Computers in biology and medicine · 2025
Facial characteristics description and classification based on 3D images of Fragile X syndrome in a retrospective cohort of young Chinese males.
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.
Chen Jieyi, Du Siyuan, Zhu Yiting, Li Dongyun, Hu Chunchun, Mei Lianni, Zhu Yunqian, Chen Huihui, Wang Sijia, Xu Xiu, Dong Xinran, Zhou Wenhao, Xu Qiong
Аннотация
PURPOSE: Fragile X syndrome (FXS) is a common cause of intellectual disability and autism. FXS presents with abnormal facial features, which in pediatric patients are subtler than what is seen in adults. The three-dimensional (3D) facial images, which contain more stereoscopic and subtle information than two-dimensional (2D) photographs, are increasingly being used to classify genetic syndromes. Here, we used 3D facial images to describe facial features and construct a classification model, especially in male patients with FXS.
METHODS: We registered the 3D facial images of 40 Chinese boys with FXS and 40 healthy boys. We utilized seven machine learning models with different features extracted from dense point cloud and sparse landmarks. A linear regression model was performed between feature reduction of regional point cloud and genomic as well as methylation subtypes.
RESULTS: The typical and subtle differences between 3D average faces of patients and controls could be quantitatively visualized. The projection of patients and controls in Fragile X-liked vectors are significantly different. The random forests model using coordinates of regional facial points (chin, eye, forehead, nose and upper lip) could perform better than expert clinicians in binary classification. Among the 63 hierarchical facial segmentation, significantly associations were found in 8 segments with genetic subtypes, and 2 segments with methylation subtypes.
CONCLUSION: The 3D facial images could assist to distinguish male patients with FXS by machine learning, in which the selected regional features performed better than the global features and sparse landmarks. The genetic and methylation status might affect regional facial features differently.
METHODS: We registered the 3D facial images of 40 Chinese boys with FXS and 40 healthy boys. We utilized seven machine learning models with different features extracted from dense point cloud and sparse landmarks. A linear regression model was performed between feature reduction of regional point cloud and genomic as well as methylation subtypes.
RESULTS: The typical and subtle differences between 3D average faces of patients and controls could be quantitatively visualized. The projection of patients and controls in Fragile X-liked vectors are significantly different. The random forests model using coordinates of regional facial points (chin, eye, forehead, nose and upper lip) could perform better than expert clinicians in binary classification. Among the 63 hierarchical facial segmentation, significantly associations were found in 8 segments with genetic subtypes, and 2 segments with methylation subtypes.
CONCLUSION: The 3D facial images could assist to distinguish male patients with FXS by machine learning, in which the selected regional features performed better than the global features and sparse landmarks. The genetic and methylation status might affect regional facial features differently.