DEVELOPING A COMPLETE AUTOMATIC 2D AND 3D CEPHALOMETRIC ANALYSIS SYSTEM

Abstract

Cephalometric is a technique employing oriented radiographs for making head measurements. Orthodontists use Cephalometric analysis for pre-treatment planning, evaluating post-treatment results and growth prediction. Cephalometric analysis is the process of extracting a set of Landmarks and performs certain analysis (computing angles and distances) from the X-Ray images. Automation of Cephalometric analysis will reduce the analysis time and improve identification of the landmarks. In order to perform Cephalometric analysis, an accurate annotation of Cephalometric landmarks has to be done. Automatic annotation of Cephalometric landmarks falls into two categories: priori-knowledge image processing techniques, Artificial Intelligence based techniques. This work considers developing a new system for automatic Cephalometric analysis in 2D and 3D with sufficient accuracy. There are three main parts in this system; automatic identification of Cephalometric landmarks, automatic analysis system for both hard (bony) and soft tissues and finally a 3D Cephalometric analysis system based on 3D reconstruction of CT slices. For the first part, four different techniques were evaluated and compared. The second part resulted in a graphical analysis with a complete trace for the Craniofacial outlines for bony and soft - tissues. The results of this work show that a hybrid model of the technique used Active Appearance Model and Active Shape Model in A multi-resolution framework based on a Leave-One-Out experiment using Skull sub-models gives an improvement in accuracy more than 25% for automatic identification of 25 Cephalometric landmarks with a complete trace for the Craniofacial outlines. This comparison between the results and previous work had also been documented. The last part in this work is the ability to perform 3D Cephalometric an lysis. Simple 3D measurements on the 3D skull were achieved. Previous work that depends on single planar radiographs has limited Cephalometric analysis to 2D; therefore, this work is huge step towards 3D Cephalometric analysis that is increasingly demanded in this area.