In this paper, we present an atlas-based level set automatic method for segmenting anatomical structures in head and neck CT data, such as mandible and brainstem. The proposed method is a hybrid method that combines two aspects. First, we register the atlas image to the image to be segmented using an intensity-based non-rigid image registration algorithm based on B-spline, the corresponding binary image of the atlas is also resampled into the reference image coordinate system according to the deformation field obtained from the registration process. Second, based on the initialization of the deformed atlas binary mask, the level set function is evolved to segment the object of interest in the test image. The proposed method was tested on CT images of head and neck and compared with expert segmentation of mandible and brainstem. The evaluation results show the method is available and effective.

Segmentation of Head and Neck CT Scans Using Atlas-based Level Set Method

This paper presents automated segmentation of structures in the Head and Neck (H&N) region, using an active contour-based joint registration and segmentation model. A new atlas selection strategy is also used. Segmentation is performed based on the dense deformation field computed from the registration of selected structures in the atlas image that have distinct boundaries, onto the patient's image. This approach results in robust segmentation of the structures of interest, even in the presence of tumors, or anatomical differences between the atlas and the patient image. For each patient, an atlas image is selected from the available atlas-database, based on the similarity metric value, computed after performing an affine registration between each image in the atlas-database and the patient's image. Unlike many of the previous approaches in the literature, the similarity metric is not computed over the entire image region; rather, it is computed only in the regions of soft tissue structures to be segmented. Qualitative and quantitative evaluation of the results is presented.

Active Contour-Based Segmentation of Head and Neck with Adaptive Atlas Selection

We present a semi-automatic algorithm for Carotid lumen segmentation on CTA images.
Our method is based on a variant of the minimal path method that models the vessel as a centerline and boundary. This is done by adding one dimension for the local radius around the centerline. The crucial step of our method is the definition of the local metrics to minimize. We have chosen to exploit the tubular structure of the vessels one wants to extract to built an anisotropic metric giving higher speed on the center of the vessels and also when the minimal path tangent is coherent with the vessel’s direction. Due to carotid stenosis or occlusions on the provided data, segmentation is refined using a region-based level sets.

Carotid Lumen Segmentation Based on Tubular Anisotropy and Contours Without Edges

This paper describes a 3D CTA image segmentation method submitted to the CLS09 contest (Carotid Lumen Segmentation and Stenosis Quantification) held in conjunction with the MICCAI 2009 conference. First, images are denoised to improve image quality and posterior segmentation. Second, region-based measurements are performed to differentiate possible vessels from other structures. Then, edge-driven metrics are used to allow vessel separation from nearby structures. Using, both edge-driven and region-based metrics a filter is used to enhance the vessels. The vessels of interest are extracted by use of the provided initialization points and of a model-driven segmentation algorithm. Using the obtained result, the final stage is devoted to stenosis quantification.

Adaptations of MARACAS algorithm to the segmentation of the carotid bifurcation and stenosis quantification in CTA images

We propose a new approach for semi-automatic segmentation of the carotid bifurcation as part of the Carotid Lumen Segmentation and Stenosis Grading Challenge MICCAI'2009 workshop. Three initial points are provided as input, belonging to the Common, the External and the Internal Carotid Arteries. Our algorithm is divided into two main steps: first, two minimal cost paths are tracked between the CCA and both the ECA and the ICA. The cost functions are based on a multiscale vesselness response. Second, after detecting the junction position and cutting or extending the paths based on the requested lengths, a level set segmentation is initialized as a thin tube around the computed paths and evolves until reaching the vessel wall or a maximal evolution time. Results on training and testing datasets are presented and compared to the manual segmentation by three observers, based on a ground truth and using four quality measures.


A Minimal Cost Path and Level Set Evolution Approach for Carotid Bifurcation Segmentation

This document describes a cost function, or metric, for use in image registration using the Insight Toolkit. The RatioVarianceImageToImageMetric class implements a metric which is calculated from the variance of the ratios of voxel values of the moving and fixed images.

Voxel Value Ratio Variance Image Registration Metric for the Insight Toolkit

The arterial lumen is modeled by a spatially continuous right generalized cylinder with piece-wise constant parameters. The method is the identifies the parameters of each cylinder piece from a series of planar contours extracted along an approximate axis of the artery. This curve is defined by a minimal path between the artery end-points. The contours are extracted by use of a 2D Fast Marching algorithm. The identification of the axial parameters is based on a geometrical analogy with piece-wise helical curves, while the identification of the surface parameters uses the Fourier series decomposition of the contours. Thus identified parameters are used as observations in a Kalman optimal estimation scheme that manages the spatial consistency from each piece to another. The method was was evaluated on 15 training and 31 testing datasets from the MICCAI 3D Segmentation in the Clinic Grand Challenge: Carotid Bifurcation Lumen Segmentation and Stenosis Grading (http://cls2009.bigr.nl/).

Carotid arteries segmentation in CT images with use of a right generalized cylinder model

Diffeomorphic demons registration was widely used for medical image registration. This document describes the addition of mask processing mode with spatial objects for itkDiffeomorphicemonsRegistrationFilter when just considering the deformation field at the region of interest.

DiffeomorphicDemonsRegistrationWithMaskFilter

The goal of the Cardiac MR Left Ventricle Segmentation Challenge at MICCAI 2009 is to compare state-of-the-art LV segmentation methods. This goal is facilitated through an evaluation system and a database of cardiac cine MR images, as well as expert contours, now freely available on the internet for research purposes. This challenge is important because the analysis and the segmentation of 3D+t sequences of MR cardiac images is fastidious, time consuming and error-prone for human operators, due to the large amount of data. Conversely, automated segmentation of cardiac images is well-known to be a challenging task.
This paper describes the method that our group submitted for the challenge, based on 4D discrete mathematical morphology. This method leads to discrete segmentations, i.e, binary masks (bitmaps) of the left ventricular myocardiums that are both spatially and temporally consistent.

4D Morphological segmentation and the MICCAI LV-segmentation grand challenge

The segmentation of left ventricular structures is necessary for the evaluation of the ejection fraction (EF) and the myocardial mass (LVM). A semi-automated 2D algorithm using connected filters and a deformable model allowing an accurate endocardial detection was proposed. The epicardial border was deduced using a deformable model restricted inside a region of interest defined from the endocardial border. Papillary muscles were detected using a fuzzy k-means algorithm. The method was applied to the challenge training and validation databases, consisting of 15 subjects each. The evaluation was performed using the tools provided by the challenge. For both datasets, results show a mean Dice metric of 0.89 for endocardial borders (0.92 for epicardial borders). Overall average perpendicular distance was 2.2 mm. Very good correlation was obtained for the EF and LVM parameters. Visual overall rating given by the challenge’s cardiologist was 1.2. Segmentation was robust and performed successfully on both datasets. 