Johannes Keustermans

Feature detection on 3D face surfaces for pose normalisation and recognition

This paper presents a SIFT algorithm adapted for 3D surfaces (called meshSIFT) and its applications to 3D face pose normalisation and recognition. The algorithm allows reliable detection of scale space extrema as local feature locations. The scale space contains the mean curvature in each vertex on different smoothed versions of the input mesh. The meshSIFT algorithm then describes the neighbourhood of every scale space extremum in a feature vector consisting of concatenated histograms of shape indices and slant angles. The feature vectors are reliably matched by comparing the angle in feature space. Using RANSAC, the best rigid transformation can be estimated based on the matched features leading to 84% correct pose normalisation of 3D faces from the Bosphorus database. Matches are mostly found between two face surfaces of the same person, allowing the algorithm to be used for 3D face recognition. Simply counting the number of matches allows 93.7% correct identification for face surfaces in the Bosphorus database and 97.7% when only frontal images are considered. In the verification scenario, we obtain an equal error rate of 15.0% to 5.1% (depending on the investigated face surfaces). These results outperform most other algorithms found in literature.

Integrating Statistical Shape Models into a Graph Cut Framework for Tooth Segmentation

The segmentation of teeth is of great importance for the computer aided planning of dental implants, orthodontic treatment, and orthognathic surgery. However, it is hampered by metallic streak artifacts present in Computed Tomography (CT) images in general, and the lack of contrast between the teeth and bone in Cone-Beam CT (CBCT) images particularly. Therefore, we propose a novel graph cut based algorithm that effectively integrates a statistical shape model based on a probabilistic shape representation. The statistical shape model is obtained from a set of training samples and imposes a Gaussian distribution on the shape space. The presented algorithm minimises an energy function that is formulated according to a maximum a posteriori criterion and consists of three terms: an image likelihood term, a segmentation likelihood term integrating the shape model into the graph cut framework, and a shape model term favoring shapes that are more likely according to the statistical shape model.

Automated cephalometric landmark localization using sparse shape and appearance models

In this paper an automated method is presented for the localization of cephalometric landmarks in craniofacial cone-beam computed tomography images. This methodmakes use of a statistical sparse appearance and shape model obtained fromtraining data. The sparse appearance model captures local image intensity patterns around each landmark. The sparse shape model, on the other hand, is constructed by embedding the landmarks in a graph. The edges of this graph represent pairwise spatial dependencies between landmarks, hence leading to a sparse shape model. The edges connecting different landmarks are defined in an automated way based on the intrinsic topology present in the training data. A maximum a posteriori approach is employed to obtain an energy function. To minimize this energy function, the problem is discretized by considering a finite set of candidate locations for each landmark, leading to a labeling problem. Using a leave-one-out approach on the training data the overall accuracy of the method is assessed. The mean and median error values for all landmarks are equal to 2.41 mm and 1.49 mm, respectively, demonstrating a clear improvement over previously published methods.

Feature-based piecewise rigid registration in 2-D medical images

Piecewise rigid registration is a fundamental problem in medical imaging involving intra-patient pose differences in multiple medical images, mostly due to articulated motion. In this paper, we propose a method to extract multiple rigid transformations in 2D medical images in the presence of outliers. First, points of interest in the images are extracted and matched with the SIFT algorithm. Secondly, multiple rigid motions are sampled and clustered by the mean shift algorithm in the special Euclidean group SE(2), a smooth manifold of 2-D rigid transformation matrices. The method proposed is evaluated for intra-subject registrations of knee fluoroscopy images, demonstrating a mean angular and translational error on the estimated motion of 0.39° and 6.65 pixels, respectively.

Automated planning of ablation targets in atrial fibrillation treatment

Catheter based radio-frequency ablation is used as an invasive treatment of atrial fibrillation. This procedure is often guided by the use of 3D anatomical models obtained from CT, MRI or rotational angiography. During the intervention the operator accurately guides the catheter to prespecified target ablation lines. The planning stage, however, can be time consuming and operator dependent which is suboptimal both from a cost and health perspective. Therefore, we present a novel statistical model-based algorithm for locating ablation targets from 3D rotational angiography images. Based on a training data set of 20 patients, consisting of 3D rotational angiography images with 30 manually indicated ablation points, a statistical local appearance and shape model is built. The local appearance model is based on local image descriptors to capture the intensity patterns around each ablation point. The local shape model is constructed by embedding the ablation points in an undirected graph and imposing that each ablation point only interacts with its neighbors. Identifying the ablation points on a new 3D rotational angiography image is performed by proposing a set of possible candidate locations for each ablation point, as such, converting the problem into a labeling problem. The algorithm is validated using a leave-one-out-approach on the training data set, by computing the distance between the ablation lines obtained by the algorithm and the manually identified ablation points. The distance error is equal to 3.8±2.9 mm. As ablation lesion size is around 5-7 mm, automated planning of ablation targets by the presented approach is sufficiently accurate.

Automated Cephalometric Landmark Identification Using Shape and Local Appearance Models

In this paper a method is presented for the automated identification of cephalometric anatomical landmarks in craniofacial cone-beam CT images. This method makes use of statistical models, incorporating both local appearance and shape knowledge obtained from training data. Firstly, the local appearance model captures the local intensity pattern around each anatomical landmark in the image. Secondly, the shape model contains a local and a global component. The former improves the flexibility, whereas the latter improves the robustness of the algorithm. Using a leave-one-out approach to the training data, we assess the overall accuracy of the method. The mean and median error values for all landmarks are equal to 2.55mm and 1.72mm, respectively.

Image Segmentation Using Graph Representations and Local Appearance and Shape Models

A generic model-based segmentation algorithm is presented. Based on a set of training data, consisting of images with corresponding object segmentations, a local appearance and local shape model is build. The object is described by a set of landmarks. For each landmark a local appearance model is build. This model describes the local intensity values in the image around each landmark. The local shape model is constructed by considering the landmarks to be vertices in an undirected graph. The edges represent the relations between neighboring landmarks. By implying the markovianity property on the graph, every landmark is only directly dependent upon its neighboring landmarks, leading to a local shape model. The objective function to be minimized is obtained from a maximum a-posteriori approach. To minimize this objective function, the problem is discretized by considering a finite set of possible candidates for each landmark. In this way the segmentation problem is turned into a labeling problem. Mean field annealing is used to optimize this labeling problem. The algorithm is validated for the segmentation of teeth from cone beam computed tomography images and for automated cephalometric analysis.