A5021471923- Cutaneous Melanoma Detection and Management
- AI in cancer detection
- Genital Health and Disease
- Nonmelanoma Skin Cancer Studies
- Dermatological diseases and infestations
- Machine Learning and Data Classification
- Face recognition and analysis
- 3D Shape Modeling and Analysis
- Contact Dermatitis and Allergies
- Digital Imaging for Blood Diseases
- Urologic and reproductive health conditions
- Forensic Anthropology and Bioarchaeology Studies
- Psoriasis: Treatment and Pathogenesis
- Urological Disorders and Treatments
- Artificial Intelligence in Healthcare and Education
- Image Retrieval and Classification Techniques
- Ethics and Social Impacts of AI
- Domain Adaptation and Few-Shot Learning
- Biomedical Text Mining and Ontologies
- Integrated Circuits and Semiconductor Failure Analysis
- Face and Expression Recognition
- Single-cell and spatial transcriptomics
- Cell Image Analysis Techniques
- Orthopedic Surgery and Rehabilitation
- Mycobacterium research and diagnosis
University Hospital of Basel
2021-2024
University of Basel
2020-2024
This paper documents how an ethically aligned co-design methodology ensures trustworthiness in the early design phase of artificial intelligence (AI) system component for healthcare. The explains decisions made by deep learning networks analyzing images skin lesions. trustworthy AI developed here used a holistic approach rather than static ethical checklist and required multidisciplinary team experts working with designers their managers. Ethical, legal, technical issues potentially arising...
Abstract AI image classification algorithms have shown promising results when applied to skin cancer detection. Most public datasets are comprised of dermoscopic photos and limited by selection bias, lack standardization, lend themselves development that can only be used skilled clinicians. The SLICE-3D (“Skin Lesion Image Crops Extracted from 3D TBP”) dataset described here addresses those concerns contains images over 400,000 distinct lesions seven dermatologic centers around the world....
Abstract Background Deep‐learning convolutional neural networks (CNNs) have outperformed even experienced dermatologists in dermoscopic melanoma detection under controlled conditions. It remains unexplored how real‐world image transformations affect CNN robustness. Objectives To investigate the consistency of risk assessment by two commercially available CNNs to help formulate recommendations for current clinical use. Methods A comparative cohort study was conducted from January July 2022 at...
Abstract Background Vulvar lichen sclerosus (VLS) is a chronic inflammatory skin condition associated with significant impairment of quality life and potential risk malignant transformation. However, diagnosis VLS often delayed due to its variable clinical presentation shame‐related late consultation. Machine learning (ML)‐trained image recognition software could potentially facilitate early VLS. Objective To develop ML‐trained image‐based model for the detection Methods Images both non‐VLS...
Hand eczema (HE) is one of the most frequent dermatoses, known to be both relapsing and remitting. Regular precise evaluation disease severity key for treatment management. Current scoring systems such as hand index (HECSI) suffer from intra- inter-observer variance. We propose an automated system based on deep learning models (DLM) quantify HE lesions' surface determine their anatomical stratification. In this retrospective study, a team 11 experienced dermatologists annotated lesions in...
The exact location of skin lesions is key in clinical dermatology. On one hand, it supports differential diagnosis (DD) since most conditions have specific predilection sites. the other matters for dermatosurgical interventions. In practice, lesion evaluation not well standardized and anatomical descriptions vary or lack altogether. Automated determination could benefit both situations.Establish an automated method to determine regions patient pictures evaluate gain DD performance a deep...
Pustular psoriasis (PP) is one of the most severe and chronic skin conditions. Its treatment difficult, measurements its severity are highly dependent on clinicians' experience. Pustules brown spots main efflorescences disease directly correlate with activity. We propose an automated deep learning model (DLM) to quantify lesions in terms count surface percentage from patient photographs.In this retrospective study, two dermatologists a student labeled 151 photographs PP patients for pustules...
Out-Of-Distribution (OOD) detection is critical to deploy deep learning models in safety-critical applications. However, the inherent hierarchical concept structure of visual data, which instrumental OOD detection, often poorly captured by conventional methods based on Euclidean geometry. This work proposes a metric framework that leverages strengths Hyperbolic geometry for detection. Inspired previous works refine decision boundary data with synthetic outliers, we extend this method space....
The growing demand for accurate and equitable AI models in digital dermatology faces a significant challenge: the lack of diverse, high-quality labeled data. In this work, we investigate potential domain-specific foundation addressing challenge. We utilize self-supervised learning (SSL) techniques to pre-train on dataset over 240,000 dermatological images from public private collections. Our study considers several SSL methods compares resulting against domain-agnostic like those pre-trained...
Most benchmark datasets for computer vision contain irrelevant images, near duplicates, and label errors. Consequently, model performance on these benchmarks may not be an accurate estimate of generalization capabilities. This is a particularly acute concern in medicine where are typically small, stakes high, annotation processes expensive error-prone. In this paper we propose SelfClean, general procedure to clean up image exploiting latent space learned with self-supervision. By relying...
This paper presents a new robust loss function, the T-Loss, for medical image segmentation. The proposed is based on negative log-likelihood of Student-t distribution and can effectively handle outliers in data by controlling its sensitivity with single parameter. parameter updated during backpropagation process, eliminating need additional computation or prior information about level spread noisy labels. Our experiments show that T-Loss outperforms traditional functions terms dice scores...
Benchmark datasets for digital dermatology unwittingly contain inaccuracies that reduce trust in model performance estimates. We propose a resource-efficient data-cleaning protocol to identify issues escaped previous curation. The leverages an existing algorithmic cleaning strategy and is followed by confirmation process terminated intuitive stopping criterion. Based on multiple dermatologists, we remove irrelevant samples near duplicates estimate the percentage of label errors six image...