Consistency of convolutional neural networks in dermoscopic melanoma recognition: A prospective real-world study about the pitfalls of augmented intelligence.

BACKGROUND: Deep-learning convolutional neural networks (CNNs) have outperformed even experienced dermatologists in dermoscopic melanoma detection under controlled conditions. It remains unexplored how real-world dermoscopic image transformations affect CNN robustness. OBJECTIVES: To investigate the consistency of melanoma risk assessment by two commercially available CNNs to help formulate recommendations for current clinical use. METHODS: A comparative cohort study was conducted from January to July 2022 at the Department of Dermatology, University Hospital Basel. Five dermoscopic images of 116 different lesions on the torso of 66 patients were captured consecutively by the same operator without deliberate rotation. Classification was performed by two CNNs (CNN-1/CNN-2). Lesions were divided into four subgroups based on their initial risk scoring and clinical dignity assessment. Reliability was assessed by variation and intraclass correlation coefficients. Excisions were performed for melanoma suspicion or two consecutively elevated CNN risk scores, and benign lesions were confirmed by expert consensus (n = 3). RESULTS: 117 repeated image series of 116 melanocytic lesions (2 melanomas, 16 dysplastic naevi, 29 naevi, 1 solar lentigo, 1 suspicious and 67 benign) were classified. CNN-1 demonstrated superior measurement repeatability for clinically benign lesions with an initial malignant risk score (mean variation coefficient (mvc): CNN-1: 49.5(±34.3)%; CNN-2: 71.4(±22.5)%; p = 0.03), while CNN-2 outperformed for clinically benign lesions with benign scoring (mvc: CNN-1: 49.7(±22.7)%; CNN-2: 23.8(±29.3)%; p = 0.002). Both systems exhibited lowest score consistency for lesions with an initial malignant risk score and benign assessment. In this context, averaging three initial risk scores achieved highest sensitivity of dignity assessment (CNN-1: 94%; CNN-2: 89%). Intraclass correlation coefficients indicated 'moderate'-to-'good' reliability for both systems (CNN-1: 0.80, 95% CI:0.71-0.87, p < 0.001; CNN-2: 0.67, 95% CI:0.55-0.77, p < 0.001). CONCLUSIONS: Potential user-induced image changes can significantly influence CNN classification. For clinical application, we recommend using the average of three initial risk scores. Furthermore, we advocate for CNN robustness optimization by cross-validation with repeated image sets. TRIAL REGISTRATION: ClinicalTrials.gov (NCT04605822).

MEK inhibition induces expression of differentiation marker Keratin 10 in human keratinocytes.

BRAF mutant metastatic melanoma was regularly treated in the past with a BRAF inhibitor (BRAFi) alone or in combination with inhibitors of the mitogen-activated protein kinase kinase (MEKi), which is still a common treatment. This combination therapy strongly reduced the occurrence of keratoacanthomas and squamous cell carcinoma, which was frequently seen when BRAFi was used as monotherapy. Here we addressed the question whether MEK inhibition counteracts squamous cell carcinoma development in part by promoting keratinocyte differentiation. Exposure of human immortalized keratinocytes to different concentrations of MEKi revealed a significant increase in the expression of differentiation-associated keratins K10 and K1 as determined by qRT-PCR and immunofluorescence staining. Taken together, the present study suggests that in a combined treatment of melanoma with BRAFi/MEKi, MEKi reduces the incidence of squamous cell carcinomas by promoting keratinocyte differentiation under combined BRAFi/MEKi treatment in melanoma. This might open further treatment perspectives for skin cancer treatment.

A flexible framework for anomaly Detection via dimensionality reduction.

Anomaly detection is challenging, especially for large datasets in high dimensions. Here, we explore a general anomaly detection framework based on dimensionality reduction and unsupervised clustering. DRAMA is released as a general python package that implements the general framework with a wide range of built-in options. This approach identifies the primary prototypes in the data with anomalies detected by their large distances from the prototypes, either in the latent space or in the original, high-dimensional space. DRAMA is tested on a wide variety of simulated and real datasets, in up to 3000 dimensions, and is found to be robust and highly competitive with commonly used anomaly detection algorithms, especially in high dimensions. The flexibility of the DRAMA framework allows for significant optimization once some examples of anomalies are available, making it ideal for online anomaly detection, active learning, and highly unbalanced datasets. Besides, DRAMA naturally provides clustering of outliers for subsequent analysis.

Improved diagnosis by automated macro- and micro-anatomical region mapping of skin photographs.

BACKGROUND: The exact location of skin lesions is key in clinical dermatology. On one hand, it supports differential diagnosis (DD) since most skin conditions have specific predilection sites. On the other hand, location matters for dermatosurgical interventions. In practice, lesion evaluation is not well standardized and anatomical descriptions vary or lack altogether. Automated determination of anatomical location could benefit both situations. OBJECTIVE: Establish an automated method to determine anatomical regions in clinical patient pictures and evaluate the gain in DD performance of a deep learning model (DLM) when trained with lesion locations and images. METHODS: Retrospective study based on three datasets: macro-anatomy for the main body regions with 6000 patient pictures partially labelled by a student, micro-anatomy for the ear region with 182 pictures labelled by a student and DD with 3347 pictures of 16 diseases determined by dermatologists in clinical settings. For each dataset, a DLM was trained and evaluated on an independent test set. The primary outcome measures were the precision and sensitivity with 95% CI. For DD, we compared the performance of a DLM trained with lesion pictures only with a DLM trained with both pictures and locations. RESULTS: The average precision and sensitivity were 85% (CI 84-86), 84% (CI 83-85) for macro-anatomy, 81% (CI 80-83), 80% (CI 77-83) for micro-anatomy and 82% (CI 78-85), 81% (CI 77-84) for DD. We observed an improvement in DD performance of 6% (McNemar test P-value 0.0009) for both average precision and sensitivity when training with both lesion pictures and locations. CONCLUSION: Including location can be beneficial for DD DLM performance. The proposed method can generate body region maps from patient pictures and even reach surgery relevant anatomical precision, e.g. the ear region. Our method enables automated search of large clinical databases and make targeted anatomical image retrieval possible.

COVID-19 related psychological burden and potential benefits of vaccination - Data from a repeated cross-sectional survey in healthcare workers.

The COVID-19 pandemic is impacting the psychological well-being, especially of health care workers, for more than two years now. Here, we followed-up on a survey we conducted at the very beginning of the pandemic, to determine potential changes in psychological strain experienced by health care workers one year later. Since our first survey in 2020, COVID-19 vaccines have been established, thus we assessed whether vaccination-status might modulate psychological burden of health care workers. We also collected data on resilience and sleep, as those might be related to successful coping. Between March and April 2021, nurses and physicians (N â€‹= â€‹286) working at the University Hospital Augsburg - with high or low exposure to COVID-19 patients - took part in an online survey. We found that fully vaccinated personnel reported lower levels of anxiety, depression, stress and exhaustion suggesting the potential positive consequences of vaccination beyond the obvious protection against a COVID-19 infection. Nurses reported more depressive symptoms, anxiety, stress and exhaustion and lower levels of job fulfilment than physicians. Individuals with high exposure to COVID-19 patients reported higher exhaustion and depersonalization. Resilience and sleep quality were significantly correlated with psychological and work-related burden, suggesting their potential role as protective resources. In general, the comparison of the present data to the survey conducted in 2020 suggests an overall increase of psychological burden in health care workers. Despite these surely alarming findings, it should be noted that being vaccinated might come along with reduced psychological strain.

Formation of Surface Impurities on Lithium-Nickel-Manganese-Cobalt Oxides in the Presence of CO2 and H2O.

Surface impurities involving parasitic reactions and gas evolution contribute to the degradation of high Ni content LiNixMnyCozO2 (NMC) cathode materials. The transient kinetic technique of temporal analysis of products (TAP), density functional theory, and infrared spectroscopy have been used to study the formation of surface impurities on varying nickel content NMC materials (NMC811, NMC622, NMC532, NMC433, NMC111) in the presence of CO2 and H2O. CO2 reactivity on a clean surface as characterized by CO2 conversion rate in the TAP reactor follows the order: NMC811 > NMC622 > NMC532 > NMC433 > NMC111. The capacity of CO2 uptake follows a different order: NMC532 > NMC433 > NMC622 > NMC811 > NMC111. Moisture pretreatment slows down the direct CO2 adsorption process and creates additional active sites for CO2 adsorption. Electronic structure calculations predict that the (012) surface is more reactive than the (1014) surface for CO2 and H2O adsorption. CO2 adsorption leading to carbonate formation is exothermic with formation of ion pairs. The average CO2 binding energies on the different materials follow the CO2 reactivity order. Water hydroxylates the (012) surface and surface OH groups favor bicarbonate formation. Water creates more active sites for CO2 adsorption on the (1014) surface due to hydrogen bonding. The composition of surface impurities formed in ambient air exposure is dependent on water concentration and the percentage of different crystal planes. Different surface reactivities suggest that battery performance degradation due to surface impurities can be mitigated by precise control of the dominant surfaces in NMC materials.

Understanding Reaction Networks through Controlled Approach to Equilibrium Experiments Using Transient Methods.

We report a combined experimental/theoretical approach to studying heterogeneous gas/solid catalytic processes using low-pressure pulse response experiments achieving a controlled approach to equilibrium that combined with quantum mechanics (QM)-based computational analysis provides information needed to reconstruct the role of the different surface reaction steps. We demonstrate this approach using model catalysts for ammonia synthesis/decomposition. Polycrystalline iron and cobalt are studied via low-pressure TAP (temporal analysis of products) pulse response, with the results interpreted through reaction free energies calculated using QM on Fe-BCC(110), Fe-BCC(111), and Co-FCC(111) facets. In TAP experiments, simultaneous pulsing of ammonia and deuterium creates a condition where the participation of reactants and products can be distinguished in both forward and reverse reaction steps. This establishes a balance between competitive reactions for D* surface species that is used to observe the influence of steps leading to nitrogen formation as the nitrogen product remains far from equilibrium. The approach to equilibrium is further controlled by introducing delay timing between NH3 and D2 which allows time for surface reactions to evolve before being driven in the reverse direction from the gas phase. The resulting isotopic product distributions for NH2D, NHD2, and HD at different temperatures and delay times and NH3/D2 pulsing order reveal the role of the N2 formation barrier in controlling the surface concentration of NHx* species, as well as providing information on the surface lifetimes of key reaction intermediates. Conclusions derived for monometallic materials are used to interpret experimental results on a more complex and active CoFe bimetallic catalyst.

CD137 (4-1BB) stimulation leads to metabolic and functional reprogramming of human monocytes/macrophages enhancing their tumoricidal activity.

Immunotherapies have heralded a new era in the cancer treatment. In addition to checkpoint inhibitors, agonistic antibodies against co-stimulatory immune receptors hold the potential to invoke efficient antitumor immunity. Targeting CD137 has gained momentum based on its ability to drive NK- and T-cell-based responses. CD137-engaging mAbs have already entered clinical trials for different types of tumors showing promising results. Despite the efforts to translate CD137-mediated immunotherapy into clinical practice, little remains known regarding the role of CD137 in human monocytes/macrophages.We found CD137 being expressed on monocytes of healthy controls and at even higher levels in patients with multiple myeloma or CLL. CD137HI(GH) monocytes displayed a distinct phenotypic, transcriptomic, and metabolic profile. They possessed an increased phagocytic capacity enabling superior antibody-dependent phagocytosis (ADPC) of multiple myeloma and lymphoma cells that were treated with anti-CD38 or anti-CD20 mAbs. Triggering CD137 promoted both metabolic and tumoricidal activity in an extracellular signal-regulated kinase (ERK)-dependent fashion. In addition, we observed a phenotypic, transcriptomic, and functional skewing towards a M1-like phenotype.Overall, we introduce CD137 as a positive immune checkpoint on human monocytes/macrophages, which can have therapeutic implications especially in view of synergistic effects when combining CD137 agonists with tumor-targeting antibodies.

Early identification of bipolar disorder among young adults - a 22-year community birth cohort.

OBJECTIVE: We set forth to build a prediction model of individuals who would develop bipolar disorder (BD) using machine learning techniques in a large birth cohort. METHODS: A total of 3748 subjects were studied at birth, 11, 15, 18, and 22 years of age in a community birth cohort. We used the elastic net algorithm with 10-fold cross-validation to predict which individuals would develop BD at endpoint (22 years) at each follow-up visit before diagnosis (from birth up to 18 years). Afterward, we used the best model to calculate the subgroups of subjects at higher and lower risk of developing BD and analyzed the clinical differences among them. RESULTS: A total of 107 (2.8%) individuals within the cohort presented with BD type I, 26 (0.6%) with BD type II, and 87 (2.3%) with BD not otherwise specified. Frequency of female individuals was 58.82% (n = 150) in the BD sample and 53.02% (n = 1868) among the unaffected population. The model with variables assessed at the 18-year follow-up visit achieved the best performance: AUC 0.82 (CI 0.75-0.88), balanced accuracy 0.75, sensitivity 0.72, and specificity 0.77. The most important variables to detect BD at the 18-year follow-up visit were suicide risk, generalized anxiety disorder, parental physical abuse, and financial problems. Additionally, the high-risk subgroup of BD showed a high frequency of drug use and depressive symptoms. CONCLUSIONS: We developed a risk calculator for BD incorporating both demographic and clinical variables from a 22-year birth cohort. Our findings support previous studies in high-risk samples showing the significance of suicide risk and generalized anxiety disorder prior to the onset of BD, and highlight the role of social factors and adverse life events.

On-site CAD templates reduce surgery time for complex craniostenosis repair in infants: a new method.

INTRODUCTION: The surgical correction of craniostenosis in children is a time-consuming and taxing procedure. To facilitate this procedure, especially in infants with complex craniostenosis, we refined the computer-aided design and manufacturing technique (CAD/CAM) based on computed tomography (CT)-generated DICOM data. We used cutting guides and molding templates, which allowed the surgeon to reshape and fixate the supraorbital bar extracorporeally on a side table and to control the intracorporal fit without removing the template. METHOD AND PATIENTS: To compare our traditional concept with the possibility of preoperative virtual planning (PVP) technique, the surgical treatment and courses of 16 infants with complex craniostenosis following fronto-orbital advancement (FOA) (age range 8-15 months) were analyzed in two groups (group 1: traditional, control group n = 8, group 2: CAD/CAM planned, n = 8). RESULTS: While in both groups, the head accurately reshaped postoperatively during the follow-up; the CAD group 2 showed a significantly shorter operating time with a mean of 4 h 25 min compared with group 1 with a mean of 5 h 37 min (p = 0.038). Additionally, the CAD group 2 had a significantly lower volume of blood loss (380 ml vs. 575 ml mean, p = 0.047), lower blood transfusion volume (285 ml vs. 400 ml mean, p = 0.108), lower fresh frozen plasma (FFP) volume (140 ml vs. 275 ml mean, p = 0.019), shorter stay in the pediatric intensive care unit (PICU) (3 vs. 5 days mean (p = 0.002), and shorter total length of hospital stay (6 days vs. 8 days mean, p = 0.002). CONCLUSION: CAD/CAM cutting guides and templates offer optimizing operative efficiency, precision, and accuracy in craniostenosis surgery in infants. As shown in this single-center observational study, the use of on-site templates significantly accelerates the reconstruction of the bandeau. The virtual 3D planning technique increases surgical precision without discernible detrimental effects.

[Assessing pain in patients with dementia]. / Schmerzerfassung bei Patienten mit Demenz.

The diagnostics of pain in older people with cognitive impairments should always consist of a subjective self-report of pain and a structured observation of pain behavior. It is important to note that the subjective self-report of pain becomes less valid with increasing cognitive decline (starting with a moderate degree of dementia). The external observation of pain behavior should include at least the three behavioral domains facial expressions, body movements and vocalization and should be performed during resting situations and during activities of daily living. Moreover, the patient should be observed for at least 3 min. Online forms of training have recently been developed and are freely available for training in external observation.

Rate/Concentration Kinetic Petals: A Transient Method to Examine the Interplay of Surface Reaction Processes.

Transient pulse response experiments are used to construct rate/concentration kinetic dependencies, RC Petals and provide a new method to distinguish the timing and interplay of adsorption, surface reaction, and product formation on complex (industrial) materials. A petal shape arises as the dynamic "reaction-diffusion" experiment forces the concentration and reaction rate to return to zero. In contrast to the typical steady-state "Langmuir-type" RC dependence, RC petals have two branches, which arise as a result of decoupled gas and surface concentrations in the non-steady-state regime. To demonstrate this approach, the characteristics of petal shapes using ammonia decomposition as a probe reaction are presented. Ammonia, hydrogen, and nitrogen transformation rates are compared on three simple materials: iron, cobalt, and a bimetallic CoFe preparation when ammonia is pulsed at 550 °C in a low-pressure diffusion reactor. All materials demonstrate a two-branch kinetic RC dependence for ammonia adsorption, and rate constants are quantified in the low-coverage regime. We found that H2 and N2 product formation was dependent on the concentration of surface intermediates for all materials with one exception: for cobalt, an additional fast hydrogen generation process was observed; the rate of which coincided with ammonia adsorption. Nitrogen generation was only significant for CoFe and cobalt and on the CoFe catalyst, a self-inhibition property was observed. A method for estimating the number of active sites based on the RC petals is presented and was applied to the iron and CoFe samples. The surface coverage and rate of formation/conversion of surface intermediates are interpreted from the examination of shape characteristics of the RC petals for each material.

Absence of herb-drug interactions of mistletoe with the tamoxifen metabolite (E/Z)-endoxifen and cytochrome P450 3A4/5 and 2D6 in vitro.

BACKGROUND: Women diagnosed with breast cancer frequently seek complementary and alternative (CAM) treatment options that can help to cope with their disease and the side effects of conventional cancer therapy. Especially in Europe, breast cancer patients use herbal products containing mistletoe (Viscum album L.). The oldest and one of the most prescribed conventional drugs for the treatment of estrogen receptor positive breast cancer is tamoxifen. Aside from positive clinical experience with the combination of tamoxifen and mistletoe, little is known about possible herb-drug interactions (HDIs) between the two products. In the present in vitro study, we investigated the effect of standardized commercial mistletoe preparations on the activity of endoxifen, the major active metabolite of tamoxifen. METHODS: The estrogen receptor positive human breast carcinoma cell line MCF-7 was treated with (E/Z)-endoxifen hydrochloride in the presence and absence of a defined estradiol concentration. Each concentration of the drug was combined with fermented Viscum album L. extracts (VAE) at clinically relevant doses, and proliferation, apoptosis and cell cycle were analyzed. In parallel, possible inhibition of CYP3A4/5 and CYP2D6 was investigated using 50-donor mixed gender pooled human liver microsomes (HLMs). RESULTS: VAE did not inhibit endoxifen induced cytostasis and cytotoxicity. At higher concentrations, VAE showed an additive inhibitory effect. VAE preparations did not cause inhibition of CYP3A4/5 and CYP2D6 catalyzed tamoxifen metabolism. CONCLUSIONS: The in vitro results suggest that mistletoe preparations can be used in combination with tamoxifen without the risk of HDIs.

Magneto-immutable turbulence in weakly collisional plasmas.

We propose that pressure anisotropy causes weakly collisional turbulent plasmas to self-organize so as to resist changes in magnetic-field strength. We term this effect "magneto-immutability" by analogy with incompressibility (resistance to changes in pressure). The effect is important when the pressure anisotropy becomes comparable to the magnetic pressure, suggesting that in collisionless, weakly magnetized (high-ß) plasmas its dynamical relevance is similar to that of incompressibility. Simulations of magnetized turbulence using the weakly collisional Braginskii model show that magneto-immutable turbulence is surprisingly similar, in most statistical measures, to critically balanced MHD turbulence. However, in order to minimize magnetic-field variation, the flow direction becomes more constrained than in MHD, and the turbulence is more strongly dominated by magnetic energy (a nonzero "residual energy"). These effects represent key differences between pressure-anisotropic and fluid turbulence, and should be observable in the ß â‰³ 1 turbulent solar wind.

Publisher's Note: Kinetic Simulations of the Interruption of Large-Amplitude Shear-Alfvén Waves in a High-ß Plasma [Phys. Rev. Lett. 119, 155101 (2017)].

This corrects the article DOI: 10.1103/PhysRevLett.119.155101.

Kinetic Simulations of the Interruption of Large-Amplitude Shear-Alfvén Waves in a High-ß Plasma.

Using two-dimensional hybrid-kinetic simulations, we explore the nonlinear "interruption" of standing and traveling shear-Alfvén waves in collisionless plasmas. Interruption involves a self-generated pressure anisotropy removing the restoring force of a linearly polarized Alfvénic perturbation, and occurs for wave amplitudes δB_{⊥}/B_{0}≳ß^{-1/2} (where ß is the ratio of thermal to magnetic pressure). We use highly elongated domains to obtain maximal scale separation between the wave and the ion gyroscale. For standing waves above the amplitude limit, we find that the large-scale magnetic field of the wave decays rapidly. The dynamics are strongly affected by the excitation of oblique firehose modes, which transition into long-lived parallel fluctuations at the ion gyroscale and cause significant particle scattering. Traveling waves are damped more slowly, but are also influenced by small-scale parallel fluctuations created by the decay of firehose modes. Our results demonstrate that collisionless plasmas cannot support linearly polarized Alfvén waves above δB_{⊥}/B_{0}∼ß^{-1/2}. They also provide a vivid illustration of two key aspects of low-collisionality plasma dynamics: (i) the importance of velocity-space instabilities in regulating plasma dynamics at high ß, and (ii) how nonlinear collisionless processes can transfer mechanical energy directly from the largest scales into thermal energy and microscale fluctuations, without the need for a scale-by-scale turbulent cascade.