Automatic segmenting teeth in X-ray images: Trends, a novel data set, benchmarking and future perspectives

Abstract: This review presents an in-depth study of the literature on segmentation methods applied in dental imaging. Ten segmentation methods were studied and categorized according to the type of the segmentation method (region-based, threshold-based, cluster-based, boundary-based or watershed-based), type of X-ray images used (intra-oral or extra-oral) and characteristics of the dataset used to evaluate the methods in the state-of-the-art works. We found that the literature has primarily focused on threshold-based segmentation methods (54%). 80% of the reviewed papers have used intra-oral X-ray images in their experiments, demonstrating preference to perform segmentation on images of already isolated parts of the teeth, rather than using extra-oral X-rays, which show tooth structure of the mouth and bones of the face. To fill a scientific gap in the field, a novel data set based on extra-oral X-ray images are proposed here. A statistical comparison of the results found with the 10 image segmentation methods over our proposed data set comprised of 1,500 images is also carried out, providing a more comprehensive source of performance assessment. Discussion on limitations of the methods conceived over the past year as well as future perspectives on exploiting learning-based segmentation methods to improve performance are also provided.

Automatic Segmenting Teeth in X-ray Images: Trends, Novel Data Set, Benchmarking, and Future Perspectives

The paper "Automatic Segmenting Teeth in X-ray Images: Trends, a Novel Data Set, Benchmarking and Future Perspectives" presents a comprehensive study on the segmentation methods specifically applied to dental imaging, thereby contributing significant insights into the field. The authors address segmenting tasks involving the identification and isolation of teeth structures from X-ray images, facilitating improved analytical capabilities for dental diagnostics and forensic identifications.

Overview and Methodology

The paper evaluates ten segmentation methods categorized into five principal types — region-based, threshold-based, cluster-based, boundary-based, and watershed-based techniques. It highlights a preference in literature for threshold-based segmentation methods (comprising 54% of surveyed studies), with 80% of past research utilizing intra-oral X-ray images. This demonstrates an inclination towards isolating teeth parts rather than employing extra-oral X-rays that encompass the complete dental and facial jaw structure.

A stark innovation in this paper is the proposed data set comprising 1,500 extra-oral X-ray images collected to bridge significant gaps identified in previous research — notably the size and variability limitations of existing public datasets. This novel data set allows a more formidable comparative evaluation of segmentation methods and assesses their performance across diverse structural variations in dental X-rays.

Results and Analysis

The authors found that methods using local variable thresholding, such as the Niblack method, performed better than those utilizing a global thresholding approach when segmenting entire dental structures. However, despite promising accuracies in precision and recall metrics, none of the evaluated segmentation methods completely isolated the teeth from the adjacent non-dental bone structures within panoramic images.

The statistical analysis of segmentation methods on the novel data set yielded key observations about how existing methodologies need adaptation or enhancement to handle the complexity posed by panoramic X-ray images. This comprehensive performance assessment and benchmarking reveal the limitations of current segmentation paradigms concerning dental radiography and point towards avenues for potential improvements.

Implications and Future Directions

This paper's findings highlight the need for advanced segmentation techniques that can scale with increased data diversity and complexity. Given the rapid advancements in artificial intelligence and machine learning methodologies, the paper underscores the potential of employing learning-based segmentation approaches. Techniques such as second-order statistical operations for clustering, semantic segmentation, and exploiting energy-minimization frameworks could revolutionize how dental X-ray images are processed.

Moreover, deep learning, with its capacity for handling vast datasets and complex pattern recognition, is poised as an attractive avenue for further exploration. The authors offer insights into how deep learning strategies, particularly those focused on semantic segmentation, could provide breakthroughs in automated dental diagnostics using X-ray imagery.

Conclusion

The paper raises significant inquiries regarding the influence of segmentation methods on dental image analysis and paves the way for future research directions. By proposing new methods and a novel data set, the authors commendably contribute to the understanding and development of practical solutions in dental imaging. This work establishes a solid foundation for future exploration and innovation in the application of AI to dental radiography, encouraging advancements that could ameliorate diagnostic accuracy and efficiency.