Hi Miguel, If you build an ‘averaged’ crown across a number of complete ones and deformably register that to your partial crown, you might be able to ‘estimate’ those missing sections. This thread on the Slicer forum can give you some insight about how you can about it. https://discourse.slicer.org/t/non-rigid-registration-of-meshes/2247. If you have some python scripting skills, you can probably automate quite a bit of it.
Assuming all works well in creating these ‘completed’ crowns and you obtain sufficiently good meshes (in Slicer or elsewhere), the question you should carefully think about is how do you know that this is an improvement and a better method of analyzing that data. When we do automated landmarking, we typically try to show that the we can reach reasonable approximation of what experts do on the same dataset, which is the gold-standard. In that sense we cannot ‘better’ the human, you can only do ‘as good’. I am not trying to discourage you or anything, but make sure you have a clear idea of how you will measure the improvement, what metric to evaluate etc… M From: [email protected] <[email protected]> On Behalf Of Miguel Eduardo Delgado Burbano Sent: Tuesday, December 15, 2020 9:16 PM To: Murat Maga <[email protected]> Cc: Morphmet <[email protected]> Subject: Re: [MORPHMET2] Re: Template-Based Mesh Completion Dear Murat, Thank you for your suggestion. I'll definitely try it, though I never used 3D Slicer. Yes, absolutely, my idea is to compare the main differences between two approaches: 1) the oclusal basin (the homologous region you have mentioned) of molars and premolars vs. 2) the completed crown using a template. I'm aware that part of my shape analysis will be a bit biased given that the reconstructed part will be, in fact, used in statistical analysis, however that is exactly what I want to know. I'm interested in these kind of methods because in anterior dentition, at least, teeth are closer and maybe the completion of the 3D meshes is better than finding a way to automatically segment and cut the labial and buccal sides of an incisor or a canine. Meshlab has some similar functions to merge two 3D surfaces but additional complications include scaling, proper alignment using rigid or not rigid registration, etc. To my knowledge neither Morpho nor Rvcg have specific functions to estimate missing portions of 3D meshes using a complete template. Again, thank you very much for all your suggestions, I'll explore in detail 3D Slicer/ SlicerMorph, it seems a robust software to GM analysis and surface processing. All the best, Miguel On Wed, Dec 16, 2020 at 1:58 AM Murat Maga <[email protected]<mailto:[email protected]>> wrote: Dear Miguel, Potentially you can use a model of full tooth, and your segmentation, import both of them into 3D Slicer as segmentaitons and use the Segment Register module to register and fuse them. There are other ways of doing this too. The more important questions is what do you expect this method to achieve, and what does this reconstruction will do to your data and analysis? You say, you want to use a landmark-free automatic shape analysis method to analyze your data. Given that your existing data is essentially only the crown of the tooth, and if you reconstruct your models in a way that I describe above (or some similar way), and feed this data into some landmark-free analysis, you will have 90+% of your data coming from regions that are estimated. You should think about what would that mean for your results (and this is assuming you get a reasonable estimate of a full tooth to begin with). Instead, what would probably work better is actually find a common region across your existing samples. You can register them into a common coordinate, and crop in that space so that every model only contains the same region across your samples and then run your landmark-free analysis on these cropped models. This still would have some issues, but at least you will be comparing actual data. You can also do this in 3D Slicer / SlicerMorph, alternatively probably in Meshlab and in R using Morpho and Rvcg libraries. On Tuesday, December 15, 2020 at 9:26:33 AM UTC-8 [email protected]<mailto:[email protected]> wrote: Dear Morphometricians, I'm working on 3D meshes of dental plaster casts to investigate human dental shape variation using free-landmark methods. Through an automatic segmentation process I obtained individual meshes of each tooth, however given the nature of plaster casts all 3D meshes are incomplete (for instance large holes in the mesial and distal parts of a tooth given the interproximal contact)(please see attached picture). Thus, I'm looking for some methods, packages or software to automatically complete each 3D mesh using a template. I need an automatic method because I’m investigating thousands of individuals simultaneously. I know there are some interesting options like MeshMonk, but unfortunately I got poor results because MeshMonk is designed to work with 3D meshes from faces which are much less complex structures than a tooth. I'm wondering if someone can suggest some packages in R, Matlab, Mathematica or specific software to reconstruct missing parts of a 3D mesh using a complete template. Suggestions are very much appreciated. Thanks in advance. Miguel -- Miguel Delgado Ph.D. National Council of Scientific and Technical Research Adjunct Professor of Biological Anthropology Anthropology Division, Faculty of Natural Sciences and Museum, National University of La Plata, Argentina Ministry of Education Key Laboratory of Contemporary Anthropology Collaborative Innovation Center of Genetics and Development School of Life Sciences and Human Phenome Institute Fudan University, Shanghai, China Phone: +54 2214191107<tel:+54%20221%20419-1107> E-mail: [email protected]<mailto:[email protected]>, [email protected]<mailto:[email protected]> Website:https://www.researchgate.net/profile/Miguel_Delgado15 -- You received this message because you are subscribed to the Google Groups "Morphmet" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]<mailto:[email protected]>. To view this discussion on the web visit https://groups.google.com/d/msgid/morphmet2/899ac525-b685-46d6-900e-36f5aafe873dn%40googlegroups.com<https://groups.google.com/d/msgid/morphmet2/899ac525-b685-46d6-900e-36f5aafe873dn%40googlegroups.com?utm_medium=email&utm_source=footer>. -- Miguel Delgado Ph.D. National Council of Scientific and Technical Research Adjunct Professor of Biological Anthropology Anthropology Division, Faculty of Natural Sciences and Museum, National University of La Plata, Argentina Ministry of Education Key Laboratory of Contemporary Anthropology Collaborative Innovation Center of Genetics and Development School of Life Sciences and Human Phenome Institute Fudan University, Shanghai, China Phone: +54 2214191107 E-mail: [email protected]<mailto:[email protected]>, [email protected]<mailto:[email protected]> Website:https://www.researchgate.net/profile/Miguel_Delgado15 -- You received this message because you are subscribed to the Google Groups "Morphmet" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]<mailto:[email protected]>. To view this discussion on the web visit https://groups.google.com/d/msgid/morphmet2/CANhKaPCdaAnRu1uG_fSRiFJfOzd-U%3D3C%2BQLBaOG%3DX1O3ZVbtWw%40mail.gmail.com<https://groups.google.com/d/msgid/morphmet2/CANhKaPCdaAnRu1uG_fSRiFJfOzd-U%3D3C%2BQLBaOG%3DX1O3ZVbtWw%40mail.gmail.com?utm_medium=email&utm_source=footer>. -- You received this message because you are subscribed to the Google Groups "Morphmet" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/morphmet2/BN8PR08MB58429B0967E78DA4C0657A22A8C50%40BN8PR08MB5842.namprd08.prod.outlook.com.
