Dental panoramic image analysis for enhancement biomarker of mandibular condyle for osteoporosis early detection

13th South-East Asian Congress of Medical Physics 2015

物理学;医药卫生

Suprijanto^1 ; Azhari^2 ; Juliastuti, E.^1 ; Septyvergy, A.^3 ; Setyagar, N.P.P.^3

Instrumentation and Control Research Group, Faculty of Industrial Technology, Institut Teknologi Bandung, Indonesia^1

Department of Dentomaxilofacial Radiology, Faculty of Dentistry, Padjadjaran University, Bandung, Indonesia^2

Engineering Physics, Faculty of Industrial Technology, Institut Teknologi Bandung, Indonesia^3

关键词: Artificial neural network methods;    Bone mineral density;    Degenerative disease;    Dual energy x ray absorptiometry (DXA);    Evaluation systems;    Panoramic images;    Radiography images;    The region of interest (ROI);   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/694/1/012066/pdf
DOI  :  10.1088/1742-6596/694/1/012066

学科分类：卫生学

来源: IOP

【 摘 要 】

Osteoporosis is a degenerative disease characterized by low Bone Mineral Density (BMD). Currently, a BMD level is determined by Dual Energy X-ray Absorptiometry (DXA) at the lumbar vertebrae and femur. Previous studies reported that dental panoramic radiography image has potential information for early osteoporosis detection. This work reported alternative scheme, that consists of the determination of the Region of Interest (ROI) the condyle mandibular in the image as biomarker and feature extraction from ROI and classification of bone conditions. The minimum value of intensity in the cavity area is used to compensate an offset on the ROI. For feature extraction, the fraction of intensity values in the ROI that represent high bone density and the ROI total area is perfomed. The classification will be evaluated from the ability of each feature and its combinations for the BMD detection in 2 classes (normal and abnormal), with the artificial neural network method. The evaluation system used 105 panoramic image data from menopause women which consist of 36 training data and 69 test data that were divided into 2 classes. The 2 classes of classification obtained 88.0% accuracy rate and 88.0% sensitivity rate.