 human readers was assessed on a subset of the data. Results Automated abdominal organ segmentation was performed with high qualitative and quantitative accuracy on UKBB and GNC data. In more than 90% of data sets, no or only minor visually detectable qualitative segmentation errors occurred. Mean Dice scores of automated segmentations compared with manual reference segmentations were well higher than 0.9 for the liver, spleen, and kidneys on UKBB and GNC data and around 0.82 and 0.89 for the pancreas on UKBB and GNC data, respectively. Mean average symmetric surface distance was between 0.3 and 1.5 mm for the liver, spleen, and kidneys and between 2 and 2.2 mm for pancreas segmentation. The quantitative accuracy of automated segmentation was comparable with the agreement between 2 human readers for all organs on UKBB and GNC data. Conclusion Automated segmentation of abdominal organs is possible with high qualitative and quantitative accuracy on whole-body MR imaging data acquired as part of UKBB and GNC. The results obtained and deep learning models trained in this study can be used as a foundation for automated analysis of thousands of MR data sets of UKBB and GNC and thus contribute to tackling topical and original scientific questions.

Major depressive disorder (MDD) is a severe psychiatric condition with key symptoms of low mood and lack of motivation, joy, and pleasure. Recently, the acid sphingomyelinase (ASM)/ceramide system has been implicated in the pathogenesis of MDD. ASM is a lysosomal glycoprotein that catalyzes the hydrolysis of sphingomyelin, an abundant component of membranes, into the bioactive sphingolipid ceramide, which impacts signaling pathways. ASM activity is inhibited by several common antidepressant drugs. Human and murine studies have confirmed that increased ASM activity and ceramide levels are correlated with MDD. To define a molecular marker for treatment monitoring, we investigated the mRNA expression of SMPD1, which encodes ASM, in primary cell culture models, a mouse study, and a human study with untreated MDD patients before and after antidepressive treatment. Our cell culture study showed that a common antidepressant inhibited ASM activity at the enzymatic level and also at the transcriptional level. In a genetically modified mouse line with depressive-like behavior, Smpd1 mRNA expression in dorsal hippocampal tissue was significantly decreased after treatment with a common antidepressant. The large human study showed that SMPD1 mRNA expression in untreated MDD patients decreased significantly after antidepressive treatment. This translational study shows that SMPD1 mRNA expression could serve as a molecular marker for treatment and adherence monitoring of MDD.

The lockdown measures imposed in Germany starting in mid-March had a significant impact on public life. While the avoidance of social contacts was possible in many areas, the responsibility to provide themselves with food was placed in the hands of private households. To investigate whether and to what extent the restrictions led to changes regarding visits to supermarkets using cars. Data were available on the parking movements on 136 parking spaces of supermarkets in Germany between the beginning of February and the end of May. The analyses focused on changes in the total number of parking events per day and slot as well as their duration. Regarding the parking events, there was a clear reduction until the end of April, interrupted by an increase around Easter. After the introduction of mandatory masking, the values recovered to just below the level before lockdown. The duration changed relatively slightly. Initially it decreased, but later rose slightly above the level observed before the restrictions. Over the medium term, there was a slight reduction of parking events and an extension of their duration by a little more than 5%. Discussion Despite the explicit exclusion of food purchases from the lockdown measures, a dynamic adaptation behaviour of the population to the changing circumstances is also evident in this context. In terms of parking movements, a lasting effect on shopping behaviour seems unlikely. Die in Deutschland ab Mitte März verhängten Ausgangsbeschränkungen hinterließen in vielen Bereichen des öffentlichen Lebens deutliche Spuren. Während das Vermeiden sozialer Kontakte in vielen Bereichen durchaus möglich war, blieb die Versorgung mit Lebensmitteln der Verantwortung der privaten Haushalte überlassen. Die vorliegende Arbeit untersucht, ob und inwiefern die Beschränkungen zu Veränderungen hinsichtlich der Besuche von Supermärkten mit Personenkraftwagen geführt haben. Zur Verfügung stehen Daten über die Parkbewegungen von 136 Parkplätzen von Supermärkten in Deutschland zwischen Anfang Februar und Ende Mai. Im Fokus der Analysen stehen Veränderungen der Gesamtzahl an Parkvorgängen pro Tag und Stellplatz sowie deren Dauer. Mit Blick auf die Parkvorgänge zeigt sich eine deutliche Reduktion bis Ende April, unterbrochen durch einen temporären Anstieg um die Osterfeiertage. Nach Einführung der Maskenpflicht erholen sich die Werte wieder bis knapp unter das Vorniveau. Die Parkdauer verändert sich nur vergleichsweise wenig. Sie sinkt mit den Beschränkungen, steigt jedoch später leicht über das Niveau vor den Beschränkungen. Mittelfristig ist eine leichte Reduzierung der Parkvorgänge und eine Verlängerung ihrer Dauer um jeweils gut 5% zu beobachten. Trotz ausdrücklicher Ausnahme der Einkäufe von Lebensmitteln von den Ausgangsbeschränkungen zeigt sich auch in diesem Bereich ein dynamisches Anpassungsverhalten der Bevölkerung an die sich verändernden Gegebenheiten. Ein dauerhafter Effekt auf die Einkaufsgewohnheiten scheint gemessen an den Parkbewegungen unwahrscheinlich.

Zusammenfassung Hintergrund Die in Deutschland ab Mitte März verhängten Ausgangsbeschränkungen hinterließen in vielen Bereichen des öffentlichen Lebens deutliche Spuren. Während das Vermeiden sozialer Kontakte in vielen Bereichen durchaus möglich war, blieb die Versorgung mit Lebensmitteln der Verantwortung der privaten Haushalte überlassen. Ziel der Arbeit Die vorliegende Arbeit untersucht, ob und inwiefern die Beschränkungen zu Veränderungen hinsichtlich der Besuche von Supermärkten mit Personenkraftwagen geführt haben. Material und Methoden Zur Verfügung stehen Daten über die Parkbewegungen von 136 Parkplätzen von Supermärkten in Deutschland zwischen Anfang Februar und Ende Mai. Im Fokus der Analysen stehen Veränderungen der Gesamtzahl an Parkvorgängen pro Tag und Stellplatz sowie deren Dauer. Ergebnisse Mit Blick auf die Parkvorgänge zeigt sich eine deutliche Reduktion bis Ende April, unterbrochen durch einen temporären Anstieg um die Osterfeiertage. Nach Einführung der Maskenpflicht erholen sich die Werte wieder bis knapp unter das Vorniveau. Die Parkdauer verändert sich nur vergleichsweise wenig. Sie sinkt mit den Beschränkungen, steigt jedoch später leicht über das Niveau vor den Beschränkungen. Mittelfristig ist eine leichte Reduzierung der Parkvorgänge und eine Verlängerung ihrer Dauer um jeweils gut 5% zu beobachten. Schlussfolgerung Trotz ausdrücklicher Ausnahme der Einkäufe von Lebensmitteln von den Ausgangsbeschränkungen zeigt sich auch in diesem Bereich ein dynamisches Anpassungsverhalten der Bevölkerung an die sich verändernden Gegebenheiten. Ein dauerhafter Effekt auf die Einkaufsgewohnheiten scheint gemessen an den Parkbewegungen unwahrscheinlich.

Individuals age differently depending on a multitude of different factors such as lifestyle, medical history and genetics. Often, the global chronological age is not indicative of the true ageing process. An organ-based age estimation would yield a more accurate health state assessment. In this work, we propose a new deep learning architecture for organ-based age estimation based on magnetic resonance images (MRI). The proposed network is a 3D convolutional neural network (CNN) with increased depth and width made possible by the hybrid utilization of inception and fire modules. We apply the proposed framework for the tasks of brain and knee age estimation. Quantitative comparisons against concurrent MR-based regression networks and different 2D and 3D data feeding strategies illustrated the superior performance of the proposed work., Submitted to IEEE EUSIPCO 2020

Brain ageing is a complex neurobiological process associated with morphological changes that can be assessed on MRI scans. Recently, Deep learning (DL)-based approaches have been proposed for the prediction of chronological brain age from MR images yielding high accuracy. These approaches, however, usually do not address quantification of uncertainty and, therefore, intrinsic physiological variability. Considering uncertainty is essential for the interpretation of the difference between predicted and chronological age. In addition, DL-based models lack in explainability compared to classical approaches like voxel-based morphometry. In this study, we aim to address both, modeling uncertainty and providing visual explanations to explore physiological patterns in brain ageing. T1-weighted brain MRI datasets of 10691 participants of the German National Cohort Study, drawn from the general population, were included in this study (chronological age from 20 to 72 years). A regression model based on a 3D Convolutional Neural Network taking into account aleatoric noise was implemented for global as well as regional brain age estimation. We observed high overall accuracy of global brain age estimation with a mean absolute error of 3.2 ± 2.5 years and mean uncertainty of 2.9 ± 0.6 years. Regional brain age estimation revealed higher estimation accuracy and lower uncertainty in central compared to peripheral brain regions. Visual explanations illustrating the importance of brain sub-regions were generated using Grad-CAM: the derived saliency maps showed a high relevance of the lateral and third ventricles, the insular lobe as well as parts of the basal ganglia and the internal capsule.

Age is an essential factor in modern diagnostic procedures. However, assessment of the true biological age (BA) remains a daunting task due to the lack of reference ground-truth labels. Current BA estimation approaches are either restricted to skeletal images or rely on non-imaging modalities that yield a whole-body BA assessment. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. In this initial study, we propose a new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans. As a first step, this framework predicts the chronological age (CA) together with the corresponding patient-dependent aleatoric uncertainty. An iterative training algorithm is then utilized to segregate atypical aging patients from the given population based on the predicted uncertainty scores. In this manner, we hypothesize that training a new model on the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain MRI dataset containing healthy individuals as well as Alzheimer's patients. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients., Comment: Accepted in ICASSP 2021. 5 pages, 4 figures. arXiv admin note: substantial text overlap with arXiv:2009.10765

Image-to-image translation plays a vital role in tackling various medical imaging tasks such as attenuation correction, motion correction, undersampled reconstruction, and denoising. Generative adversarial networks have been shown to achieve the state-of-the-art in generating high fidelity images for these tasks. However, the state-of-the-art GAN-based frameworks do not estimate the uncertainty in the predictions made by the network that is essential for making informed medical decisions and subsequent revision by medical experts and has recently been shown to improve the performance and interpretability of the model. In this work, we propose an uncertainty-guided progressive learning scheme for image-to-image translation. By incorporating aleatoric uncertainty as attention maps for GANs trained in a progressive manner, we generate images of increasing fidelity progressively. We demonstrate the efficacy of our model on three challenging medical image translation tasks, including PET to CT translation, undersampled MRI reconstruction, and MRI motion artefact correction. Our model generalizes well in three different tasks and improves performance over state of the art under full-supervision and weak-supervision with limited data. Code is released here: https://github.com/ExplainableML/UncerGuidedI2I.

Privacy concerns around sharing personally identifiable information are a major barrier to data sharing in medical research. In many cases, researchers have no interest in a particular individual's information but rather aim to derive insights at the level of cohorts. Here, we utilise generative adversarial networks (GANs) to create medical imaging datasets consisting entirely of synthetic patient data. The synthetic images ideally have, in aggregate, similar statistical properties to those of a source dataset but do not contain sensitive personal information. We assess the quality of synthetic data generated by two GAN models for chest radiographs with 14 radiology findings and brain computed tomography (CT) scans with six types of intracranial haemorrhages. We measure the synthetic image quality by the performance difference of predictive models trained on either the synthetic or the real dataset. We find that synthetic data performance disproportionately benefits from a reduced number of classes. Our benchmark also indicates that at low numbers of samples per class, label overfitting effects start to dominate GAN training. We conducted a reader study in which trained radiologists discriminate between synthetic and real images. In accordance with our benchmark results, the classification accuracy of radiologists improves with an increasing resolution. Our study offers valuable guidelines and outlines practical conditions under which insights derived from synthetic images are similar to those that would have been derived from real data. Our results indicate that synthetic data sharing may be an attractive alternative to sharing real patient-level data in the right setting., npj Digital Medicine, 4 (1), ISSN:2398-6352

The concept of biological age (BA), although important in clinical practice, is hard to grasp mainly due to the lack of a clearly defined reference standard. For specific applications, especially in pediatrics, medical image data are used for BA estimation in a routine clinical context. Beyond this young age group, BA estimation is mostly restricted to whole-body assessment using non-imaging indicators such as blood biomarkers, genetic and cellular data. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. Thus, a whole-body assessment of the BA does not reflect the deviations of aging behavior between organs. To this end, we propose a new imaging-based framework for organ-specific BA estimation. In this initial study, we focus mainly on brain MRI. As a first step, we introduce a chronological age (CA) estimation framework using deep convolutional neural networks (Age-Net). We quantitatively assess the performance of this framework in comparison to existing state-of-the-art CA estimation approaches. Furthermore, we expand upon Age-Net with a novel iterative data-cleaning algorithm to segregate atypical-aging patients (BA $\not \approx$ CA) from the given population. We hypothesize that the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain magnetic resonance image (MRI) dataset containing healthy individuals as well as Alzheimer's patients with different dementia ratings. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients. A statistical and visualization-based analysis has provided evidence regarding the potential and current challenges of the proposed methodology., Accepted to IEEE Transcations on Medical Imaging 2021. 13 pages, 14 figures, 4 tables

The number of patients with diabetes is expected to increase by 51% worldwide within the next 25 years, with diabetes complications resulting in high health costs (1). Consequently, comprehensive diabetes care is essential to control disease progression. Pharmacists are able to contribute to diabetes care in different settings (2,3). However, large randomized controlled trials for community pharmacists are rare. Therefore, GLICEMIA 2.0 was developed to examine whether intensive pharmacist-delivered care in community pharmacies can effectively improve glycemic control and reduce risk for complications. GLICEMIA 2.0 was a multicenter, cluster-randomized controlled trial in 26 German pharmacies randomly assigned to the control or intervention group. Pharmacists from each center were trained prior to the start of the trial for 1 day regarding how to conduct the study. Pharmacists of the intervention group received an additional day of training that included medication analysis and counseling of behavioral changes. The Freiburg Ethics Commission International, Freiburg, Germany, approved the trial (016/1826). Eligible participants had type 2 diabetes, BMI of at least 25 kg/m2, and HbA1c ≥7% (53 mmol/mol). Data were collected at baseline and after 6 and 12 months. The assessment included HbA1c (point of care; Abbott Afinion AS 100 Analyzer), fasting blood glucose, weight, and blood pressure. Pharmacists supported the participants of the intervention group for 1 year in a protocol-defined lifestyle change …

OBJECTIVES Longitudinal assessment of changes in apparent diffusion coefficient (ADC)-values in multiple myeloma (MM) patients and their potential role for classifying disease activity. METHODS Retrospective analysis of whole-body-MRI data in 73 stage III MM patients undergoing systemic treatment. Bone marrow involvement was evaluated using a standardized unenhanced 4-sequences whole-body-MRI protocol. We measured ADC-values in focal lesions (FL) and diffusely involved bone marrow (DIBM) areas. Response to treatment was based on the course of hematologic parameters. The time points of MRI-examinations were baseline, 1st (mean, 3 months), 2nd (mean, 10 months), and 3rd (mean, 18 months) follow up (FU). RESULTS Mean IgG and IgA serum values at baseline were 2.1 mg/dl and 1.8 mg/dl, respectively. Patients were classified into responders (n = 59) and non-responders (n = 34). Some patients were re-enrolled for new treatment regimens as they became therapy-refractory. Patterns of medullary involvement were focal (n = 44), diffuse (n = 61) and mixed (n = 30). In FL, a subgroup of myeloma patients undergoing short-term 1st FU experienced a significant increase in ADC in responders (p = 0.001), but not in non-responders (p = 0.9). In the further course of the study, ADC levels decreased continuously in responders (p = 0.02) and increased slightly in non-responders (p = 0.8). In patients with DIBM, ADC values decreased in the responders (p

PURPOSE The purpose of this study was to develop and validate a deep learning-based framework for automated segmentation and vessel shape analysis on non-contrast-enhanced magnetic resonance (MR) data of the thoracic aorta within the German National Cohort (GNC) MR study. MATERIALS AND METHODS One hundred data sets acquired in the GNC MR study were included (56 men, average age 53 y [22 to 72 y]). All participants had undergone non-contrast-enhanced MR imaging of the thoracic vessels. Automated vessel segmentation of the thoracic aorta was performed using a Convolutional Neural Network in a supervised setting with manually annotated data sets as the ground truth. Seventy data sets were used for training; 30 data sets were used for quantitative and qualitative evaluation. Automated shape analysis based on centerline extraction from segmentation masks was performed to derive a diameter profile of the vessel. For comparison, 2 radiologists measured vessel diameters manually. RESULTS Overall, automated aortic segmentation was successful, providing good qualitative analyses with only minor irregularities in 29 of 30 data sets. One data set with severe MR artifacts led to inadequate automated segmentation results. The mean Dice score of automated vessel segmentation was 0.85. Automated aortic diameter measurements were similar to manual measurements (average difference -0.9 mm, limits of agreement: -5.4 to 3.9 mm), with minor deviations in the order of the interreader agreement between the 2 radiologists (average difference -0.5 mm, limits of agreement: -5.8 to 4.8 mm). CONCLUSION Automated segmentation and shape analysis of the thoracic aorta is feasible with high accuracy on non-contrast-enhanced MR imaging using the proposed deep learning approach.

Background Attenuation correction (AC) of PET data is usually performed using a second imaging for the generation of attenuation maps. In certain situations however—when CT- or MR-derived attenuation maps are corrupted or CT acquisition solely for the purpose of AC shall be avoided—it would be of value to have the possibility of obtaining attenuation maps only based on PET information. The purpose of this study was to thus develop, implement, and evaluate a deep learning-based method for whole body [18F]FDG-PET AC which is independent of other imaging modalities for acquiring the attenuation map. Methods The proposed method is investigated on whole body [18F]FDG-PET data using a Generative Adversarial Networks (GAN) deep learning framework. It is trained to generate pseudo CT images (CTGAN) based on paired training data of non-attenuation corrected PET data (PETNAC) and corresponding CT data. Generated pseudo CTs are then used for subsequent PET AC. One hundred data sets of whole body PETNAC and corresponding CT were used for training. Twenty-five PET/CT examinations were used as test data sets (not included in training). On these test data sets, AC of PET was performed using the acquired CT as well as CTGAN resulting in the corresponding PET data sets PETAC and PETGAN. CTGAN and PETGAN were evaluated qualitatively by visual inspection and by visual analysis of color-coded difference maps. Quantitative analysis was performed by comparison of organ and lesion SUVs between PETAC and PETGAN. Results Qualitative analysis revealed no major SUV deviations on PETGAN for most anatomic regions; visually detectable deviations were mainly observed along the diaphragm and the lung border. Quantitative analysis revealed mean percent deviations of SUVs on PETGAN of − 0.8 ± 8.6% over all organs (range [− 30.7%, + 27.1%]). Mean lesion SUVs showed a mean deviation of 0.9 ± 9.2% (range [− 19.6%, + 29.2%]). Conclusion Independent AC of whole body [18F]FDG-PET is feasible using the proposed deep learning approach yielding satisfactory PET quantification accuracy. Further clinical validation is necessary prior to implementation in clinical routine applications.

Background Non-adherence (NA) to immunosuppressants (IS) among renal transplant recipients (RTRs) is associated with higher risk of allograft rejection, graft loss, and mortality. A precise measurement of NA is indispensable, although its prevalence differs greatly depending on the respective measurement methods. The objective of this study was to assess the accuracy and concordance of different measurement methods of NA in patients after renal transplantation. Design and methods This was a single-center prospective observational study. At baseline (T0), NA was measured via physicians’ estimates (PE), self-reports (SR), and tacrolimus trough level variability (CV%) in 78 RTRs. A Visual Analogue Scale (VAS, 0–100%) was applied both for SR and PE. In addition, we used BAASIS© for SR and a 5-point Likert scale for PE. NA was measured prospectively via electronic monitoring (EM, VAICA©) during a three month period. Meanwhile, all participants received phone calls in a two week interval (T1-T6) during which SRs were given. Results Seventy-eight RTRs participated in our study. At t0, NA rates of 6.4%, 28.6%, and 15.4% were found for PE, SR, and CV%, respectively. No correlation was found between these methods. During the study, the percentages of self-reported and electronically monitored adherence remained high, with a minimum mean of 91.2% for the strictest adherence measure (Timing Adherence ±30 min). Our results revealed a moderate to high association between SR and EM. In contrast to PE and CV%, SR significantly predicted electronically monitored adherence. Overall, a decreasing effect of electronically monitored adherence was found for both taking and timing adherence (±2 h, ±30 min) over the course of the study. Discussion The moderate to high concordance of SR and EM suggests that both methods measure NA equally accurately. SR seems to be a method that can adequately depict electronically monitored NA and may represent a good and economical instrument to assess NA in clinical practice. The increased adherence at the beginning of the study and its subsequent decrease suggests an intervention effect. Surveillance of IS intake via EM with intermittent phone calls could improve adherence on a short-term basis. To establish long-term effects, further research is necessary.

PURPOSE: To develop and validate an automated morphometric analysis framework for the quantitative analysis of geometric hip joint parameters in MR images from the German National Cohort (GNC) study. MATERIALS AND METHODS: A secondary analysis on 40 participants (mean age, 51 years; age range, 30–67 years; 25 women) from the prospective GNC MRI study (2015–2016) was performed. Based on a proton density–weighted three-dimensional fast spin-echo sequence, a morphometric analysis approach was developed, including deep learning−based landmark localization, bone segmentation of the femora and pelvis, and a shape model for annotation transfer. The centrum-collum-diaphyseal, center-edge (CE), three alpha angles, head-neck offset (HNO), and HNO ratio along with the acetabular depth, inclination, and anteversion were derived. Quantitative validation was provided by comparison with average manual assessments of radiologists in a cross-validation format. Paired-sample t tests with a Bonferroni-corrected significance level of .005 were employed alongside mean differences and 10th/90th percentiles, median absolute deviations (MADs), and intraclass correlation coefficients (ICCs). RESULTS: High agreement in mean Dice similarity coefficients was achieved (average of 97.52% ± 0.46 [standard deviation]). The subsequent morphometric analysis produced results with low mean MAD values, with the highest values of 3.34° (alpha 03:00 o’clock position) and 0.87 mm (HNO) and ICC values ranging between 0.288 (HNO ratio) and 0.858 (CE) compared with manual assessments. These values were in line with interreader agreements, which at most had MAD values of 4.02° (alpha 12:00 o’clock position) and 1.07 mm (HNO) and ICC values ranging between 0.218 (HNO ratio) and 0.777 (CE). CONCLUSION: Automatic extraction of geometric hip parameters from MRI is feasible using a morphometric analysis approach with deep learning. Keywords: Computer-Aided Diagnosis (CAD), Interventional-MSK, MR-Imaging, Neural Networks, Skeletal-Appendicular, Hip, Anatomy, Computer Applications-3D, Segmentation, Vision, Application Domain, Quantification Supplemental material is available for this article. © RSNA, 2021

Whole-body positron emission tomography/magnetic resonance imaging (wbPET/MRI) is a promising diagnostic tool of recurrent prostate cancer (PC), but its role in primary staging of high-risk PC (hrPC) is not well defined. Thus, the aim was to compare the diagnostic accuracy for T-staging of PET-blinded reading (PBR) and PET/MRI. In this prospective study, hrPC patients scheduled to radical prostatectomy (RPx) with extended lymphadenectomy (eLND) were staged with wbPET/MRI and either 68Ga-PSMA-11 or 11C-choline including simultaneous multiparametric MRI (mpMRI). Images were assessed in two sessions, first as PBR (mpMRI and wbMRI) and second as wbPET/MRI. Prostate Imaging Reporting and Data System criteria (PIRADS v2) were used for T-staging. Results were correlated with the exact anatomical localization and extension as defined by histopathology. Diagnostic accuracy of cTNM stage according to PBR was compared to pathological pTNM stage as reference standard. Thirty-four patients underwent wbPET/MRI of 68Ga-PSMA-11 (n = 17) or 11C-choline (n = 17). Twenty-four patients meeting the inclusion criteria of localized disease ± nodal disease based on imaging results underwent RPx and eLND, whereas ten patients were excluded from analysis due to metastatic disease. T-stage was best defined by mpMRI with underestimation of tumor lesion size by PET for both tracers. N-stage yielded a per patient sensitivity/specificity comparable to PBR. MpMRI is the primary modality for T-staging in hrPC as PET underestimated T-stage in direct comparison to final pathology. In this selected study, cohort MRI shows no inferiority compared to wbPET/MRI considering N-staging.

Purpose To perform follow-up brain MRI in volunteer participants who had previously received multiple doses of gadobutrol and to assess for changes in signal intensities and relaxation times. Materials and Methods This prospective study included 160 participants who received gadobutrol only between 2007 and 2017. The participants were separated into two groups, including participants with at least five contrast agent-enhanced examinations and normal kidney function (group 1) or at least one examination and impaired renal function (group 2). Control groups with normal and impaired renal function (groups 3 and 4) without history of contrast agent exposure were included for comparison. Unenhanced brain MRI was performed in 220 participants (76, 84, 25, and 35 participants in groups 1-4, respectively) with T1-weighted spin-echo and T1 and T2 mapping to determine visual signal intensity changes, signal intensity ratios (globus pallidus-to-thalamus and dentate nucleus-to-pons ratios), and T1 and T2 relaxation times. Results In groups 1 and 2, neither visual signal alterations nor differences in signal intensity ratio or T2 mapping were found. T1 mapping showed no changes for dentate nucleus, pons, and thalamus. However, shorter T1 relaxation times in the globus pallidus were found in group 1 compared with group 3 (difference of -26.2 msec; P = .002), which correlated with the number of previous gadobutrol doses in this group (P = .001). Conclusion In study participants who had previously received gadobutrol, brain MRI showed no differences relative to healthy control participants without gadobutrol exposure. However, quantitative T1 measurements might indicate gadolinium retention in the globus pallidus. © RSNA, 2018 Online supplemental material is available for this article.

Local deformations in medical modalities are common phenomena due to a multitude of factors such as metallic implants or limited field of views in magnetic resonance imaging (MRI). Completion of the missing or distorted regions is of special interest for automatic image analysis frameworks to enhance post-processing tasks such as segmentation or classification. In this work, we propose a new generative framework for medical image inpainting, titled ipA-MedGAN. It bypasses the limitations of previous frameworks by enabling inpainting of arbitrary shaped regions without a prior localization of the regions of interest. Thorough qualitative and quantitative comparisons with other inpainting and translational approaches have illustrated the superior performance of the proposed framework for the task of brain MR inpainting., Comment: Submitted to IEEE ICIP 2020

Summary Background: Penalization and regularization techniques for statistical modeling have attracted increasing attention in biomedical research due to their advantages in the presence of high-dimensional data. A special focus lies on algorithms that incorporate automatic variable selection like the least absolute shrinkage operator (lasso) or statistical boosting techniques. Objectives: Focusing on the linear regression framework, this article compares the two most-common techniques for this task, the lasso and gradient boosting, both from a methodological and a practical perspective. Methods: We describe these methods highlighting under which circumstances their results will coincide in low-dimensional settings. In addition, we carry out extensive simulation studies comparing the performance in settings with more predictors than observations and investigate multiple combinations of noise-to-signal ratio and number of true non-zero coeffcients. Finally, we examine the impact of different tuning methods on the results. Results: Both methods carry out penalization and variable selection for possibly highdimensional data, often resulting in very similar models. An advantage of the lasso is its faster run-time, a strength of the boosting concept is its modular nature, making it easy to extend to other regression settings. Conclusions: Although following different strategies with respect to optimization and regularization, both methods imply similar constraints to the estimation problem leading to a comparable performance regarding prediction accuracy and variable selection in practice.