Assessment of Diagnostic Performance of Dermatologists Cooperating With a Convolutional Neural Network in a Prospective Clinical Study: Human With Machine

Artificial intelligence and machine learning in dermatological research and healthcare: British Society for Investigative Dermatology Skin Club Report, Southampton, April 2024

The British Society of Investigative Dermatology is the annual meeting of the UK’s skin research community. At this year’s meeting in Southampton, there was a discussion on the history and potential of artificial intelligence (‘AI’) in health care.

The four experts who spoke at the meeting have summarized their lectures in this paper. There is a piece on Alan Turing, who proposed the ‘Turing Test’ to find out if a machine could think like a human. Concepts like ‘machine learning’ (a key tool in AI) are explained. Next, there is a piece on the challenges of using AI decision-making tools in the skin cancer pathway. We discuss AI/machine learning approaches to grouping patients and choosing the best treatments for people with ‘atopic dermatitis’ (or ‘eczema’). Finally, potential pitfalls in AI are highlighted, including the need to explain how AI makes decisions and approaches to achieving this.

There is much excitement about AI, and this paper captures the discussion from the meeting of the current state of AI in dermatology health care.

Patient and dermatologists' perspectives on augmented intelligence for melanoma screening: A prospective study

BACKGROUND

Artificial intelligence (AI) shows promising potential to enhance human decision-making as synergistic augmented intelligence (AuI), but requires critical evaluation for skin cancer screening in a real-world setting.

OBJECTIVES

To investigate the perspectives of patients and dermatologists after skin cancer screening by human, artificial and augmented intelligence.

METHODS

A prospective comparative cohort study conducted at the University Hospital Basel included 205 patients (at high-risk of developing melanoma, with resected or advanced disease) and 8 dermatologists. Patients underwent skin cancer screening by a dermatologist with subsequent 2D and 3D total-body photography (TBP). Any suspicious and all melanocytic skin lesions ≥3 mm were imaged with digital dermoscopes and classified by corresponding convolutional neural networks (CNNs). Excisions were performed based on dermatologist's melanoma suspicion, study-defined elevated CNN risk-scores and/or melanoma suspicion by AuI. Subsequently, all patients and dermatologists were surveyed about their experience using questionnaires, including quantification of patient's safety sense following different examinations (subjective safety score (SSS): 0-10).

RESULTS

Most patients believed AI could improve diagnostic performance (95.5%, n = 192/201). In total, 83.4% preferred AuI-based skin cancer screening compared to examination by AI or dermatologist alone (3D-TBP: 61.3%; 2D-TBP: 22.1%, n = 199). Regarding SSS, AuI induced a significantly higher feeling of safety than AI (mean-SSS (mSSS): 9.5 vs. 7.7, p < 0.0001) or dermatologist screening alone (mSSS: 9.5 vs. 9.1, p = 0.001). Most dermatologists expressed high trust in AI examination results (3D-TBP: 90.2%; 2D-TBP: 96.1%, n = 205). In 68.3% of the examinations, dermatologists felt that diagnostic accuracy improved through additional AI-assessment (n = 140/205). Especially beginners (<2 years' dermoscopic experience; 61.8%, n = 94/152) felt AI facilitated their clinical work compared to experts (>5 years' dermoscopic experience; 20.9%, n = 9/43). Contrarily, in divergent risk assessments, only 1.5% of dermatologists trusted a benign CNN-classification more than personal malignancy suspicion (n = 3/205).

CONCLUSIONS

While patients already prefer AuI with 3D-TBP for melanoma recognition, dermatologists continue to rely largely on their own decision-making despite high confidence in AI-results.

TRIAL REGISTRATION

ClinicalTrials.gov (NCT04605822).

European consensus-based interdisciplinary guideline for melanoma. Part 1: Diagnostics - Update 2024

This guideline was developed in close collaboration with multidisciplinary experts from the European Association of Dermato-Oncology (EADO), the European Dermatology Forum (EDF) and the European Organization for Research and Treatment of Cancer (EORTC). Recommendations for the diagnosis and treatment of melanoma were developed on the basis of systematic literature research and consensus conferences. Cutaneous melanoma (CM) is the most dangerous form of skin tumor and accounts for 90 % of skin cancer mortality. The diagnosis of melanoma can be made clinically and must always be confirmed by dermoscopy. If melanoma is suspected, a histopathological examination is always required. Sequential digital dermoscopy and whole-body photography can be used in high-risk patients to improve the detection of early-stage melanoma. If available, confocal reflectance microscopy can also improve the clinical diagnosis in special cases. Melanoma is classified according to the 8th version of the American Joint Committee on Cancer classification. For thin melanomas up to a tumor thickness of 0.8 mm, no further diagnostic imaging is required. From stage IB, lymph node sonography is recommended, but no further imaging examinations. From stage IIB/C, whole-body examinations with computed tomography or positron emission tomography CT in combination with magnetic resonance imaging of the brain are recommended. From stage IIB/C and higher, a mutation test is recommended, especially for the BRAF V600 mutation. It is important to perform a structured follow-up to detect relapses and secondary primary melanomas as early as possible. A stage-based follow-up regimen is proposed, which in the experience of the guideline group covers the optimal requirements, although further studies may be considered. This guideline is valid until the end of 2026.

Image-Based Artificial Intelligence in Psoriasis Assessment: The Beginning of a New Diagnostic Era?

Psoriasis, a chronic inflammatory skin disease, affects millions of people worldwide. It imposes a significant burden on patients' quality of life and healthcare systems, creating an urgent need for optimized diagnosis, treatment, and management. In recent years, image-based artificial intelligence (AI) applications have emerged as promising tools to assist physicians by offering improved accuracy and efficiency. In this review, we provide an overview of the current landscape of image-based AI applications in psoriasis. Emphasis is placed on machine learning (ML) algorithms, a key subset of AI, which enable automated pattern recognition for various tasks. Key AI applications in psoriasis include lesion detection and segmentation, differentiation from other skin conditions, subtype identification, automated area involvement, and severity scoring, as well as personalized treatment selection and response prediction. Furthermore, we discuss two commercially available systems that utilize standardized photo documentation, automated segmentation, and semi-automated Psoriasis Area and Severity Index (PASI) calculation for patient assessment and follow-up. Despite the promise of AI in this field, many challenges remain. These include the validation of current models, integration into clinical workflows, the current lack of diversity in training-set data, and the need for standardized imaging protocols. Addressing these issues is crucial for the successful implementation of AI technologies in clinical practice. Overall, we underscore the potential of AI to revolutionize psoriasis management, highlighting both the advancements and the hurdles that need to be overcome. As technology continues to evolve, AI is expected to significantly improve the accuracy, efficiency, and personalization of psoriasis treatment.

Skin Cancer Image Classification Using Artificial Intelligence Strategies: A Systematic Review

The increasing incidence of and resulting deaths associated with malignant skin tumors are a public health problem that can be minimized if detection strategies are improved. Currently, diagnosis is heavily based on physicians' judgment and experience, which can occasionally lead to the worsening of the lesion or needless biopsies. Several non-invasive imaging modalities, e.g., confocal scanning laser microscopy or multiphoton laser scanning microscopy, have been explored for skin cancer assessment, which have been aligned with different artificial intelligence (AI) strategies to assist in the diagnostic task, based on several image features, thus making the process more reliable and faster. This systematic review concerns the implementation of AI methods for skin tumor classification with different imaging modalities, following the PRISMA guidelines. In total, 206 records were retrieved and qualitatively analyzed. Diagnostic potential was found for several techniques, particularly for dermoscopy images, with strategies yielding classification results close to perfection. Learning approaches based on support vector machines and artificial neural networks seem to be preferred, with a recent focus on convolutional neural networks. Still, detailed descriptions of training/testing conditions are lacking in some reports, hampering reproduction. The use of AI methods in skin cancer diagnosis is an expanding field, with future work aiming to construct optimal learning approaches and strategies. Ultimately, early detection could be optimized, improving patient outcomes, even in areas where healthcare is scarce.

Computerizing the first step of the two-step algorithm in dermoscopy: A convolutional neural network for differentiating melanocytic from non-melanocytic skin lesions

IMPORTANCE

Convolutional neural networks (CNN) have shown performance equal to trained dermatologists in differentiating benign from malignant skin lesions. To improve clinicians' management decisions, additional classifications into diagnostic categories might be helpful.

METHODS

A convenience sample of 100 pigmented/non-pigmented skin lesions was used for a cross-sectional two-level reader study including 96 dermatologists (level I: dermoscopy only; level II: clinical close-up images, dermoscopy, and textual information). Dermoscopic images were classified by a binary CNN trained to differentiate melanocytic from non-melanocytic lesions (FotoFinder Systems, Bad Birnbach, Germany). Primary endpoint was the accuracy of the CNN's classification in comparison with dermatologists reviewing level-II information. Secondary endpoints included dermatologists' accuracies according to their level of experience and the CNN's area under the curve (AUC) of receiver operating characteristics (ROC).

RESULTS

The CNN revealed an accuracy and ROC AUC with corresponding 95 % confidence intervals (CI) of 91.0 % (83.8 % to 95.2 %) and 0.981 (0.962 to 1). In level I, dermatologists showed a mean accuracy of 83.7 % (82.5 % to 84.8 %). With level II information, the accuracy improved to 87.8 % (86.7 % to 88.9 %; p < 0.001). When comparing accuracies of CNN and dermatologists in level II, the CNN's accuracy was higher (91.0 % versus 87.8 %, p < 0.001). For experts with level II information results were on par with the CNN (91.0 % versus 90.4 %, p = 0.368).

CONCLUSIONS

The tested CNN accurately differentiated melanocytic from non-melanocytic skin lesions and outperformed dermatologists. The CNN may support clinicians and could be used in an ensemble approach combined with other CNN models.

Skin lesion segmentation using deep learning algorithm with ant colony optimization

BACKGROUND

Segmentation of skin lesions remains essential in histological diagnosis and skin cancer surveillance. Recent advances in deep learning have paved the way for greater improvements in medical imaging. The Hybrid Residual Networks (ResUNet) model, supplemented with Ant Colony Optimization (ACO), represents the synergy of these improvements aimed at improving the efficiency and effectiveness of skin lesion diagnosis.

OBJECTIVE

This paper seeks to evaluate the effectiveness of the Hybrid ResUNet model for skin lesion classification and assess its impact on optimizing ACO performance to bridge the gap between computational efficiency and clinical utility.

METHODS

The study used a deep learning design on a complex dataset that included a variety of skin lesions. The method includes training a Hybrid ResUNet model with standard parameters and fine-tuning using ACO for hyperparameter optimization. Performance was evaluated using traditional metrics such as accuracy, dice coefficient, and Jaccard index compared with existing models such as residual network (ResNet) and U-Net.

RESULTS

The proposed hybrid ResUNet model exhibited excellent classification accuracy, reflected in the noticeable improvement in all evaluated metrics. His ability to describe complex lesions was particularly outstanding, improving diagnostic accuracy. Our experimental results demonstrate that the proposed Hybrid ResUNet model outperforms existing state-of-the-art methods, achieving an accuracy of 95.8%, a Dice coefficient of 93.1%, and a Jaccard index of 87.5.

CONCLUSION

The addition of ResUNet to ACO in the proposed Hybrid ResUNet model significantly improves the classification of skin lesions. This integration goes beyond traditional paradigms and demonstrates a viable strategy for deploying AI-powered tools in clinical settings.

FUTURE WORK

Future investigations will focus on increasing the version's abilities by using multi-modal imaging information, experimenting with alternative optimization algorithms, and comparing real-world medical applicability. There is also a promising scope for enhancing computational performance and exploring the model's interpretability for more clinical adoption.

Prospective multicenter study using artificial intelligence to improve dermoscopic melanoma diagnosis in patient care

BACKGROUND

Early detection of melanoma, a potentially lethal type of skin cancer with high prevalence worldwide, improves patient prognosis. In retrospective studies, artificial intelligence (AI) has proven to be helpful for enhancing melanoma detection. However, there are few prospective studies confirming these promising results. Existing studies are limited by low sample sizes, too homogenous datasets, or lack of inclusion of rare melanoma subtypes, preventing a fair and thorough evaluation of AI and its generalizability, a crucial aspect for its application in the clinical setting.

METHODS

Therefore, we assessed "All Data are Ext" (ADAE), an established open-source ensemble algorithm for detecting melanomas, by comparing its diagnostic accuracy to that of dermatologists on a prospectively collected, external, heterogeneous test set comprising eight distinct hospitals, four different camera setups, rare melanoma subtypes, and special anatomical sites. We advanced the algorithm with real test-time augmentation (R-TTA, i.e., providing real photographs of lesions taken from multiple angles and averaging the predictions), and evaluated its generalization capabilities.

RESULTS

Overall, the AI shows higher balanced accuracy than dermatologists (0.798, 95% confidence interval (CI) 0.779-0.814 vs. 0.781, 95% CI 0.760-0.802; p = 4.0e-145), obtaining a higher sensitivity (0.921, 95% CI 0.900-0.942 vs. 0.734, 95% CI 0.701-0.770; p = 3.3e-165) at the cost of a lower specificity (0.673, 95% CI 0.641-0.702 vs. 0.828, 95% CI 0.804-0.852; p = 3.3e-165).

CONCLUSION

As the algorithm exhibits a significant performance advantage on our heterogeneous dataset exclusively comprising melanoma-suspicious lesions, AI may offer the potential to support dermatologists, particularly in diagnosing challenging cases.

Deep Learning Techniques for the Dermoscopic Differential Diagnosis of Benign/Malignant Melanocytic Skin Lesions: From the Past to the Present

There has been growing scientific interest in the research field of deep learning techniques applied to skin cancer diagnosis in the last decade. Though encouraging data have been globally reported, several discrepancies have been observed in terms of study methodology, result presentations and validation in clinical settings. The present review aimed to screen the scientific literature on the application of DL techniques to dermoscopic melanoma/nevi differential diagnosis and extrapolate those original studies adequately by reporting on a DL model, comparing them among clinicians and/or another DL architecture. The second aim was to examine those studies together according to a standard set of statistical measures, and the third was to provide dermatologists with a comprehensive explanation and definition of the most used artificial intelligence (AI) terms to better/further understand the scientific literature on this topic and, in parallel, to be updated on the newest applications in the medical dermatologic field, along with a historical perspective. After screening nearly 2000 records, a subset of 54 was selected. Comparing the 20 studies reporting on convolutional neural network (CNN)/deep convolutional neural network (DCNN) models, we have a scenario of highly performant DL algorithms, especially in terms of low false positive results, with average values of accuracy (83.99%), sensitivity (77.74%), and specificity (80.61%). Looking at the comparison with diagnoses by clinicians (13 studies), the main difference relies on the specificity values, with a +15.63% increase for the CNN/DCNN models (average specificity of 84.87%) compared to humans (average specificity of 64.24%) with a 14,85% gap in average accuracy; the sensitivity values were comparable (79.77% for DL and 79.78% for humans). To obtain higher diagnostic accuracy and feasibility in clinical practice, rather than in experimental retrospective settings, future DL models should be based on a large dataset integrating dermoscopic images with relevant clinical and anamnestic data that is prospectively tested and adequately compared with physicians.

A Narrative Review: Opportunities and Challenges in Artificial Intelligence Skin Image Analyses Using Total Body Photography

Artificial intelligence (AI) algorithms for skin lesion classification have reported accuracy at par with and even outperformance of expert dermatologists in experimental settings. However, the majority of algorithms do not represent real-world clinical approach where skin phenotype and clinical background information are considered. We review the current state of AI for skin lesion classification and present opportunities and challenges when applied to total body photography (TBP). AI in TBP analysis presents opportunities for intrapatient assessment of skin phenotype and holistic risk assessment by incorporating patient-level metadata, although challenges exist for protecting patient privacy in algorithm development and improving explainable AI methods.

Multicenter prospective blinded melanoma detection study with a handheld elastic scattering spectroscopy device

Background

The elastic scattering spectroscopy (ESS) device (DermaSensor Inc., Miami, FL) is a noninvasive, painless, adjunctive tool for skin cancer detection.

Objectives

To investigate the performance of the ESS device in the detection of melanoma.

Methods

A prospective, investigator-blinded, multicenter study was conducted at 8 United States (US) and 2 Australian sites. All eligible skin lesions were clinically concerning for melanoma, examined with the ESS device, subsequently biopsied according to dermatologists' standard of care, and evaluated with histopathology. A total of 311 participants with 440 lesions were enrolled, including 44 melanomas (63.6% in situ and 36.4% invasive) and 44 severely dysplastic nevi.

Results

The observed sensitivity of the ESS device for melanoma detection was 95.5% (95% CI, 84.5% to 98.8%, 42 of 44 melanomas), and the observed specificity was 32.5% (95% CI, 27.2% to 38.3%). The positive and negative predictive values were 16.0% and 98.1%, respectively.

Limitations

The device was tested in a high-risk population with lesions selected for biopsy based on clinical and dermoscopic assessments of board-certified dermatologists. Most enrolled lesions were pigmented.

Conclusion

The ESS device's high sensitivity and NPV for the detection of melanoma suggest the device may be a useful adjunctive, point-of-care tool for melanoma detection.

Minimum labelling requirements for dermatology artificial intelligence-based Software as Medical Device (SaMD): A consensus statement

BACKGROUND/OBJECTIVES

Artificial intelligence (AI) holds remarkable potential to improve care delivery in dermatology. End users (health professionals and general public) of AI-based Software as Medical Devices (SaMD) require relevant labelling information to ensure that these devices can be used appropriately. Currently, there are no clear minimum labelling requirements for dermatology AI-based SaMDs.

METHODS

Common labelling recommendations for AI-based SaMD identified in a recent literature review were evaluated by an Australian expert panel in digital health and dermatology via a modified Delphi consensus process. A nine-point Likert scale was used to indicate importance of 10 items, and voting was conducted to determine the specific characteristics to include for some items. Consensus was achieved when more than 75% of the experts agreed that inclusion of information was necessary.

RESULTS

There was robust consensus supporting inclusion of all proposed items as minimum labelling requirements; indication for use, intended user, training and test data sets, algorithm design, image processing techniques, clinical validation, performance metrics, limitations, updates and adverse events. Nearly all suggested characteristics of the labelling items received endorsement, except for some characteristics related to performance metrics. Moreover, there was consensus that uniform labelling criteria should apply across all AI categories and risk classes set out by the Therapeutic Goods Administration.

CONCLUSIONS

This study provides critical evidence for setting labelling standards by the Therapeutic Goods Administration to safeguard patients, health professionals, consumers, industry, and regulatory bodies from AI-based dermatology SaMDs that do not currently provide adequate information about how they were developed and tested.

Performance of an automated total body mapping algorithm to detect melanocytic lesions of clinical relevance

IMPORTANCE

Total body photography for skin cancer screening is a well-established tool allowing documentation and follow-up of the entire skin surface. Artificial intelligence-based systems are increasingly applied for automated lesion detection and diagnosis.

DESIGN AND PATIENTS

In this prospective observational international multicentre study experienced dermatologists performed skin cancer screenings and identified clinically relevant melanocytic lesions (CRML, requiring biopsy or observation). Additionally, patients received 2D automated total body mapping (ATBM) with automated lesion detection (ATBM master, Fotofinder Systems GmbH). Primary endpoint was the percentage of CRML detected by the bodyscan software. Secondary endpoints included the percentage of correctly identified "new" and "changed" lesions during follow-up examinations.

RESULTS

At baseline, dermatologists identified 1075 CRML in 236 patients and 999 CRML (92.9%) were also detected by the automated software. During follow-up examinations dermatologists identified 334 CRMLs in 55 patients, with 323 (96.7%) also being detected by ATBM with automated lesions detection. Moreover, all new (n = 13) or changed CRML (n = 24) during follow-up were detected by the software. Average time requirements per baseline examination was 14.1 min (95% CI [12.8-15.5]). Subgroup analysis of undetected lesions revealed either technical (e.g. covering by clothing, hair) or lesion-specific reasons (e.g. hypopigmentation, palmoplantar sites).

CONCLUSIONS

ATBM with lesion detection software correctly detected the vast majority of CRML and new or changed CRML during follow-up examinations in a favourable amount of time. Our prospective international study underlines that automated lesion detection in TBP images is feasible, which is of relevance for developing AI-based skin cancer screenings.

Artificial intelligence in the detection of skin cancer: State of the art

The incidence of melanoma is increasing rapidly. This cancer has a good prognosis if detected early. For this reason, various systems of skin lesion image analysis, which support imaging diagnostics of this neoplasm, are developing very dynamically. To detect and recognize neoplastic lesions, such systems use various artificial intelligence (AI) algorithms. This area of computer science applications has recently undergone dynamic development, abounding in several solutions that are effective tools supporting diagnosticians in many medical specialties. In this contribution, a number of applications of different classes of AI algorithms for the detection of this skin melanoma are presented and evaluated. Both classic systems based on the analysis of dermatoscopic images as well as total body systems, enabling the analysis of the patient's whole body to detect moles and pathologic changes, are discussed. These increasingly popular applications that allow the analysis of lesion images using smartphones are also described. The quantitative evaluation of the discussed systems with particular emphasis on the method of validation of the implemented algorithms is presented. The advantages and limitations of AI in the analysis of lesion images are also discussed, and problems requiring a solution for more effective use of AI in dermatology are identified.

Performance of Commercial Dermatoscopic Systems That Incorporate Artificial Intelligence for the Identification of Melanoma in General Practice: A Systematic Review

BACKGROUND

Cutaneous melanoma remains an increasing global public health burden, particularly in fair-skinned populations. Advancing technologies, particularly artificial intelligence (AI), may provide an additional tool for clinicians to help detect malignancies with a more accurate success rate. This systematic review aimed to report the performance metrics of commercially available convolutional neural networks (CNNs) tasked with detecting MM.

METHODS

A systematic literature search was performed using CINAHL, Medline, Scopus, ScienceDirect and Web of Science databases.

RESULTS

A total of 16 articles reporting MM were included in this review. The combined number of melanomas detected was 1160, and non-melanoma lesions were 33,010. The performance of market-approved technology and clinician performance for classifying melanoma was highly heterogeneous, with sensitivity ranging from 16.4 to 100.0%, specificity between 40.0 and 98.3% and accuracy between 44.0 and 92.0%. Less heterogeneity was observed when clinicians worked in unison with AI, with sensitivity ranging between 83.3 and 100.0%, specificity between 83.7 and 87.3%, and accuracy between 86.4 and 86.9%.

CONCLUSION

Instead of focusing on the performance of AI versus clinicians for classifying melanoma, more consistent performance has been obtained when clinicians' work is supported by AI, facilitating management decisions and improving health outcomes.

Artificial intelligence and skin cancer

Artificial intelligence is poised to rapidly reshape many fields, including that of skin cancer screening and diagnosis, both as a disruptive and assistive technology. Together with the collection and availability of large medical data sets, artificial intelligence will become a powerful tool that can be leveraged by physicians in their diagnoses and treatment plans for patients. This comprehensive review focuses on current progress toward AI applications for patients, primary care providers, dermatologists, and dermatopathologists, explores the diverse applications of image and molecular processing for skin cancer, and highlights AI's potential for patient self-screening and improving diagnostic accuracy for non-dermatologists. We additionally delve into the challenges and barriers to clinical implementation, paths forward for implementation and areas of active research.

Artificial Intelligence in Skin Cancer Diagnosis: A Reality Check

The field of skin cancer detection offers a compelling use case for the application of artificial intelligence (AI) within the realm of image-based diagnostic medicine. Through the analysis of large datasets, AI algorithms have the capacity to classify clinical or dermoscopic images with remarkable accuracy. Although these AI-based applications can operate both autonomously and under human supervision, the best results are achieved through a collaborative approach that leverages the expertise of both AI and human experts. However, it is important to note that most studies focus on assessing the diagnostic accuracy of AI in artificial settings rather than in real-world scenarios. Consequently, the practical utility of AI-assisted diagnosis in a clinical environment is still largely unknown. Furthermore, there exists a knowledge gap concerning the optimal use cases and deployment settings for these AI systems as well as the practical challenges that may arise from widespread implementation. This review explores the advantages and limitations of AI in a variety of real-world contexts, with a specific focus on its value to consumers, general practitioners, and dermatologists.

Augmented Decision-Making in wound Care: Evaluating the clinical utility of a Deep-Learning model for pressure injury staging

BACKGROUND

Precise categorization of pressure injury (PI) stages is critical in determining the appropriate treatment for wound care. However, the expertise necessary for PI staging is frequently unavailable in residential care settings.

OBJECTIVE

This study aimed to develop a convolutional neural network (CNN) model for classifying PIs and investigate whether its implementation can allow physicians to make better decisions for PI staging.

METHODS

Using 3,098 clinical images (2,614 and 484 from internal and external datasets, respectively), a CNN was trained and validated to classify PIs and other related dermatoses. A two-part survey was conducted with 24 dermatology residents, ward nurses, and medical students to determine whether the implementation of the CNN improved initial PI classification decisions.

RESULTS

The top-1 accuracy of the model was 0.793 (95% confidence interval [CI], 0.778-0.808) and 0.717 (95% CI, 0.676-0.758) over the internal and external testing sets, respectively. The accuracy of PI staging among participants was 0.501 (95% CI, 0.487-0.515) in Part I, improving by 17.1% to 0.672 (95% CI, 0.660-0.684) in Part II. Furthermore, the concordance between participants increased significantly with the use of the CNN model, with Fleiss' κ of 0.414 (95% CI, 0.410-0.417) and 0.641 (95% CI, 0.638-0.644) in Parts I and II, respectively.

CONCLUSIONS

The proposed CNN model can help classify PIs and relevant dermatoses. In addition, augmented decision-making can improve consultation accuracy while ensuring concordance between the clinical decisions made by a diverse group of health professionals.

Diagnostic performance of augmented intelligence with 2D and 3D total body photography and convolutional neural networks in a high-risk population for melanoma under real-world conditions: A new era of skin cancer screening?

BACKGROUND

Convolutional neural networks (CNNs) have outperformed dermatologists in classifying pigmented skin lesions under artificial conditions. We investigated, for the first time, the performance of three-dimensional (3D) and two-dimensional (2D) CNNs and dermatologists in the early detection of melanoma in a real-world setting.

METHODS

In this prospective study, 1690 melanocytic lesions in 143 patients with high-risk criteria for melanoma were evaluated by dermatologists, 2D-FotoFinder-ATBM and 3D-Vectra WB360 total body photography (TBP). Excision was based on the dermatologists' dichotomous decision, an elevated CNN risk score (study-specific malignancy cut-off: FotoFinder >0.5, Vectra >5.0) and/or the second dermatologist's assessment with CNN support. The diagnostic accuracy of the 2D and 3D CNN classification was compared with that of the dermatologists and the augmented intelligence based on histopathology and dermatologists' assessment. Secondary end-points included reproducibility of risk scores and naevus counts per patient by medical staff (gold standard) compared to automated 3D and 2D TBP CNN counts.

RESULTS

The sensitivity, specificity, and receiver operating characteristics area under the curve (ROC-AUC) for risk-score-assessments compared to histopathology of 3D-CNN with 95% confidence intervals (CI) were 90.0%, 64.6% and 0.92 (CI 0.85-1.00), respectively. While dermatologists and augmented intelligence achieved the same sensitivity (90%) and comparable classification ROC-AUC (0.91 [CI 0.80-1.00], 0.88 [CI 0.77-1.00]) with 3D-CNN, their specificity was superior (92.3% and 86.2%, respectively). The 2D-CNN (sensitivity: 70%, specificity: 40%, ROC-AUC: 0.68 [CI 0.46-0.90]) was outperformed by 3D CNN and dermatologists. The 3D-CNN showed a higher correlation coefficient for repeated measurements of 246 lesions (R = 0.89) than the 2D-CNN (R = 0.79). The mean naevus count per patient varied significantly (gold standard: 210 lesions; 3D-CNN: 469; 2D-CNN: 1324; p < 0.0001).

CONCLUSIONS

Our study emphasises the importance of validating the classification of CNNs in real life. The novel 3D-CNN device outperformed the 2D-CNN and achieved comparable sensitivity with dermatologists. The low specificity of CNNs and the lack of automated counting of TBP nevi currently limit the use of augmented intelligence in clinical practice.

Accuracy of a Smartphone-Based Artificial Intelligence Application for Classification of Melanomas, Melanocytic Nevi, and Seborrheic Keratoses

Current artificial intelligence algorithms can classify melanomas at a level equivalent to that of experienced dermatologists. The objective of this study was to assess the accuracy of a smartphone-based "You Only Look Once" neural network model for the classification of melanomas, melanocytic nevi, and seborrheic keratoses. The algorithm was trained using 59,090 dermatoscopic images. Testing was performed on histologically confirmed lesions: 32 melanomas, 35 melanocytic nevi, and 33 seborrheic keratoses. The results of the algorithm's decisions were compared with those of two skilled dermatologists and five beginners in dermatoscopy. The algorithm's sensitivity and specificity for melanomas were 0.88 (0.71-0.96) and 0.87 (0.76-0.94), respectively. The algorithm surpassed the beginner dermatologists, who achieved a sensitivity of 0.83 (0.77-0.87). For melanocytic nevi, the algorithm outclassed each group of dermatologists, attaining a sensitivity of 0.77 (0.60-0.90). The algorithm's sensitivity for seborrheic keratoses was 0.52 (0.34-0.69). The smartphone-based "You Only Look Once" neural network model achieved a high sensitivity and specificity in the classification of melanomas and melanocytic nevi with an accuracy similar to that of skilled dermatologists. However, a bigger dataset is required in order to increase the algorithm's sensitivity for seborrheic keratoses.