This paper presents an experimental study for assessing the applicability of general-purpose 3D segmentation algorithms for analyzing dental periapical lesions in cone-beam computed tomography (CBCT) scans. In the field of Endodontics, clinical studies have been unable to determine if a periapical granuloma can heal with non-surgical methods. Addressing this issue, Simon et al. recently proposed a diagnostic technique which non-invasively classifies target lesions using CBCT. Manual segmentation exploited in their study, however, is too time consuming and unreliable for real world adoption. On the other hand, many technically advanced algorithms have been proposed to address segmentation problems in various biomedical and non-biomedical contexts, but they have not yet been applied to the field of dentistry. Presented in this paper is a novel application of such segmentation algorithms to the clinically-significant dental problem. This study evaluates three state-of-the-art graph-based algorithms: a normalized cut algorithm based on a generalized eigen-value problem, a graph cut algorithm implementing energy minimization techniques, and a random walks algorithm derived from discrete electrical potential theory. In this paper, we extend the original 2D formulation of the above algorithms to segment 3D images directly and apply the resulting algorithms to the dental CBCT images. We experimentally evaluate quality of the segmentation results for 3D CBCT images, as well as their 2D cross sections. The benefits and pitfalls of each algorithm are highlighted.
Landmark placement is crucial in manual demarcation and registration of
anatomical structures, registration of different image modalities (i.e. MRI/CT), labeling training data for lip and face principal component models, training for neural networks, and signal interpolation to name some applications. Although landmark placement at curvature and coordinate extrema (e.g. corners of the mouth, lowest point on the lower lip) is fairly unambiguous, the placement of point landmarks along a linear contour is subjective. Unfortunately the user's choice of landmark placement determines the quality of the resulting registration.
In this paper, we present an algorithm to remove these undesired degrees of freedom by re-placing landmarks along the contour. Ambiguous landmarks are moved so as to minimize a thin plate spline energy while constraining the landmarks to the originally specified contour. The resulting landmark placement results in a smoother registration while still interpolating the contours and fixed landmarks. The results show that the ambiguity of manual landmark placement along contours does affect the smoothness of the interpolated registration, and that significantly smoother interpolations can be achieved using our approach. This procedure may also benefit other applications employing landmarks by eliminating unintended curvature (variation) from the landmark data.
Currently, two-dimensional photographs are most commonly used to facilitate visualization, assessment and treatment of facial abnormalities in craniofacial care but are subject to errors because of perspective, projection, lack metric and 3-dimensional information. One can find in the literature a variety of methods to generate 3-dimensional facial images such as laser scans, stereo-photogrammetry, infrared imaging and even CT however each of these methods contain inherent limitations and as such no systems are in common clinical use. In this paper we will focus on development of indirect 3-dimensional landmark location and measurement of facial soft-tissue with light-based techniques. In this paper we will statistically evaluate and validate a current three-dimensional image-based face modeling technique using a plaster head model. We will also develop computer graphics tools for indirect anthropometric measurements in a three-dimensional head model (or polygonal mesh) including linear distances currently used in anthropometry. The measurements will be tested against a validated 3-dimensional digitizer (MicroScribe 3DX).