Summary
Background
Over the last few years, several articles on dermoscopy of non‐neoplastic dermatoses have been published, yet there is poor consistency in the terminology among different studies.
...Objectives
We aimed to standardize the dermoscopic terminology and identify basic parameters to evaluate in non‐neoplastic dermatoses through an expert consensus.
Methods
The modified Delphi method was followed, with two phases: (i) identification of a list of possible items based on a systematic literature review and (ii) selection of parameters by a panel of experts through a three‐step iterative procedure (blinded e‐mail interaction in rounds 1 and 3 and a face‐to‐face meeting in round 2). Initial panellists were recruited via e‐mail from all over the world based on their expertise on dermoscopy of non‐neoplastic dermatoses.
Results
Twenty‐four international experts took part in all rounds of the consensus and 13 further international participants were also involved in round 2. Five standardized basic parameters were identified: (i) vessels (including morphology and distribution); (ii) scales (including colour and distribution); (iii) follicular findings; (iv) ‘other structures’ (including colour and morphology); and (v) ‘specific clues’. For each of them, possible variables were selected, with a total of 31 different subitems reaching agreement at the end of the consensus (all of the 29 proposed initially plus two more added in the course of the consensus procedure).
Conclusions
This expert consensus provides a set of standardized basic dermoscopic parameters to follow when evaluating inflammatory, infiltrative and infectious dermatoses. This tool, if adopted by clinicians and researchers in this field, is likely to enhance the reproducibility and comparability of existing and future research findings and uniformly expand the universal knowledge on dermoscopy in general dermatology.
What's already known about this topic?
Over the last few years, several papers have been published attempting to describe the dermoscopic features of non‐neoplastic dermatoses, yet there is poor consistency in the terminology among different studies.
What does this study add?
The present expert consensus provides a set of standardized basic dermoscopic parameters to follow when evaluating inflammatory, infiltrative and infectious dermatoses.
This consensus should enhance the reproducibility and comparability of existing and future research findings and uniformly expand the universal knowledge on dermoscopy in general dermatology.
Linked Editorial: Bahadoran. Br J Dermatol 2020; 182:260–261.
Plain language summary available online
Whether machine-learning algorithms can diagnose all pigmented skin lesions as accurately as human experts is unclear. The aim of this study was to compare the diagnostic accuracy of state-of-the-art ...machine-learning algorithms with human readers for all clinically relevant types of benign and malignant pigmented skin lesions.
For this open, web-based, international, diagnostic study, human readers were asked to diagnose dermatoscopic images selected randomly in 30-image batches from a test set of 1511 images. The diagnoses from human readers were compared with those of 139 algorithms created by 77 machine-learning labs, who participated in the International Skin Imaging Collaboration 2018 challenge and received a training set of 10 015 images in advance. The ground truth of each lesion fell into one of seven predefined disease categories: intraepithelial carcinoma including actinic keratoses and Bowen's disease; basal cell carcinoma; benign keratinocytic lesions including solar lentigo, seborrheic keratosis and lichen planus-like keratosis; dermatofibroma; melanoma; melanocytic nevus; and vascular lesions. The two main outcomes were the differences in the number of correct specific diagnoses per batch between all human readers and the top three algorithms, and between human experts and the top three algorithms.
Between Aug 4, 2018, and Sept 30, 2018, 511 human readers from 63 countries had at least one attempt in the reader study. 283 (55·4%) of 511 human readers were board-certified dermatologists, 118 (23·1%) were dermatology residents, and 83 (16·2%) were general practitioners. When comparing all human readers with all machine-learning algorithms, the algorithms achieved a mean of 2·01 (95% CI 1·97 to 2·04; p<0·0001) more correct diagnoses (17·91 SD 3·42 vs 19·92 4·27). 27 human experts with more than 10 years of experience achieved a mean of 18·78 (SD 3·15) correct answers, compared with 25·43 (1·95) correct answers for the top three machine algorithms (mean difference 6·65, 95% CI 6·06–7·25; p<0·0001). The difference between human experts and the top three algorithms was significantly lower for images in the test set that were collected from sources not included in the training set (human underperformance of 11·4%, 95% CI 9·9–12·9 vs 3·6%, 0·8–6·3; p<0·0001).
State-of-the-art machine-learning classifiers outperformed human experts in the diagnosis of pigmented skin lesions and should have a more important role in clinical practice. However, a possible limitation of these algorithms is their decreased performance for out-of-distribution images, which should be addressed in future research.
None.
Summary
Background
Little is known about the variability of the dermoscopic criteria of squamous cell carcinoma (SCC) according to the histopathological differentiation grade.
Objectives
To evaluate ...whether specific dermoscopic criteria can predict the diagnosis of poorly differentiated SCC compared with well‐ and moderately differentiated SCC.
Methods
Clinical and dermoscopic images of SCCs were retrospectively evaluated for the presence of predefined criteria. Univariate and adjusted odds ratios were calculated. Discriminant functions were used to plot receiver–operator characteristic curves.
Results
Of 143 SCCs included, 48 (33·5%) were well differentiated, 45 (31·5%) were moderately differentiated and 50 (35·0%) were poorly differentiated. Flat tumours had a fourfold increased probability of being poorly differentiated. Dermoscopically, the presence of a predominantly red colour posed a 13‐fold possibility of poor differentiation, whereas a predominantly white and white–yellow colour decreased the odds of poorly differentiated SCC by 97% each. The presence of vessels in more than 50% of the tumour's surface, a diffuse distribution of vessels and bleeding were significantly associated with poor differentiation, while scale/keratin was a potent predictor of well‐ or moderately differentiated tumours.
Conclusions
Dermoscopy may be regarded as a reliable preoperative tool to distinguish poorly from well‐ and moderately differentiated SCC. Given that poor differentiation of SCC represents an independent risk factor for recurrence, metastasis and disease‐specific death, identifying poorly differentiated tumours in vivo may enhance their appropriate management.
What's already known about this topic?
While the dermoscopic criteria of squamous cell carcinoma (SCC) have been well described, little is currently known about the variability of these criteria with respect to the histopathological grade of differentiation in SCC.
What does this study add?
Poorly differentiated SCC is dermoscopically typified by a predominantly red colour, attributed to the presence of bleeding and/or dense vascularity.
Identifying poorly differentiated tumours in vivo may enhance their appropriate management.
Summary
Background The ability to diagnose malignant skin tumours accurately and to distinguish them from benign lesions is vital in ensuring appropriate patient management. Little is known about ...the effects of mobile teledermatology services on diagnostic accuracy and their appropriateness for skin tumour surveillance.
Objectives To evaluate the diagnostic accuracy of clinical and dermoscopic image tele‐evaluation for mobile skin tumour screening.
Methods Over a 3‐month period up to three clinical and dermoscopic images were obtained of 113 skin tumours from 88 patients using a mobile phone camera. Dermoscopic images were taken with a dermatoscope applied to the camera lens. Clinical and dermoscopic images of each lesion together with clinical information were separately teletransmitted for decision‐making. Results were compared with those obtained by face‐to‐face examination and histopathology as the gold standard.
Results A total of 322 clinical and 278 dermoscopic images were acquired; two (1%) clinical and 18 (6%) dermoscopic pictures were inadequate for decision‐making. After excluding inadequate images, the majority of which were dermoscopic pictures, only 104 of the 113 skin tumours from 80 of 88 patients could be tele‐evaluated. Among these 104 lesions, 25 (24%) benign nonmelanocytic, 15 (14%) benign melanocytic, 58 (56%) malignant nonmelanocytic and six (6%) malignant melanocytic lesions were identified. Clinical and dermoscopic tele‐evaluations demonstrated strong concordance with the gold standard (κ = 0·84 for each) and similar high sensitivity and specificity for all diagnostic categories. With regard to the detailed diagnoses, clinical image tele‐evaluation was superior to teledermoscopy resulting in 16 vs. 22 discordant cases.
Conclusions Clinical image tele‐evaluation might be the method of choice for mobile tumour screening.
See also the Commentary by Varma
Background
Deep learning convolutional neural networks (CNN) may assist physicians in the diagnosis of melanoma. The capacity of a CNN to differentiate melanomas from combined naevi, the latter ...representing well‐known melanoma simulators, has not been investigated.
Objective
To assess the diagnostic performance of a CNN when used to differentiate melanomas from combined naevi in comparison with dermatologists.
Methods
In this study, a CNN with regulatory approval for the European market (Moleanalyzer‐Pro, FotoFinder Systems GmbH, Bad Birnbach, Germany) was used. We attained a dichotomous classification (benign, malignant) in dermoscopic images of 36 combined naevi and 36 melanomas with a mean Breslow thickness of 1.3 mm. Primary outcome measures were the CNN's sensitivity, specificity and the diagnostic odds ratio (DOR) in comparison with 11 dermatologists with different levels of experience.
Results
The CNN revealed a sensitivity, specificity and DOR of 97.1% (95% CI 82.7–99.6), 78.8% (95% CI 62.8–89.1.3) and 34 (95% CI 4.8–239), respectively. Dermatologists showed a lower mean sensitivity, specificity and DOR of 90.6% (95% CI 84.1–94.7; P = 0.092), 71.0% (95% CI 62.6–78.1; P = 0.256) and 24 (95% CI 11.6–48.4; P = 0.1114). Under the assumption that dermatologists use the CNN to verify their (initial) melanoma diagnosis, dermatologists achieve an increased specificity of 90.3% (95% CI 79.8–95.6) at an almost unchanged sensitivity. The largest benefit was observed in ‘beginners’, who performed worst without CNN verification (DOR = 12) but best with CNN verification (DOR = 98).
Conclusion
The tested CNN more accurately classified combined naevi and melanomas in comparison with trained dermatologists. Their diagnostic performance could be improved if the CNN was used to confirm/overrule an initial melanoma diagnosis. Application of a CNN may therefore be of benefit to clinicians.
Linked Commentary: M.V. Heppt et al. J Eur Acad Dermatol Venereol 2020; 34: 1134–1135. https://doi.org/10.1111/jdv.16577.
Summary
Background
Dermoscopy is limited in differentiating accurately between pigmented lentigo maligna (LM) and pigmented actinic keratosis (PAK). This might be related to the fact that most ...studies have focused on pigmented criteria only, without considering additional recognizable features.
Objectives
To investigate the diagnostic accuracy of established dermoscopic criteria for pigmented LM and PAK, but including in the evaluation features previously associated with nonpigmented facial actinic keratosis.
Methods
Retrospectively enrolled cases of histopathologically diagnosed LM, PAK and solar lentigo/early seborrhoeic keratosis (SL/SK) were dermoscopically evaluated for the presence of predefined criteria. Univariate and multivariate regression analyses were performed and receiver operating characteristic curves were used.
Results
The study sample consisted of 70 LMs, 56 PAKs and 18 SL/SKs. In a multivariate analysis, the most potent predictors of LM were grey rhomboids (sixfold increased probability of LM), nonevident follicles (fourfold) and intense pigmentation (twofold). In contrast, white circles, scales and red colour were significantly correlated with PAK, posing a 14‐fold, eightfold and fourfold probability for PAK, respectively. The absence of evident follicles also represented a frequent LM criterion, characterizing 71% of LMs.
Conclusions
White and evident follicles, scales and red colour represent significant diagnostic clues for PAK. Conversely, intense pigmentation and grey rhomboidal lines appear highly suggestive of LM.
What's already known about this topic?
Dermoscopy is insufficient to differentiate between lentigo maligna (LM) and pigmented actinic keratosis (PAK).
What does this study add?
White and evident follicles, scales and red colour represent significant diagnostic clues for PAK.
Intense pigmentation and grey rhomboidal lines appear highly suggestive of LM.
These novel findings might improve the early detection of LM, while reducing unnecessary biopsies for PAK.
Convolutional neural networks (CNNs) efficiently differentiate skin lesions by image analysis. Studies comparing a market-approved CNN in a broad range of diagnoses to dermatologists working under ...less artificial conditions are lacking.
One hundred cases of pigmented/non-pigmented skin cancers and benign lesions were used for a two-level reader study in 96 dermatologists (level I: dermoscopy only; level II: clinical close-up images, dermoscopy, and textual information). Additionally, dermoscopic images were classified by a CNN approved for the European market as a medical device (Moleanalyzer Pro, FotoFinder Systems, Bad Birnbach, Germany). Primary endpoints were the sensitivity and specificity of the CNN’s dichotomous classification in comparison with the dermatologists’ management decisions. Secondary endpoints included the dermatologists’ diagnostic decisions, their performance according to their level of experience, and the CNN’s area under the curve (AUC) of receiver operating characteristics (ROC).
The CNN revealed a sensitivity, specificity, and ROC AUC with corresponding 95% confidence intervals (CI) of 95.0% (95% CI 83.5% to 98.6%), 76.7% (95% CI 64.6% to 85.6%), and 0.918 (95% CI 0.866–0.970), respectively. In level I, the dermatologists’ management decisions showed a mean sensitivity and specificity of 89.0% (95% CI 87.4% to 90.6%) and 80.7% (95% CI 78.8% to 82.6%). With level II information, the sensitivity significantly improved to 94.1% (95% CI 93.1% to 95.1%; P < 0.001), while the specificity remained unchanged at 80.4% (95% CI 78.4% to 82.4%; P = 0.97). When fixing the CNN’s specificity at the mean specificity of the dermatologists’ management decision in level II (80.4%), the CNN’s sensitivity was almost equal to that of human raters, at 95% (95% CI 83.5% to 98.6%) versus 94.1% (95% CI 93.1% to 95.1%); P = 0.1. In contrast, dermatologists were outperformed by the CNN in their level I management decisions and level I and II diagnostic decisions. More experienced dermatologists frequently surpassed the CNN’s performance.
Under less artificial conditions and in a broader spectrum of diagnoses, the CNN and most dermatologists performed on the same level. Dermatologists are trained to integrate information from a range of sources rendering comparative studies that are solely based on one single case image inadequate.
•A market-approved convolutional neural network (CNN) trained on dermoscopic images was tested against 96 dermatologists.•Test data included a broad range of skin lesions and was compiled from external sources not involved in CNN training.•Dermatologists indicated their management decisions after reviewing clinical, dermoscopic, and textual case information.•In this setting dermatologists performed on par with the CNN's classifications based on dermoscopic images alone.
Background
Several dermoscopic and in vivo reflectance confocal microscopy (RCM) diagnostic criteria of lentigo maligna (LM)/lentigo maligna melanoma (LMM) have been identified. However, no study ...compared the diagnostic accuracy of these techniques.
Objective
We evaluated the diagnostic accuracy of dermoscopy and RCM for LM/LMM using a holistic assessment of the images.
Methods
A total of 223 facial lesions were evaluated by 21 experts. Diagnostic accuracy of the clinical, dermoscopic and RCM examination was compared. Interinvestigator variability and confidence level in the diagnosis were also evaluated.
Results
Overall diagnostic accuracy of the two imaging techniques was good (area under the curve of the sROC function: 0.89). RCM was more sensitive (80%, vs. 61%) and less specific (81% vs. 92%) than dermoscopy for LM/LMM. In particular, RCM showed a higher sensitivity for hypomelanotic and recurrent LM/LMM. RCM had a higher interinvestigator agreement and a higher confidence level in the diagnosis than dermoscopy.
Conclusion
Reflectance confocal microscopy and dermoscopy are both useful techniques for the diagnosis of facial lesions and in particular LM/LMM. RCM is particularly suitable for the identification of hypomelanotic and recurrent LM/LMM.
Summary
Pigmented Bowen disease (pBD) is an uncommon variant of squamous cell carcinoma in situ. Sometimes it can show clinical and dermoscopic features that are seen in other pigmented lesions of ...the skin and mucosa, making the diagnosis difficult. We report six cases of pBD occurring on the anogenital area, and discuss the importance of dermoscopy for improving the diagnostic accuracy in pBD.
Dermoscopic features of nonpigmented eccrine poroma Conforti, C.; Giuffrida, R.; Seabra Resende, F. S. ...
Clinical and experimental dermatology,
December 2019, 2019-Dec, 2019-12-00, 20191201, Letnik:
44, Številka:
8
Journal Article