International journal of computer assisted radiology and surgery · 2026
Synthetic skin image generation using a physics-based, object-to-image computational pipeline.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Sizikova Elena, Saharkhiz Niloufar, Kim Andrea, Lago Miguel, Delfino Jana G, Badano Aldo
Аннотация
PURPOSE: We describe an extended overview and evaluation of the S-SYNTH skin simulation approach (Kim et al. in MICCAI, Springer, pp 734-744, 2024), a knowledge-based skin simulation approach to generate synthetic skin images for aiding artificial intelligence (AI) algorithm development.
METHODS: The skin object model allows for controlled variation in skin appearance, including skin color, presence of hair, lesion shape, and blood fraction. Using this approach, we study the effect of model variations on AI models for skin lesion segmentation.
RESULTS: Results obtained using synthetic data follow similar comparative trends as patient dermatologic images and have the potential to mitigate biases and limitations of existing patient datasets including small dataset size and poor representativeness. We also demonstrate a novel use case where S-SYNTH can enhance diffusion model performance for dermatological applications.
CONCLUSION: Synthetic images have the ability to address many limitations of available patient datasets in skin imaging. Simulating datasets allows for tight control over the parameter space and the ability to create rare examples and annotations that are typically missing from patient datasets.
METHODS: The skin object model allows for controlled variation in skin appearance, including skin color, presence of hair, lesion shape, and blood fraction. Using this approach, we study the effect of model variations on AI models for skin lesion segmentation.
RESULTS: Results obtained using synthetic data follow similar comparative trends as patient dermatologic images and have the potential to mitigate biases and limitations of existing patient datasets including small dataset size and poor representativeness. We also demonstrate a novel use case where S-SYNTH can enhance diffusion model performance for dermatological applications.
CONCLUSION: Synthetic images have the ability to address many limitations of available patient datasets in skin imaging. Simulating datasets allows for tight control over the parameter space and the ability to create rare examples and annotations that are typically missing from patient datasets.