In dentistry, panoramic X-ray images are extensively used by dentists for tooth structure analysis and disease diagnosis. However, the manual analysis of these images is time-consuming and prone to misdiagnosis or ove...In dentistry, panoramic X-ray images are extensively used by dentists for tooth structure analysis and disease diagnosis. However, the manual analysis of these images is time-consuming and prone to misdiagnosis or overlooked. While deep learning techniques have been employed to segment teeth in panoramic X-ray images, accurate segmentation of individual teeth remains an underexplored area. In this study, we propose an end-to-end deep learning method that effectively addresses this challenge by employing an improved combinatorial loss function to separate the boundaries of adjacent teeth, enabling precise segmentation of individual teeth in panoramic X-ray images. We validate the feasibility of our approach using a challenging dataset. By training our segmentation network on 115 panoramic X-ray images, we achieve an intersection over union (IoU) of 86.56% for tooth segmentation and an accuracy of 65.52% in tooth counting on 87 test set images. Experimental results demonstrate the significant improvement of our proposed method in single tooth segmentation compared to existing methods.展开更多
文摘In dentistry, panoramic X-ray images are extensively used by dentists for tooth structure analysis and disease diagnosis. However, the manual analysis of these images is time-consuming and prone to misdiagnosis or overlooked. While deep learning techniques have been employed to segment teeth in panoramic X-ray images, accurate segmentation of individual teeth remains an underexplored area. In this study, we propose an end-to-end deep learning method that effectively addresses this challenge by employing an improved combinatorial loss function to separate the boundaries of adjacent teeth, enabling precise segmentation of individual teeth in panoramic X-ray images. We validate the feasibility of our approach using a challenging dataset. By training our segmentation network on 115 panoramic X-ray images, we achieve an intersection over union (IoU) of 86.56% for tooth segmentation and an accuracy of 65.52% in tooth counting on 87 test set images. Experimental results demonstrate the significant improvement of our proposed method in single tooth segmentation compared to existing methods.
