Mol A, van der Stelt P F
Department of Oral Radiology, Academic Center for Dentistry (ACTA), Free University, The Netherlands.
IEEE Trans Biomed Eng. 1991 Apr;38(4):357-9. doi: 10.1109/10.133231.
This paper describes an application of digital image analysis for the description of periapical bone lesion in dental radiography. These techniques enable the problem of observer variability to be circumvented, and can eventually be used for the quantitative assessment of bone lesions. Since periapical bone lesions appear radiographically as dark areas compared with their surrounding tissues, an edge-detection method was developed to extract the boundaries between anatomy and pathology. The original data were transformed in such a way that only the outlines remained. The lesion was assumed to be demarcated by a continuous outline, so all the contour points not being part of a continuous outline were excluded. With only limited operator interaction, the lesion contour could be projected successfully onto the original image. The result of this operation was a simplified diagnostic examination process promising a higher degree of objectivity in periapical bone lesion detection. It is concluded that the results encourage further development of image processing techniques suitable for the definitive detection and diagnosis of periapical bone lesions.
本文描述了数字图像分析在牙科放射成像中根尖周骨病变描述方面的应用。这些技术能够规避观察者变异性问题，并最终可用于骨病变的定量评估。由于根尖周骨病变在放射影像上与其周围组织相比呈现为暗区，因此开发了一种边缘检测方法来提取解剖结构与病理之间的边界。原始数据经过变换，使得仅保留轮廓。假定病变由连续轮廓划定，因此排除所有不属于连续轮廓的轮廓点。只需有限的操作员交互，病变轮廓就能成功投影到原始图像上。该操作的结果是简化了诊断检查过程，有望在根尖周骨病变检测中实现更高程度的客观性。得出的结论是，这些结果鼓励进一步开发适用于根尖周骨病变明确检测和诊断的图像处理技术。
