Computation And Application Of Tooth-root Orientation From3D CT Images

In virtual dental surgery planning and dental implantology using computed to-mography (CT) images, the estimation of the quality, the quantity of the operating areaand the examination of the axis of neighboring and biting teeth is important to makesure the success of the manipulating dental implantation. Once the size, position andangle of the implant were selected incorrectly, the surgical complications such as theperforation of sinus and the damage of nerve etc would occur. In orthodontic treatmen-t, the alignment of the tooth roots in parallel axial inclinations is critical for a stableorthodontic result. In medical informatics, it is important to acquire normal shape androot orientations of teeth——such statistical data is useful in accurately extracting den-tal root or locating fractured roots or root tips.However, due to its high connectivity to neighboring teeth and jawbones, a toothand its axis is very elusive to automatically identify in dental CT images. According toanatomical knowledge and clinical experience, we note that most single root teeth havethe orientations similar as the center axis of their pulp cavity regions (or pulp canalregions). Based on this knowledge, we present some novel methods and a frameworkof tooth-root orientation computation and analysis system, it mainly includes:（1）. Interactivetooth-rootorientationcomputationonsingletooth-root. Firstly,a3D model of the bone structure of the jaw was reconstructed from3D CT image.Then, a virtual cutting-plane is placed artifcially at the position near each tooth-rootof interest. We can interactively adjust the orientation and position of the tooth-rootbasedonhuman’sobservation. Finally,thetooth-rootorientationcanbeobtainedbasedon the direction along which the plane spread. Incidentally, to convert the2D tooth-root orientation into a3D tooth-root orientation, a2D to3D points converting method was developed.（2）. Semi-automatic tooth-root orientation computation on single tooth-root.Tosimplifythemanipulation of the user, weimproveasemi-automatic methodtocom-pute the tooth-root orientation. Similarly, we generate a virtual plane cutting throughthe pulp cavity of the interested tooth. Then, the pulp cavity from3D images for theinterested tooth was segmented using region grow method. At last, we calculate thecenter of each slice, and the center axis from the segmented pulp cavity region wasextracted using the least squares ft method, which, is the wanted tooth-root orientationof the interested tooth.（3）. Multirooted tooth-root orientation computation based on PCA (PrincipalComponent Analysis) method. Since the multirooted tooth has multiple and irregulartooth-root, each of the slice may not has a clear center. So, the semi-automatic tooth-root orientation computation may not work well under this circumstance. A multiroot-ed toot-root orientation computation method was developed based on PCA method.Firstly, we segment the pulp cavity region of the multirooted tooth-root and calculatetheaxisofthepulpcavityregionforareference. Unfortunately, accordingtoourobser-vation, the calculated axis of the pulp cavity was not parallel with the rough orientationof the multirooted tooth-root. Consequently, a tooth segmentation method based on in-teractive bafe was developed, and the multirooted toot-root orientation computationmethod based on PCA method was verifed in this situation.（4）. Estimation of the missing tooth-root orientation by a quaternion inter-polation technique. We assume that both missing and existent teeth are normally dis-tributed along the dental arch, the root orientations of missing teeth was deduced fromroot orientation of the existent teeth adjacent to the missing teeth. The estimation ofthe root orientations of missing teeth can be described as the interpolation problemof some known3D orientation(namely, root orientations of existent teeth) along a3Ddental arch curve. Generally, the linear interpolation technique is not suitable for s-moothly interpolating3D orientations along a3D curve, a quaternion interpolationtechnique(squad method) was applied to this problem.（5）. The application of tooth-root orientation computation in the reconstruc-tion of virtual cutting-plane. The purpose the virtual cutting-plane is to re-express the 3D CT data into a2D cutting-plane. Once a reconstruction method based on the cut-ting path was proposed which can spread the interested region according to a certainpath. However, the cutting-direction of each point along this path is not scrupulouslycalculated. We applied the tooth-root orientation computation method into this recon-struction method. And the cutting-direction at each point was re-calculated accordingto some controlled orientation(several root orientations). Finally, we got a much moreaccurate result of the interested tooth cutting-plane.We cooperate with Shanghai Eastern Hospital in developing this system. Sever-al data sets were tested and some satisfed results can demonstrate the practical andreliable of these methods.