Abstract:Melanoma, the deadliest form of skin cancer, has seen a steady increase in incidence rates worldwide, posing a significant challenge to dermatologists. Early detection is crucial for improving patient survival rates. However, performing total body screening (TBS), i.e., identifying suspicious lesions or ugly ducklings (UDs) by visual inspection, can be challenging and often requires sound expertise in pigmented lesions. To assist users of varying expertise levels, an artificial intelligence (AI) decision support tool was developed. Our solution identifies and characterizes UDs from real-world wide-field patient images. It employs a state-of-the-art object detection algorithm to locate and isolate all skin lesions present in a patient's total body images. These lesions are then sorted based on their level of suspiciousness using a self-supervised AI approach, tailored to the specific context of the patient under examination. A clinical validation study was conducted to evaluate the tool's performance. The results demonstrated an average sensitivity of 95% for the top-10 AI-identified UDs on skin lesions selected by the majority of experts in pigmented skin lesions. The study also found that the tool increased dermatologists' confidence when formulating a diagnosis, and the average majority agreement with the top-10 AI-identified UDs reached 100% when assisted by our tool. With the development of this AI-based decision support tool, we aim to address the shortage of specialists, enable faster consultation times for patients, and demonstrate the impact and usability of AI-assisted screening. Future developments will include expanding the dataset to include histologically confirmed melanoma and validating the tool for additional body regions.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges in the early detection of malignant melanoma. Specifically, the paper focuses on the following aspects: 1. **Improving the early detection rate**: Malignant melanoma is a deadly skin cancer. If it can be detected early and treated in a timely manner, the patient's survival rate will be significantly improved. However, identifying suspicious lesions or "Ugly Ducklings (UDs)" during the Total Body Screening (TBS) is a complex and time - consuming task that requires extensive professional knowledge. 2. **Reducing the dependence on specialized doctors**: Currently, accurate diagnosis of skin cancer requires specialized skills and experience, which are usually only possessed by dermatologists who have received special training. However, due to the insufficient number of dermatologists, especially in some regions, this has led to problems such as long waiting times for patients and diagnosis delays. 3. **Developing reliable auxiliary tools**: In order to assist professionals at different levels in conducting total body examinations, the paper proposes an artificial intelligence (AI) decision - support tool based on self - supervised learning. This tool aims to automatically identify and classify suspicious lesions from the total body examination images of patients, thereby improving the accuracy and efficiency of diagnosis. ### Main contributions 1. **Using real - world data**: The study used total body examination images (RWD TBS) that were standardly collected during routine consultations in the dermatology clinic of the University Hospital Zurich for training, validating, and testing the model. 2. **Self - supervised learning method**: A self - supervised learning method based on the context of a single patient was proposed for feature representation and the identification of "Ugly Duckling" signs. 3. **Clinical verification**: Through clinical verification of multiple groups of participants, the accuracy of the algorithm's prediction and its impact as an AI - assisted screening tool were evaluated. ### Method overview 1. **Data collection and pre - processing**: - Images were collected using a high - resolution polarized whole - body imaging device (FotoFinder© ATBM). - The images were pre - processed, including segmentation, annotation, and feature extraction. 2. **Skin lesion detector**: - The YOLOR model was used for real - time object detection to identify all skin lesions. - By adjusting the model parameters and data augmentation strategies, the detection accuracy for small objects was improved. 3. **"Ugly Duckling" detector**: - A self - supervised learning method (DINO) was used to generate meaningful feature representations. - By calculating the similarity in the embedding space, the most suspicious lesions (i.e., "Ugly Ducklings") were identified. 4. **Clinical verification**: - Participants with different backgrounds and experience levels (including dermatologists, general practitioners, and non - medical students) were invited to perform the lesion selection task. - The performance of the participants with and without AI assistance was evaluated, including sensitivity, confidence, and the number of selected lesions. ### Results 1. **Skin lesion detector**: - The model achieved an average recall rate of 95% and a precision rate of 75% on four unseen patients. - For the identification of "Ugly Ducklings", the model had a recall rate of 100%. 2. **"Ugly Duckling" detector**: - The results of clinical verification showed that with AI assistance, the average sensitivity of dermatologists increased from 80 - 84% to 92 - 98%, the average sensitivity of students increased from 63% to 81%, and the average sensitivity of general practitioners increased from 68% to 82%. - The average sensitivity of AI prediction reached 95%, and even reached 100% in expert evaluations. In conclusion, through the development and verification of an AI - assisted tool based on self - supervised learning, this paper aims to improve the early detection rate of malignant melanoma, reduce the dependence on specialized doctors, and improve the diagnostic accuracy and confidence of professionals at different levels.

Automatized self-supervised learning for skin lesion screening

Unsupervised sub-segmentation for pigmented skin lesions.

Automated Machine Learning (AutoML) for the Diagnosis of Melanoma Skin Lesions From Consumer-Grade Camera Photos

Clinical Melanoma Diagnosis with Artificial Intelligence: Insights from a Prospective Multicenter Study

[The Rise of Artificial Intelligence - High Prediction Accuracy in Early Detection of Pigmented Melanoma]

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

An Intelligent Decision Support System for Skin Cancer Detection from Dermoscopic Images

Implementation of artificial intelligence algorithms for melanoma screening in a primary care setting

Artificial Intelligence Support for Skin Lesion Triage in Primary Care and Dermatology

Artificial Intelligence Support for Skin Lesion Triage in Primary Care and Dermatology (Preprint)

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

Validation of artificial intelligence prediction models for skin cancer diagnosis using dermoscopy images: the 2019 International Skin Imaging Collaboration Grand Challenge

Computer Aided Diagnosis of Melanoma Using Deep Neural Networks and Game Theory: Application on Dermoscopic Images of Skin Lesions

Intelligent Dermatologist Tool for Classifying Multiple Skin Cancer Subtypes by Incorporating Manifold Radiomics Features Categories

Design and validation of a new machine-learning-based diagnostic tool for the differentiation of dermatoscopic skin cancer images

Melanoma Detection Using Deep Learning-Based Classifications

Automated deep learning approach for classification of malignant melanoma and benign skin lesions

Monitoring of Pigmented Skin Lesions Using 3D Whole Body Imaging

Artificial Intelligence in the detection of skin cancer: state of the art

Skin Cancer Classification Utilizing a Hybrid Model of Machine Learning Models Trained on Dermoscopic Images

Artificial intelligence and skin melanoma