Does sex matter? Analysis of sex-related differences in the diagnostic performance of a market-approved convolutional neural network for skin cancer detection.

BACKGROUND: Advances in biomedical artificial intelligence may introduce or perpetuate sex and gender discriminations. Convolutional neural networks (CNN) have proven a dermatologist-level performance in image classification tasks but have not been assessed for sex and gender biases that may affect training data and diagnostic performance. In this study, we investigated sex-related imbalances in training data and diagnostic performance of a market-approved CNN for skin cancer classification (Moleanalyzer Pro®, Fotofinder Systems GmbH, Bad Birnbach, Germany). METHODS: We screened open-access dermoscopic image repositories widely used for CNN training for distribution of sex. Moreover, the sex-related diagnostic performance of the market-approved CNN was tested in 1549 dermoscopic images stratified by sex (female n = 773; male n = 776). RESULTS: Most open-access repositories showed a marked under-representation of images originating from female (40%) versus male (60%) patients. Despite these imbalances and well-known sex-related differences in skin anatomy or skin-directed behaviour, the tested CNN achieved a comparable sensitivity of 87.0% [80.9%-91.3%] versus 87.1% [81.1%-91.4%], specificity of 98.7% [97.4%-99.3%] versus 96.9% [95.2%-98.0%] and ROC-AUC of 0.984 [0.975-0.993] versus 0.979 [0.969-0.988] in dermoscopic images of female versus male origin, respectively. In the sample at hand, sex-related differences in ROC-AUCs were not statistically significant in the per-image analysis nor in an additional per-individual analysis (p ≥ 0.59). CONCLUSION: Design and training of artificial intelligence algorithms for medical applications should generally acknowledge sex and gender dimensions. Despite sex-related imbalances in open-access training data, the diagnostic performance of the tested CNN showed no sex-related bias in the classification of skin lesions.

Monitoring patients at risk for melanoma: May convolutional neural networks replace the strategy of sequential digital dermoscopy?

BACKGROUND: Sequential digital dermoscopy (SDD) is applied for early melanoma detection by uncovering dynamic changes of monitored lesions. Convolutional neural networks (CNN) are capable of high diagnostic accuracies similar to trained dermatologists. OBJECTIVES: To investigate the capability of CNN to correctly classify melanomas originally diagnosed by mere dynamic changes during SDD. METHODS: A retrospective cross-sectional study using image quartets of 59 high-risk patients each containing one melanoma diagnosed by dynamic changes during SDD and three nevi (236 lesions). Two validated CNN classified quartets at baseline or after SDD follow-up at the time of melanoma diagnosis. Moreover, baseline quartets were rated by 26 dermatologists. The main outcome was the number of quartets with correct classifications. RESULTS: CNN-1 correctly classified 9 (15.3%) and CNN-2 8 (13.6%) of 59 baseline quartets. In baseline images, CNN-1 attained a sensitivity of 25.4% (16.1%-37.8%) and specificity of 92.7% (87.8%-95.7%), whereas CNN-2 of 28.8% (18.8%-41.4%) and 75.7% (68.9%-81.4%). Expectedly, after SDD follow-up CNN more readily detected melanomas resulting in improved sensitivities (CNN-1: 44.1% [32.2%-56.7%]; CNN-2: 49.2% [36.8%-61.6%]). Dermatologists were told that each baseline quartet contained one melanoma, and on average, correctly classified 24 (22-27) of 59 quartets. Correspondingly, accepting a baseline quartet to be appropriately classified whenever the highest malignancy score was assigned to the melanoma within, CNN-1 and CNN-2 correctly classified 28 (47.5%) and 22 (37.3%) of 59 quartets, respectively. CONCLUSIONS: The tested CNN could not replace the strategy of SDD. There is a need for CNN capable of integrating information on dynamic changes into analyses.

Collective human intelligence outperforms artificial intelligence in a skin lesion classification task.

BACKGROUND AND OBJECTIVES: Convolutional neural networks (CNN) enable accurate diagnosis of medical images and perform on or above the level of individual physicians. Recently, collective human intelligence (CoHI) was shown to exceed the diagnostic accuracy of individuals. Thus, diagnostic performance of CoHI (120 dermatologists) versus individual dermatologists versus two state-of-the-art CNN was investigated. PATIENTS AND METHODS: Cross-sectional reader study with presentation of 30 clinical cases to 120 dermatologists. Six diagnoses were offered and votes collected via remote voting devices (quizzbox®, Quizzbox Solutions GmbH, Stuttgart, Germany). Dermatoscopic images were classified by a binary and multiclass CNN (FotoFinder Systems GmbH, Bad Birnbach, Germany). Three sets of diagnostic classifications were scored against ground truth: (1) CoHI, (2) individual dermatologists, and (3) CNN. RESULTS: CoHI attained a significantly higher accuracy [95 % confidence interval] (80.0 % [62.7 %-90.5 %]) than individual dermatologists (75.7 % [73.8 %-77.5 %]) and CNN (70.0 % [52.1 %-83.3 %]; all P < 0.001) in binary classifications. Moreover, CoHI achieved a higher sensitivity (82.4 % [59.0 %-93.8 %]) and specificity (76.9 % [49.7 %-91.8 %]) than individual dermatologists (sensitivity 77.8 % [75.3 %-80.2 %], specificity 73.0 % [70.6 %-75.4 %]) and CNN (sensitivity 70.6 % [46.9 %-86.7 %], specificity 69.2 % [42.4 %-87.3 %]). The diagnostic accuracy of CoHI was superior to that of individual dermatologists (P < 0.001) in multiclass evaluation, with the accuracy of the latter comparable to multiclass CNN. CONCLUSIONS: Our analysis revealed that the majority vote of an interconnected group of dermatologists (CoHI) outperformed individuals and CNN in a demanding skin lesion classification task.

Dark corner artefact and diagnostic performance of a market-approved neural network for skin cancer classification.

BACKGROUND AND OBJECTIVES: Convolutional neural networks (CNN) have proven dermatologist-level performance in skin lesion classification. Prior to a broader clinical application, an assessment of limitations is crucial. Therefore, the influence of a dark tubular periphery in dermatoscopic images (also called dark corner artefact [DCA]) on the diagnostic performance of a market-approved CNN for skin lesion classification was investigated. PATIENTS AND METHODS: A prospective image set of 233 skin lesions (60 malignant, 173 benign) without DCA (control-set) was modified to show small, medium or large DCA. All 932 images were analyzed by a market-approved CNN (Moleanalyzer-Pro® , FotoFinder Systems), providing malignancy scores (range 0-1) with the cut-off > 0.5 indicating malignancy. RESULTS: In the control-set the CNN achieved a sensitivity of 90.0 % (79.9 % - 95.3 %), a specificity of 96.5 % (92.6 % - 98.4 %), and an area under the curve (AUC) of receiver operating characteristics (ROC) of 0.961 (0.932 - 0.989). Comparable diagnostic performance was observed in the DCAsmall-set and DCAmedium-set. Conversely, in the DCAlarge-set significantly increased malignancy scores triggered a significantly decreased specificity (87.9 % [82.2 % - 91.9 %], P < 0.001), non-significantly increased sensitivity (96.7 % [88.6 % - 99.1 %]) and unchanged ROC-AUC of 0.962 (0.935 - 0.989). CONCLUSIONS: Convolutional neural network classification was robust in images with small and medium DCA, but impaired in images with large DCA. Physicians should be aware of this limitation when submitting images to CNN classification.

Association between different scale bars in dermoscopic images and diagnostic performance of a market-approved deep learning convolutional neural network for melanoma recognition.

BACKGROUND: Studies systematically unravelling possible causes for false diagnoses of deep learning convolutional neural networks (CNNs) are scarce, yet needed before broader application. OBJECTIVES: The objective of the study was to investigate whether scale bars in dermoscopic images are associated with the diagnostic accuracy of a market-approved CNN. METHODS: This cross-sectional analysis applied a CNN trained with more than 150,000 images (Moleanalyzer-pro®, FotoFinder Systems Inc., Bad Birnbach, Germany) to investigate seven dermoscopic image sets depicting the same 130 melanocytic lesions (107 nevi, 23 melanomas) without or with digitally superimposed scale bars of different manufacturers. Sensitivity, specificity and area under the curve (AUC) of receiver operating characteristics (ROC) for the CNN's binary classification of images with or without superimposed scale bars were assessed. RESULTS: Six dermoscopic image sets with different scale bars and one control set without scale bars (overall 910 images) were submitted to CNN analysis. In images without scale bars, the CNN attained a sensitivity [95% confidence interval] of 87.0% [67.9%-95.5%] and a specificity of 87.9% [80.3%-92.8%]. ROC AUC was 0.953 [0.914-0.992]. Scale bars were not associated with significant changes in sensitivity (range 87%-95.7%, all p ≥ 1.0). However, four scale bars induced a decrease of the CNN's specificity (range 0%-43.9%, all p < 0.001). Moreover, ROC AUC was significantly reduced by two scale bars (range 0.520-0.848, both p ≤ 0.042). CONCLUSIONS: Superimposed scale bars in dermoscopic images may impair the CNN's diagnostic accuracy, mostly by increasing the rate of the false-positive diagnoses. We recommend avoiding scale bars in images intended for CNN analysis unless specific measures counteracting effects are implemented. CLINICAL TRIAL NUMBER: This study was registered at the German Clinical Trial Register (DRKS-Study-ID: DRKS00013570; URL: https://www.drks.de/drks_web/).

Past and present of computer-assisted dermoscopic diagnosis: performance of a conventional image analyser versus a convolutional neural network in a prospective data set of 1,981 skin lesions.

BACKGROUND: Convolutional neural networks (CNNs) have shown a dermatologist-level performance in the classification of skin lesions. We aimed to deliver a head-to-head comparison of a conventional image analyser (CIA), which depends on segmentation and weighting of handcrafted features, to a CNN trained by deep learning. METHODS: Cross-sectional study using a real-world, prospectively acquired, dermoscopic dataset of 1981 skin lesions to compare the diagnostic performance of a market-approved CNN (Moleanalyzer-Pro™, developed in 2018) to a CIA (Moleanalyzer-3™/Dynamole™; developed in 2004, all FotoFinder Systems Inc, Germany). As a reference standard, we used histopathological diagnoses (n = 785) or, in non-excised benign lesions (n = 1196), expert consensus plus an uneventful follow-up by sequential digital dermoscopy for at least 2 years. RESULTS: A total of 281 malignant lesions and 1700 benign lesions from 435 patients (62.2% male, mean age: 52 years) were prospectively imaged. The CNN showed a sensitivity of 77.6% (95% confidence interval [CI]: [72.4%-82.1%]), specificity of 95.3% (95% CI: [94.2%-96.2%]), and receiver operating characteristic (ROC)-area under the curve (AUC) of 0.945 (95% CI: [0.930-0.961]). In contrast, the CIA achieved a sensitivity of 53.4% (95% CI: [47.5%-59.1%]), specificity of 86.6% (95% CI: [84.9%-88.1%]) and ROC-AUC of 0.738 (95% CI: [0.701-0.774]). The data set included melanomas originally diagnosed by dynamic changes during sequential digital dermoscopy (52 of 201, 20.6%), which reduced the sensitivities of both classifiers. Pairwise comparisons of sensitivities, specificities, and ROC-AUCs indicated a clear outperformance by the CNN (all p < 0.001). CONCLUSIONS: The superior diagnostic performance of the CNN argues against a continued application of former CIAs as an aide to physicians' clinical management decisions.

Melanoma recognition by a deep learning convolutional neural network-Performance in different melanoma subtypes and localisations.

BACKGROUND: Deep learning convolutional neural networks (CNNs) show great potential for melanoma diagnosis. Melanoma thickness at diagnosis among others depends on melanoma localisation and subtype (e.g. advanced thickness in acrolentiginous or nodular melanomas). The question whether CNN may counterbalance physicians' diagnostic difficulties in these melanomas has not been addressed. We aimed to investigate the diagnostic performance of a CNN with approval for the European market across different melanoma localisations and subtypes. METHODS: The current market version of a CNN (Moleanalyzer-Pro®, FotoFinder Systems GmbH, Bad Birnbach, Germany) was used for classifications (malignant/benign) in six dermoscopic image sets. Each set included 30 melanomas and 100 benign lesions of related localisations and morphology (set-SSM: superficial spreading melanomas and macular nevi; set-LMM: lentigo maligna melanomas and facial solar lentigines/seborrhoeic keratoses/nevi; set-NM: nodular melanomas and papillomatous/dermal/blue nevi; set-Mucosa: mucosal melanomas and mucosal melanoses/macules/nevi; set-AMskin: acrolentiginous melanomas and acral (congenital) nevi; set-AMnail: subungual melanomas and subungual (congenital) nevi/lentigines/ethnical type pigmentations). RESULTS: The CNN showed a high-level performance in set-SSM, set-NM and set-LMM (sensitivities >93.3%, specificities >65%, receiver operating characteristics-area under the curve [ROC-AUC] >0.926). In set-AMskin, the sensitivity was lower (83.3%) at a high specificity (91.0%) and ROC-AUC (0.928). A limited performance was found in set-mucosa (sensitivity 93.3%, specificity 38.0%, ROC-AUC 0.754) and set-AMnail (sensitivity 53.3%, specificity 68.0%, ROC-AUC 0.621). CONCLUSIONS: The CNN may help to partly counterbalance reduced human accuracies. However, physicians need to be aware of the CNN's limited diagnostic performance in mucosal and subungual lesions. Improvements may be expected from additional training images of mucosal and subungual sites.

Photoaging smartphone app promoting poster campaign to reduce smoking prevalence in secondary schools: the Smokerface Randomized Trial: design and baseline characteristics.

INTRODUCTION: Smoking is the largest cause of preventable death globally. Most smokers smoke their first cigarette in early adolescence. We took advantage of the widespread availability of mobile phones and adolescents' interest in appearance to develop a free photoaging app which is promoted via a poster campaign in secondary schools. This study aims to evaluate its effectiveness regarding smoking prevalence and students' attitudes towards smoking. METHODS AND ANALYSIS: A randomised controlled trial is conducted with 9851 students of both genders with an average age of 12 years in grades 6 and 7 of 126 secondary schools in Germany. At present, cigarette smoking prevalence in our sample is 4.7%, with 4.6% of the students currently using e-cigarettes (1.6% use both). The prospective experimental study design includes measurements at baseline and at 6, 12 and 24 months postintervention via a questionnaire plus a random cotinine saliva sample at 24 months postintervention. The study groups consist of randomised schools receiving the Smokerface poster campaign and control schools with comparable baseline data (no intervention). The primary end point is the difference of change in smoking prevalence in the intervention group versus the difference in the control group at 24 months follow-up. Longitudinal changes in smoking-related attitudes, the number of new smokers and quitters and the change in the number of never-smokers will be compared between the two groups as secondary outcomes. ETHICS AND DISSEMINATION: Ethical approval was obtained from the ethics committee of the University of Gießen and the ministries of cultural affairs, both in Germany. Results will be disseminated at conferences, in peer-reviewed journals, on our websites and throughout the multinational Education Against Tobacco network. TRIAL REGISTRATION NUMBER: NCT02544360, Pre-results.

Small-Molecule Stabilization of the 14-3-3/Gab2 Protein-Protein Interaction (PPI) Interface.

Small-molecule modulation of protein-protein interactions (PPIs) is one of the most promising new areas in drug discovery. In the vast majority of cases only inhibition or disruption of PPIs is realized, whereas the complementary strategy of targeted stabilization of PPIs is clearly under-represented. Here, we report the example of a semi-synthetic natural product derivative--ISIR-005--that stabilizes the cancer-relevant interaction of the adaptor protein 14-3-3 and Gab2. The crystal structure of ISIR-005 in complex with 14-3-3 and the binding motif of Gab2 comprising two phosphorylation sites (Gab2pS210pT391) showed how the stabilizing molecule binds to the rim-of-the-interface of the protein complex. Only in the direct vicinity of 14-3-3/Gab2pT391 site is a pre-formed pocket occupied by ISIR-005; binding of the Gab2pS210 motif to 14-3-3 does not create an interface pocket suitable for the molecule. Accordingly, ISIR-005 only stabilizes the binding of the Gab2pT391 but not the Gab2pS210 site. This study represents structural and biochemical proof of the druggability of the 14-3-3/Gab2 PPI interface with important implications for the development of PPI stabilizers.