Nuclear medicine and molecular imaging · 2016
Intra-patient Variability of FDG Standardized Uptake Values in Mediastinal Blood Pool, Liver, and Myocardium during R-CHOP Chemotherapy in Patients with Diffuse Large B-cell Lymphoma.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
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.
Alsagob Maysoon, Salih Mustafa A, Hamad Muddathir H A, Al-Yafee Yusra, Al-Zahrani Jawaher, Al-Bakheet Albandary, Nester Michael, Sakati Nadia, Wakil Salma M, AlOdaib Ali, Colak Dilek, Kaya Namik
Аннотация
RATIONALE: Periodontal diseases are highly prevalent and largely preventable. The challenge of ensuring sustained adherence to preventive measures, such as mechanical plaque control, remains unresolved despite their strong scientific support. Embedding emerging strategies within the framework of predictive, preventive, and personalised medicine (PPPM) offers a promising path to improve adherence, enable early risk prediction, and tailor interventions. Within this paradigm, digital image biomarkers are increasingly recognised as essential tools for supporting proactive, system-oriented oral health management.
WORKING HYPOTHESIS AND METHODOLOGY: This study hypothesised that a patient-centered artificial intelligence (AI) model could automatically detect dental plaque and gingival inflammation, advancing predictive and preventive strategies for oral health. To verify the working hypothesis, a calibrated periodontist annotated 504 intraoral images, generating target masks (TM) as ground truth. A YOLOv8Seg-based deep learning model was trained for simultaneously segmented teeth, dental plaque, and gingival health status (healthy vs. inflamed), generating predicted masks (PM) that were subsequently used for classification tasks, including the calculation of gingival and plaque indices.
RESULTS: The model achieved moderate segmentation performance (IoU = 47%, DSC = 61%), with higher accuracy for tooth regions (mAP = 71% and 77%). Detection of dental plaque, healthy gingiva, and inflamed gingiva showed moderate precision (52%). Plaque index classification performed strongly (DSC = 95%, recall = 91%), whereas gingival inflammation showed moderate but clinically meaningful accuracy (DSC = 70%, recall = 92%), supporting early identification of inflammatory burden.
CONCLUSIONS: The proposed model functions as a practical digital tool for predictive oral healthcare by generating actionable, image-based biomarkers for patient phenotyping, early risk flagging, and site-specific behavioural reinforcement. Its lightweight architecture enables future integration into mobile platforms for longitudinal digital health monitoring and precision prevention. Expert recommendations include embedding AI-based plaque and gingivitis assessment into mobile health tools to enhance participatory self-monitoring; integrating imaging-derived biomarkers with behavioural, microbiological and socioeconomic information to support multimodal diagnostics and more accurate patient profiling; operationalising targeted prevention through site-specific alerts and personalised recall strategies; and deploying lightweight AI solutions in community-level screening programmes to reduce the burden of chronic inflammatory oral conditions. This work supports the transition from reactive treatment to a proactive, system-oriented PPPM.
SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s13167-025-00432-5.
WORKING HYPOTHESIS AND METHODOLOGY: This study hypothesised that a patient-centered artificial intelligence (AI) model could automatically detect dental plaque and gingival inflammation, advancing predictive and preventive strategies for oral health. To verify the working hypothesis, a calibrated periodontist annotated 504 intraoral images, generating target masks (TM) as ground truth. A YOLOv8Seg-based deep learning model was trained for simultaneously segmented teeth, dental plaque, and gingival health status (healthy vs. inflamed), generating predicted masks (PM) that were subsequently used for classification tasks, including the calculation of gingival and plaque indices.
RESULTS: The model achieved moderate segmentation performance (IoU = 47%, DSC = 61%), with higher accuracy for tooth regions (mAP = 71% and 77%). Detection of dental plaque, healthy gingiva, and inflamed gingiva showed moderate precision (52%). Plaque index classification performed strongly (DSC = 95%, recall = 91%), whereas gingival inflammation showed moderate but clinically meaningful accuracy (DSC = 70%, recall = 92%), supporting early identification of inflammatory burden.
CONCLUSIONS: The proposed model functions as a practical digital tool for predictive oral healthcare by generating actionable, image-based biomarkers for patient phenotyping, early risk flagging, and site-specific behavioural reinforcement. Its lightweight architecture enables future integration into mobile platforms for longitudinal digital health monitoring and precision prevention. Expert recommendations include embedding AI-based plaque and gingivitis assessment into mobile health tools to enhance participatory self-monitoring; integrating imaging-derived biomarkers with behavioural, microbiological and socioeconomic information to support multimodal diagnostics and more accurate patient profiling; operationalising targeted prevention through site-specific alerts and personalised recall strategies; and deploying lightweight AI solutions in community-level screening programmes to reduce the burden of chronic inflammatory oral conditions. This work supports the transition from reactive treatment to a proactive, system-oriented PPPM.
SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s13167-025-00432-5.