OSTEODENT: Using model-base image analysis of dental panoramic radiographs to identify sujects with osteopenia
TL;DRAbstract
Objectives: To develop image analysis software that can robustly and accurately measure the width of the inferior mandibular cortex on dental panoramic radiographs as an indicator of reduced bone mineral density (BMD). \nMethods: An Active Shape Model of the cortex was trained by expert annotation of 132 radiographs and used to locate the superior and inferior boundaries of the cortex automatically in a leave-one-out cross-validation test. Anatomical accuracy of the search is increased by employing a straightforward user interaction. Width was measured at 50 locations bilaterally between the antegonion and the mental foramen and averaged. Correlation between width and BMD (determined by DXA at the hip and spine) was measured and found to be highest in a region slightly medial to the antegonion – the “optimal” region. Taking osteopenia (T≤-1.0) as the categorical variable, we conducted ROC analysis using average width for both automatic and interactive search (tests 1 and 2) and “op
Chat with Paper
AI Agents for this Paper
Objectives: To develop image analysis software that can robustly and accurately measure the width of the inferior mandibular cortex on dental panoramic radiographs as an indicator of reduced bone mineral density (BMD). \nMethods: An Active Shape Model of the cortex was trained by expert annotation of 132 radiographs and used to locate the superior and inferior boundaries of the cortex automatically in a leave-one-out cross-validation test. Anatomical accuracy of the search is increased by employing a straightforward user interaction. Width was measured at 50 locations bilaterally between the antegonion and the mental foramen and averaged. Correlation between width and BMD (determined by DXA at the hip and spine) was measured and found to be highest in a region slightly medial to the antegonion – the “optimal” region. Taking osteopenia (T≤-1.0) as the categorical variable, we conducted ROC analysis using average width for both automatic and interactive search (tests 1 and 2) and “op
Keywords
Chat
Click to start Chat