Kelvin Weng Chiong Foong

Medical Image Segmentation Using K-Means Clustering and Improved Watershed Algorithm

We propose a methodology that incorporates k-means and improved watershed segmentation algorithm for medical image segmentation. The use of the conventional watershed algorithm for medical image analysis is widespread because of its advantages, such as always being able to produce a complete division of the image. However, its drawbacks include over-segmentation and sensitivity to false edges. We address the drawbacks of the conventional watershed algorithm when it is applied to medical images by using k-means clustering to produce a primary segmentation of the image before we apply our improved watershed segmentation algorithm to it. The k-means clustering is an unsupervised learning algorithm, while the improved watershed segmentation algorithm makes use of automated thresholding on the gradient magnitude map and post-segmentation merging on the initial partitions to reduce the number of false edges and over-segmentation. By comparing the number of partitions in the segmentation maps of 50 images, we showed that our proposed methodology produced segmentation maps which have 92% fewer partitions than the segmentation maps produced by the conventional watershed algorithm

Tooth segmentation of dental study models using range images

The accurate segmentation of the teeth from the digitized representation of a dental study model is an important component in computer-based algorithms for orthodontic feature detection and measurement and in the simulation of orthodontic procedures such as tooth rearrangement. This paper presents an automated method for tooth segmentation from the three-dimensional (3-D) digitized image captured by a laser scanner. We avoid the complexity of directly processing 3-D mesh data by proposing the innovative idea of detecting features on two range images computed from the 3-D image. The dental arch is first obtained from the plan-view range image. Using the arch as the reference, a panoramic range image of the dental model can be computed. The interstices between the teeth are detected separately in the two range images, and results from both views are combined for a determination of interstice locations and orientations. Finally, the teeth are separated from the gums by delineating the gum margin. The algorithm was tested on 34 dental models representing a variety of malocclusions and was found to be robust and accurate.

A robust global and local mixture distance based non-rigid point set registration

We present a robust global and local mixture distance (GLMD) based non-rigid point set registration method which consists of an alternating two-step process: correspondence estimation and transformation updating. We first define two distance features for measuring global and local structural differences between two point sets, respectively. The two distances are then combined to form a GLMD based cost matrix which provides a flexible way to estimate correspondences by minimizing global or local structural differences using a linear assignment solution. To improve the correspondence estimation and enhance the interaction between the two steps, an annealing scheme is designed to gradually change the cost minimization from local to global and the thin plate spline transformation from rigid to non-rigid during registration. We test the performance of our method in contour registration, sequence images and real images, and compare with six state-of-the-art methods where our method shows the best alignments in most scenarios. HighlightsWe develop a multi-feature based non-rigid point set registration algorithm.We define two distance based features.A multi-feature based cost matrix is designed for correspondence estimation.We solve issues in the current iterative non-rigid point set registration methods.The performance of our method outperforms state-of-the-art methods in this area.

A level-set based approach for anterior teeth segmentation in cone beam computed tomography images.

Cone beam CT (CBCT) has gained popularity in dentistry for 3D imaging of the jaw bones and teeth due to its high resolution and relatively lower radiation exposure compared to multi-slice CT (MSCT). However, image segmentation of the tooth from CBCT is more complex than from MSCT due to lower bone signal-to-noise. This paper describes a level-set method to extract tooth shape from CBCT images of the head. We improve the variational level set framework with three novel energy terms: (1) dual intensity distribution models to represent the two regions inside and outside the tooth; (2) a robust shape prior to impose a shape constraint on the contour evolution; and (3) using the thickness of the tooth dentine wall as a constraint to avoid leakage and shrinkage problems in the segmentation process. The proposed method was compared with several existing methods and was shown to give improved segmentation accuracy.

Autotransplantation of 28 premolar donor teeth in 24 orthodontic patients.

OBJECTIVE To test the null hypothesis that premolar autotransplantation is not successful for orthodontic patients. MATERIALS AND METHODS In the present study, 28 premolar transplants from 24 orthodontic patients were associated with orthodontic treatment. At a routine 3-month appointment, patients underwent a dental radiograph and a chair-side observation for periodontal problems. Three sets of dental radiographs were taken by one dental assistant using a custom holder at: preoperation (T0), 2 year postoperation (T1), and retention (T2) (4- to 14-year follow-up observation) stages. All transplants were conducted in a one-phase operation by one operator (Dr Inoue). Recipient sites were: ten for missing maxillary canines, nine for maxillary centrals and laterals, eight for lower second premolar, and five for other missing premolar sites. All recipients maintained the retained primary tooth with a socket. RESULTS The success ratio of all 28 transplants was 100%, although four transplants shorter than a 4-year period of observation were omitted. Two transplant patients, one with a medical history of histiocytosis and the other with a history of osteomyelitis of the maxilla without a recipient socket, were also excluded from this study. Eleven of 22 premolar transplants had a root canal treatment (RCT), four of which had RCT within 2 years after the operation. CONCLUSIONS The null hypothesis was rejected. The success ratio of premolar transplants was 100%.

Computer-based extraction of the inferior alveolar nerve canal in 3-D space

Computed tomography (CT) is increasingly employed in orthodontic treatment. One of the most successful applications is in dental implantology, in which an artificial root is surgically inserted into the jawbone to provide anchorage for a dental prosthesis. For successful implant surgery, it is crucial to locate internal structures such as the inferior alveolar nerve canal (IAC). This paper presents a computerized technique for extracting the IAC. To facilitate the extraction, we first generate panoramic CT images (panoramics) by reformatting the original CT images. The panoramics are a series of cross-sectional images along curved planes through the mandible (lower jawbone). Hollow canals are subsequently detected by analyzing the voxel intensities and 3-D gradient orientations in the panoramics. The axis of the IAC is then traced out by a novel 3-D line-tracking technique. The method is effective for extracting the IAC despite the open structure of the surrounding bone.

Medical image segmentation using watershed segmentation with texture-based region merging

The use of the watershed algorithm for image segmentation is widespread because it is able to produce a complete division of the image. However, it is susceptible to over-segmentation and in medical image segmentation, this meant that that we do not have good representations of the anatomy. We address this issue by thresholding the gradient magnitude image and performing post-segmentation merging on the initial segmentation map. The automated thresholding technique is based on the histogram of the gradient magnitude map while the post-segmentation merging is based on the similarity in textural features (namely angular second moment, contrast, entropy and inverse difference moment) belonging to two neighboring partitions. When applied to the segmentation of various facial anatomical structures from magnetic resonance (MR) images, the proposed method achieved an overlap index of 92.6% compared to manual contour tracings. It is able to merge more than 80% of the initial partitions, which indicates that a large amount of over-segmentation has been reduced. Results produced using watershed algorithm with and without the proposed and proposed post-segmentation merging are presented for comparisons.

Morphologic quantification of the maxilla and the mandible with cone-beam computed tomography

INTRODUCTION The purpose of this pilot study was to use cone-beam computed tomography (CBCT) to determine the volumes of the maxilla and the mandible in subjects with skeletal Class I, Class II, and Class III malocclusions. Hypothesis 1 was that the volume (size) of a skeletal Class II maxilla is larger than those of Class I and Class III. Hypothesis 2 was that the volume of a skeletal Class III mandible is larger than those of Class I and Class II. METHODS Thirty women patients were classified into 3 groups according to their skeletal pattern: skeletal Class I (0 degrees < or =ANB <6 degrees ), Class II (ANB > or =6 degrees ) and Class III (ANB <0 degrees ). The volumes of the maxilla and the mandible were measured with CBCT. CB MercuRay (Hitachi Medico, Tokyo, Japan) and CB works software (CyberMed, Seoul, Korea) were used to process the images. RESULTS There was a trend that skeletal Class III subjects might have significantly greater mandibular volume compared with Class II subjects (P = 0.089). The ratios of maxilla-to-mandible volumes between the skeletal Class II and Class III groups were significantly different (P = 0.005). Differences were observed in the ratios of maxillary and mandibular volumes across the 3 groups. CONCLUSIONS Hypotheses 1 and 2 were rejected; there was no trend for Class III subjects to have larger mandibles (P = 0.089) compared with Class II subjects. The ratio of the maxilla and mandible volumes in skeletal Class III subjects was significantly larger (P = 0.005) compared with Class II subjects.

Efficient partial-surface registration for 3D objects

We present a new method for three-dimensional partial-surface registration that utilizes both regional surface properties and shape rigidity constraint to align a partial object surface and its corresponding complete surface. The statistical properties of the vertices on the object surface are first computed and compared with each other to find the initial candidate correspondences. We use the overall object-shape rigidity constraint and a clustering method to obtain an approximation of the transformation parameters while, at the same time, rejecting correspondence outliers. The transformation parameters can be further refined with an iterative approach. The algorithm does not require any feature extraction or initial pose estimation, and is especially applicable when the object surfaces are formed by a large number of vertices, smooth with few salient features, and contain many regionally similar surface patches. Experiments confirm that the proposed scheme can achieve accurate registration results in an efficient manner.

Improved spin images for 3D surface matching using signed angles

Despite the popularity of spin images in surface matching and registration, disadvantages such as noise sensitivity and low discriminative ability still hindered their usefulness in real applications. In this paper, a novel approach was proposed for improving the spin images. The proposed method modified the standard spin images by using angle information between the normals of reference point and neighboring points. This information largely increased the robustness to noise without losing the intrinsic advantages of spin images. Moreover, signs were defined to incorporate the directions of angles which were shown to be able to further improve the descriptive power. Experiments were also conducted to show the outperformance of improved spin images under different levels of noise, and good agreements were obtained by comparing with the standard spin images and a recent popular 3D descriptor.