V. Grau

Improved watershed transform for medical image segmentation using prior information

The watershed transform has interesting properties that make it useful for many different image segmentation applications: it is simple and intuitive, can be parallelized, and always produces a complete division of the image. However, when applied to medical image analysis, it has important drawbacks (oversegmentation, sensitivity to noise, poor detection of thin or low signal to noise ratio structures). We present an improvement to the watershed transform that enables the introduction of prior information in its calculation. We propose to introduce this information via the use of a previous probability calculation. Furthermore, we introduce a method to combine the watershed transform and atlas registration, through the use of markers. We have applied our new algorithm to two challenging applications: knee cartilage and gray matter/white matter segmentation in MR images. Numerical validation of the results is provided, demonstrating the strength of the algorithm for medical image segmentation.

An Advanced System for the Simulation and Planning of Orthodontic Treatments

This paper describes a new method for 3D orthodontic treatment simulation developed for an orthodontic planning system (MAGA-LLANES). We develop an original system for three-dimensional reconstruction of dental anatomy. Data are acquired directly from the patient with low cost 3D digitizers avoiding use of dental casts in orthodontic treatments. We apply these 3D dental models to simulate three-dimensional movement of teeth, including rotations, during orthodontic treatment. We develop an original simplified model of arch-wire behavior and a viscoplastic behavior law for the alveolar bone, to simulate teeth displacements during orthodontic treatments. The proposed algorithm enables to quantify the effect of orthodontic appliances on teeth movement. Preliminary results are very promising.

Automatically Generated, Anatomically Accurate Meshes for Cardiac Electrophysiology Problems

Significant advancements in imaging technology and the dramatic increase in computer power over the last few years broke the ground for the construction of anatomically realistic models of the heart at an unprecedented level of detail. To effectively make use of high-resolution imaging datasets for modeling purposes, the imaged objects have to be discretized. This procedure is trivial for structured grids. However, to develop generally applicable heart models, unstructured grids are much preferable. In this study, a novel image-based unstructured mesh generation technique is proposed. It uses the dual mesh of an octree applied directly to segmented 3-D image stacks. The method produces conformal, boundary-fitted, and hexahedra-dominant meshes. The algorithm operates fully automatically with no requirements for interactivity and generates accurate volume-preserving representations of arbitrarily complex geometries with smooth surfaces. The method is very well suited for cardiac electrophysiological simulations. In the myocardium, the algorithm minimizes variations in element size, whereas in the surrounding medium, the element size is grown larger with the distance to the myocardial surfaces to reduce the computational burden. The numerical feasibility of the approach is demonstrated by discretizing and solving the monodomain and bidomain equations on the generated grids for two preparations of high experimental relevance, a left ventricular wedge preparation, and a papillary muscle.

Outlining of the prostate using snakes with shape restrictions based on the wavelet transform (Doctoral Thesis: Dissertation)

This paper considers the problem of deformable contour initialization and modeling for segmentation of the human prostate in medical images. We propose a new technique for elastic deformation restriction to particular object shapes of any closed planar curve using localized multiscale contour parameterization based on the 1D dyadic wavelet transform. For this purpose we define internal curve deformation forces as a result of multiscale parametrical contour analysis. The form restricted contour deformation and its initialization by template matching are performed in a coarse to fine segmentation process based on a multiscale image edge representation containing the important edges of the image at various scales. The method is useful for 3D conformal radiotherapy planning and automatic prostate volume measurements in ultrasonographic diagnosis.

Automatic Localization of Cephalometric Landmarks

A system for automatic detection of cephalometric landmarks is presented. Landmark detection is carried out in two steps: a line detection module searches for significant, well-contrasted lines of the image, such as the jaw line or the nasal spine. The landmark detection module uses the lines located in the first module to determine the search areas and then applies a pattern detection algorithm, based on mathematical morphology techniques. Relations between landmarks and lines are determined by means of a training process. The system has been tested for the detection of 17 landmarks on 20 images: more than 90% of the landmarks are accurately identified.

Local-phase based 3D boundary detection using monogenic signal and its application to real-time 3-D echocardiography images

Ultrasound images are characterized by their speckle appearance, low contrast, and poor signal-to-noise ratio. It is always challenging to extract important clinical information from these images. An important step before automatic measurement is to transform the image into significant features of interest. Intensity based methods do not perform particularly well on ultrasound images. However, it has been shown previously that ultrasound images respond well to local phase-based methods which are theoretically intensityinvariant and thus suitable for low-contrast nature of ultrasound images. We extend the local phase-based method of feature asymmetry measure computation to detect 3D features using the monogenic signal, which is an isotropic extension of the analytic signal to higher dimensional functions. The proposed method is applied to real-time 3D echocardiography images and the visual results for the endocardial and epicardial boundary detection are presented.

Segmentation of trabeculated structures using an anisotropic Markov random field: application to the study of the optic nerve head in glaucoma

The study of the architecture of the optic nerve head (ONH) may provide valuable information about the development and progression of glaucoma. To this end, we have generated three-dimensional datasets from monkey eyes under controlled intraocular pressure (IOP). Segmentation of the connective tissues in this area is crucial to obtain an accurate measurement of geometrical parameters and to build mechanical models. However, this segmentation is made difficult by the complicated geometry and the artifacts introduced in the dataset building process. We present a novel segmentation algorithm, based on expectation-maximization, which incorporates an anisotropic Markov random field (MRF) to introduce prior knowledge about the geometry of the structure. The structure tensor is used to characterize the predominant structure direction and the spatial coherence at each point. The algorithm, which has been validated on an artificial validation dataset that mimics our ONH datasets, shows significant improvement over an isotropic MRF. Results on the real datasets demonstrate the ability of the new algorithm to obtain accurate, spatially consistent segmentations of this structure.

Comparison of Rule-Based and DTMRI-Derived Fibre Architecture in a Whole Rat Ventricular Computational Model

The anisotropic electrical conduction within myocardial tissue due to preferential cardiac myocyte orientation (`fibre orientation) is known to impact strongly in electrical wavefront dynamics, particularly during arrhythmogenesis. Faithful representation of cardiac fibre architecture within computational cardiac models which seek to investigate such phenomena is thus imperative. Drawbacks in derivation of fibre structure from imaging modalities often render rule-based representations based on a priori knowledge preferential. However, the validity of using such rule-based approaches within whole ventricular models remains unclear. Here, we present the development of a generic computational framework to directly compare the fibre architecture predicted by rule-based methods used within whole ventricular models against fibre structure derived from DTMRI data, and assess how relative differences influence propagation dynamics throughout the ventricles. Results demonstrate the close overall match between the methods within the rat ventricles, and highlight regions for potential rule-adaption.

Real-time 3D fusion echocardiography.

OBJECTIVESnThis study assessed 3-dimensional fusion echocardiography (3DFE), combining several real-time 3-dimensional echocardiography (RT3DE) full volumes from different transducer positions, for improvement in quality and completeness of the reconstructed image.nnnBACKGROUNDnThe RT3DE technique has limited image quality and completeness of datasets even with current matrix transducers.nnnMETHODSnRT3DE datasets were acquired in 32 participants (mean age 33.7 +/- 18.8 years; 27 men, 5 women). The 3DFE technique was also performed on a cardiac phantom. The endocardial border definition of RT3DE and 3DFE segments was graded for quality: good (2), intermediate (1), poor (0), or out of sector. Short-axis and apical images were compared in RT3DE and 3DFE, yielding 2,048 segments. The images were processed to generate 3DFE and then compared with cardiac magnetic resonance.nnnRESULTSnIn the heart phantom, fused datasets showed improved contrast to noise ratio from 49.4 +/- 25.1 (single dataset) to 125.4 +/- 25.1 (6 datasets fused together). In subjects, more segments were graded as good quality with 3DFE (805 vs. 435; p < 0.0001) and fewer as intermediate (184 vs. 283; p = 0.017), poor (31 vs. 265; p < 0.0001), or out of sector (4 vs. 41; p < 0.001) compared with the single 3-dimensional dataset. End-diastolic volume (EDV) and end-systolic volume (ESV) obtained from 3-dimensional fused datasets were equivalent to those from single datasets (EDV 118.2 +/- 39 ml vs. 119.7 +/- 43 ml; p = 0.41; ESV 48.1 +/- 30 ml vs. 48.4 +/- 35 ml; p = 0.87; ejection fraction [EF] 61.0 +/- 10% vs. 61.8 +/- 10%; p = 0.44). Bland-Altman analysis showed good 95% limits of agreement for the nonfused datasets (EDV +/-46 ml; ESV +/-36 ml; EF +/-14%) and the fused datasets (EDV +/-45 ml; ESV +/-35 ml; EF +/-16%), when compared with cardiac magnetic resonance.nnnCONCLUSIONSnFusion of full-volume datasets resulted in an improvement in endocardial borders, image quality, and completeness of the datasets.

An advanced system for the simulation and planning of orthodontic treatment

This paper presents a new system for three-dimensional (3-D) orthodontic treatment planning and movement of teeth. We describe a computer vision technique for the acquisition and processing of 3-D images of the profile of hydrocolloid dental imprints. Profile measurement is based on the triangulation method which detects deformation of the projection of a laser line on the dental imprints. The system is computer-controlled and designed to achieve depth and lateral resolutions of 0.1 and 0.2 mm, respectively, within a depth range of 40 mm. The 3-D image of the imprint is segmented in order to identify different teeth. Two operators are presented: one for the detection of molars and premolars based on a directional gradient, and one for incisors and canines based on 3-D registration with dental models contained in a database. We apply these 3-D dental models to simulate the 3-D movement of teeth, including rotations, during orthodontic treatment. With this objective, we have developed an original simplified model of arch-wire behaviour and a viscoplastic behaviour law for the alveolar bone in order to simulate teeth displacements during orthodontic treatment. The contribution of the paper is part of a diagnosis system (called MAGALLANES) that is designed to replace manual measurement methods, which use costly plaster models, with computer measurement methods and teeth movement simulation using cheap hydrocolloid dental wafers. This procedure will reduce the cost and acquisition time of orthodontic data and facilitate the conduct of epidemiological studies.