Multiphase Level Set Segmentation for ROI Extraction from Dental Radiographs
Image guided surgery, quantitative analysis and visual understanding along with proper medical interpretation have led to the development of segmentation techniques in image processing in a more precise manner. The deformable models provide an explicit representation of the boundary and shape of the object. Various features like inherent connectivity and smoothness which counteract noise and boundary irregularities are present in such models. Based on the region of interest they incorporate knowledge of the nearby regions. While using parametric model we faced certain pitfalls. Firstly, the conditions when the initial model and region of interest boundary differ greatly in size and shape, resulting in high error rate and lesser accuracy. Secondly, it was observed that when region of interest is large, then the model has to be applied more than once separately for each case. To overcome such situations, the level set deformable models also referred as geometric deformable model which provide an elegant solution to address the primary limitation of parametric deformable model. It includes no parameterisation of the contour, topological flexibility and good numerical stability. The results produced by the proposed method were compared with the ground truth results obtained with two domain experts. Based on the evaluation measures, the proposed method is found to be very effective in identification of interested region. The proposed method can be useful for assistance to the dental medical practitioners during their endodontic treatment of the tooth. The method can further be evaluated rigorously and can also be integrated in any computer based diagnostic tool.
KeywordsImage segmentation Level set method Single phase level set segmentation Multiphase level set segmentation Vitality detailing Bunching property Gradient descent method Contouring Deformable models Pixel connectivity Parametric model
- 58.Tejaswi, P., & Subhani, S. K. M. (2014). A region based variational method for image segmentation and bias correction with application to MRI. International Journal of Scientific Engineering and Research, 2(9), 40–46.Google Scholar
- 62.Chen, S., & Radke, R. J. (2009) Level set segmentation with both shape and intensity priors. In Proceedings of the IEEE international conference on computer vision, pp. 763–770.Google Scholar
- 65.Pock, T., Cremers, D., Bischof, H., & Chambolle, A. (2009). An algorithm for minimizing the Mumford-Shah functional. In Proceedings of the IEEE international conference on computer vision, pp. 1133–1140.Google Scholar
- 68.Jomier, J., Rault, E., & Aylward, S. R. (2004). Automatic Quantification of Pupil Dilation Under Stress. International Symposium on Biomedical Imaging (ISBI), 249–252.Google Scholar