Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists.

Background: Deep learning convolutional neural networks (CNN) may facilitate melanoma detection, but data comparing a CNN's diagnostic performance to larger groups of dermatologists are lacking. Methods: Google's Inception v4 CNN architecture was trained and validated using dermoscopic images and corresponding diagnoses. In a comparative cross-sectional reader study a 100-image test-set was used (level-I: dermoscopy only; level-II: dermoscopy plus clinical information and images). Main outcome measures were sensitivity, specificity and area under the curve (AUC) of receiver operating characteristics (ROC) for diagnostic classification (dichotomous) of lesions by the CNN versus an international group of 58 dermatologists during level-I or -II of the reader study. Secondary end points included the dermatologists' diagnostic performance in their management decisions and differences in the diagnostic performance of dermatologists during level-I and -II of the reader study. Additionally, the CNN's performance was compared with the top-five algorithms of the 2016 International Symposium on Biomedical Imaging (ISBI) challenge. Results: In level-I dermatologists achieved a mean (±standard deviation) sensitivity and specificity for lesion classification of 86.6% (±9.3%) and 71.3% (±11.2%), respectively. More clinical information (level-II) improved the sensitivity to 88.9% (±9.6%, P = 0.19) and specificity to 75.7% (±11.7%, P < 0.05). The CNN ROC curve revealed a higher specificity of 82.5% when compared with dermatologists in level-I (71.3%, P < 0.01) and level-II (75.7%, P < 0.01) at their sensitivities of 86.6% and 88.9%, respectively. The CNN ROC AUC was greater than the mean ROC area of dermatologists (0.86 versus 0.79, P < 0.01). The CNN scored results close to the top three algorithms of the ISBI 2016 challenge. Conclusions: For the first time we compared a CNN's diagnostic performance with a large international group of 58 dermatologists, including 30 experts. Most dermatologists were outperformed by the CNN. Irrespective of any physicians' experience, they may benefit from assistance by a CNN's image classification. Clinical trial number: This study was registered at the German Clinical Trial Register (DRKS-Study-ID: DRKS00013570; https://www.drks.de/drks_web/).

Integrin αE(CD103) Is Involved in Regulatory T-Cell Function in Allergic Contact Hypersensitivity.

Murine contact hypersensitivity (CHS) is a dendritic cell (DC)-dependent T-cell-mediated inflammation with CD8+ T cells as effectors and CD4+ T cells as regulators (Treg cells) that models human allergic contact dermatitis. The integrin αE(CD103) is expressed by some T-cell and DC subsets and has been implicated in epithelial lymphocyte localization, but its role in immune regulation remains enigmatic. We have identified a function for CD103 in the development of cutaneous allergic immune responses. CHS responses, but not irritant contact dermatitis, were significantly augmented in CD103-deficient mice in hapten-challenged skin. Phenotype and function of skin DCs during sensitization were normal, whereas adoptive transfer experiments revealed that the elevated CHS response in CD103-deficient mice is transferred by primed T cells and is independent of resident cells in recipient mice. While T-cell counts were elevated in challenged skin of CD103-deficient mice, the FoxP3 expression level of CD4+CD25+ Treg cells was significantly reduced, indicating impaired functionality. Indeed, Treg cells from CD103-deficient mice were not able to suppress CHS reactions during the elicitation phase. Further, CD103 on FoxP3+ Treg cells was involved in Treg retention to inflamed skin. These findings indicate an unexpected dichotomous functional role for CD103 on Treg cells by modulating FoxP3 expression.

Inositoylated platelet-activating factor (Ino-C2-PAF) modulates dynamic lymphocyte-endothelial cell interactions and alleviates psoriasis-like skin inflammation in two complementary mouse models.

Psoriasis, a tumor necrosis factor alpha (TNFα)-governed inflammatory disorder with prominent dysregulation of cutaneous vascular functions, has evolved into a model disorder for studying anti-inflammatory therapies. We present experimental in vitro and in vivo data on 1-O-octadecyl-2-O-(2-(myo-inositolyl)-ethyl)-sn-glycero-3-(R/S)-phosphatidyl-choline (Ino-C2-PAF), the lead compound of a class of synthetic glycosylated phospholipids, in anti-inflammatory therapy. Ino-C2-PAF strongly induced apoptosis only in TNFα-stimulated, but not in untreated human vascular endothelial cells. Moreover, TNFα-induced endothelial adhesion molecules that mediated the rolling and firm adhesion of leukocytes (vascular cell adhesion protein-1 (VCAM-1), E-selectin, and ICAM-1) were selectively downregulated by Ino-C2-PAF. Similarly, expression of L-selectin, VCAM-1 receptor α4ß1 integrin , and lymphocyte function-associated antigen-1 on human peripheral blood mononuclear cells was reduced without induction of apoptosis. Functionally, these changes were accompanied by significant impairment of rolling and adhesion of human peripheral blood lymphocytes on TNFα-activated endothelial cells in a dynamic flow chamber system. When the therapeutic potential of Ino-C2-PAF was assessed in two complementary mouse models of psoriasis, K5.hTGFß1 transgenic and JunB/c-Jun-deficient mice, Ino-C2-PAF led to significant alleviation of the clinical symptoms and normalized the pathological cutaneous changes including vascularization. There were no overt adverse effects. These findings suggested that Ino-C2-PAF is a potential candidate in the therapy of inflammatory skin diseases that include abnormal vascular functions.

TNF receptor I on human keratinocytes is a binding partner for staphylococcal protein A resulting in the activation of NF kappa B, AP-1, and downstream gene transcription.

Primary human keratinocytes and immortalized HaCaT cells were analysed for their capacity to bind purified staphylococcal protein A (SpA). Co-incubation with FITC-labelled SpA led to a dose-depending attachment. Pull-down experiments with cellular extracts revealed the TNFα receptor I (TNF RI) as binding partner on keratinocytes. Thus, we next looked for expression of this receptor in human epidermis and cultured keratinocytes. TNF RI is strongly expressed on all keratinocytes analysed, both at the mRNA and protein level and activation by SpA at optimal doses of 50-100 µg/ml resulted in the phosphorylation of the TNF RI downstream kinases MEK1/2, JNK1/2, and p38 subsequently leading to translocation of the p65 NF kappa B subunit and AP-1 into the nucleus. This translocation was then followed by increased expression of IL-8 and COX-2, two known NF kappa B-induced pro-inflammatory genes. To further test the relevance of our findings, we analysed in vitro production of over 100 strains isolated from atopic eczema showing that more than 85% of the tested strains produced extracellular SpA in substantial amounts. Thus, besides superantigens, haemolysins, and other cell wall components, Staphylococcus aureus exerts pro-inflammatory stimuli on human keratinocytes through the production of SpA signalling through TNF RI.