Dense deformation field estimation for atlas registration using the active contour framework

A key research area in computer vision is image segmentation. Image segmentation aims at extracting objects of interest in images or video sequences. These objects contain relevant information for a given application. For example, a video surveillance application generally requires to extract moving objects (vehicles, persons or animals) from a sequence of images in order to check that their path stays conformed to the regulation rules set for the observed scene. Image segmentation is not an easy task. In many applications, the contours of the objects of interest are difficult to delineate, even manually. The problems linked to segmentation are often due to low contrast, fuzzy contours or too similar intensities with adjacent objects. In some cases, the objects to be extracted have no real contours in the image. This kind of objects is called virtual objects. Virtual objects appear especially in medical applications. To draw them, medical experts usually estimate their position from surrounding objects. The problems related to image segmentation can be greatly simplified with information known in advance on the objects to be extracted (the prior knowledge). A widely used method consists to extract the needed prior knowledge from a reference image often called atlas. The goal of the atlas is to describe the image to be segmented like a map would describe the components of a geographical area. An atlas can contain three types of information on each object being part of the image: an estimation of its position in the image, a description of its shape and texture, and the features of its adjacent objects. The atlas-based segmentation method is rather used when the atlas can characterize a range of images. This method is thus especially adapted to medical images due to the existing consistency between anatomical structures of same type. There exist two types of atlas: the determinist atlas and the statistical atlas. The determinist atlas is an image which has been selected or computed, to be the most representative of an image category to be segmented. This image is called intensity atlas. The contours of the objects of interest (the objects to be extracted in images of the same type) have been traced manually on the intensity atlas, or by using a semi-automatic method. A label is often attributed to each one of these objects in order to differentiate them. In this way, we obtain a labeled version of the atlas called labeled atlas. The statistical atlas is an atlas created from a database of images in order to be the most representative of a certain type of images to be segmented. In this atlas, the position and the features of the objects of interest depend on statistical measures. In this thesis, we are focused on the use of determinist atlases for image segmentation. The segmentation process with a determinist atlas consists to deform the objects delineated in the atlas in order to better align them with their corresponding objects in the image to be segmented. To perform this task, we have distinguished two types of approaches in the literature. The first approach consists to reduce the segmentation problem in an image registration problem. First of all, a dense deformation field that registers (i.e. puts in point-to-point spatial correspondence) the atlas to the image to be segmented, is explicitly computed. Then, this transformation is used to project the assigned labels onto each atlas structure on the image to be segmented. The advantage of this approach is that the deformation field computed from the registration of visible contours allows to easily estimate the position of virtual objects or objects with fuzzy contours. However, the methods currently used for the atlas registration are often only based on the intensity atlas. That means that they do not exploit the object-based information that can be obtained by combining the intensity atlas with its labeled version. In the second approach, the atlas contours selected by the labeled atlas are directly deformed without using a geometrical deformation. For that, this approach is based on matching contour techniques, generally called deformable models. In this thesis, we are interested to a particular type of deformable models, which are the active contour segmentation models. The advantage of the active contour method is that this segmentation technique has been designed to exploit the image information directly linked to the object to be delineated. By using object-based information, active contour models are frequently able to extract regions where the atlas-based segmentation method by registration fails. On the other hand, the result of this local segmentation method is very sensitive to the initial atlas contour position regarding to the target contours. On the other hand, this local segmentation method is very sensitive to the initial position of the atlas contours: the closer they are to the contours to be detected, the more robust the active contour-based segmentation will be. Besides, this segmentation technique needs prior shape models to be able to estimate the position of virtual objects. The main objective of this thesis is to design an algorithm for atlas-based segmentation which combines the advantages of the dense deformation field computed by the registration algorithms, with local segmentation constraints coming from the active contour framework. This implies to design a model where the registration and segmentation by active contours are jointly performed. The atlas registration algorithm that we propose is based on a formulation allowing the integration of any segmentation or contour regularization forces derived from the theory of the active contours in a non parametric registration process. Our algorithm led us to introduce the concept of hierarchical atlas registration. Its principle is that the registration of the main image objects helps the registration of depending objects. This allows to bring progressively the atlas contours closer to their target and thus, to limit the risk to be stuck in a local minimum. Our model had been designed to be easily adaptable to various types of segmentation problems. At the end of the thesis, we present several examples of atlas registration applications in medical imaging. These applications highlight the integration of manual constraints in an atlas registration process, the modeling of a tumor growth in the atlas, the labelization of the thalamus for a statistical study on neuronal connections, the localization of the subthalamic nucleus (STN) for deep brain stimulation (DBS) and the compensation of intra-operative brain shift for neuronavigation systems.

Nonrigid medical image registration

Francisco Javier Sanchez Castro

Over the last two decades medical image processing has become a fundamental tool in medicine, and brain imaging is a paradigm thereof. The fast development of the imaging acquisition techniques allows medical staff to dispose of a variety of data coming from different imaging modalities. However, the analysis of such amount of data based on visual inspection is very difficult and highly time consuming. Often, the information contained in a medical image cannot be entirely seen by a human eye and vice-versa, computers do not have human common sense or the experience acquired by medical experts. Therefore, it is desirable to combine both medical experts and computers in an optimal trade-off to improve the outcome of the wide range of applications where medical imaging is nowadays applied. Among these applications we can mention diagnosis, surgical planning and intra-operative surgical navigation. Image registration is one of the most popular image processing techniques used in clinical environments. It consists of bringing two or more images into a single spatial representation, i.e. finding the optimal geometric transformation between the data. Image registration overcomes the difficulties encountered by clinicians to fuse images mentally since it enables to integrate different images into one representation. The applications of image registration in medicine are numerous and can be grouped in clinical, related to detection and diagnosis, and surgical. Some of these applications are: preprocedural planning and simulation, interventional radiology, diagnostic radiology, minimally invasive procedures, radiation therapy, intraoperative navigation, robot-assisted interventions, etc. During the decade of the 90s a large number of registration algorithms were proposed, linked to a variety of medical applications. Nowadays, rigid registration is considered to be a well-covered domain with efficient solutions for the most common applications. Current research mostly focuses on nonrigid registration methods. However, the complex framework of nonrigid registration makes it a complex problem where there is not a universal solution. This complex framework includes a number of components that interact: the dimensionality of the data involved, the feature space to base the registration on, the kind and domain of the transformation model, the modalities of the data and the subjects involved, the user interaction, the similarity metric to assess the alignment quantitatively, the search space where the optimal solution has to be found, and the search strategy including optimization of the objective function and implementation details. We can summarize the motivation of this thesis in the following sentence: Validation of nonrigid registration algorithms is a very difficult task and rather an application-dependent problem. Therefore, this dissertation is structured to face the problem of brain image registration in three closely correlated parts: algorithms, applications and validation. In the algorithmic context, we propose a control point based nonrigid registration algorithm which aims to improve the performance of classical point based algorithms. In the framework of the implementation of the algorithm we develop a control point detector based on information theory tools and a transformation model based on a type of radial basis functions specially suited for interpolation of scattered and nonuniform distributed data samples. The algorithm is evaluated for typical inter-subject monomodal and multimodal brain image registration. In the context of clinical applications we face a neurosurgical application: the subthalamic nucleus targeting for deep brain stimulation in Parkinson's disease surgery. The extensive study covers from state-of-the-art targeting procedures to automatic targeting methods through different registration algorithms. Moreover, we analyze the influence of surrounding anatomical features on the target location. We also study the feasibility of an atlas of landmarks aiming to improve the targeting accuracy and at the same time to reduce both the computational cost and the preprocessing needed compared to the other automatic targeting methods proposed. All the proposed methods and techniques are statistically cross-validated against state-of-the-art techniques and expert's targeting variability through an application-adapted validation scheme. The results show that automatic subthalamic nucleus localization is possible and as accurate as the methods currently used. We also provide valuable conclusions about the anatomical structures that must be used to infer more accurate targets. In the context of validation of brain image registration, we develop a general (non application-oriented) validation scheme that provides an independent and high dimensional means of assessing the quality of the alignment given by a registration algorithm. This procedure is applied to structural magnetic resonance images and is based on the information carried by the diffusion tensor images. It provides an error measure per voxel allowing an independent evaluation for different regions. Almost 800 registered images are analyzed to provide with strength the conclusions derived from the statistical analysis.

EPFL2008

A Cross Validation Study of Deep Brain Stimulation Targeting: From Experts to Atlas-Based, Segmentation-Based and Automatic Registration Algorithms

Meritxell Bach Cuadra, Olivier Cuisenaire, Reto Meuli, Francisco Javier Sanchez Castro, Jean-Philippe Thiran

Validation of image registration algorithms is a difficult task and open-ended problem, usually application-dependent. In this paper we focus on deep brain stimulation (DBS) targeting for the treatment of movement disorders like Parkinson's disease and essential tremor. DBS involves implantation of an electrode deep inside the brain to electrically stimulate specific areas shutting down the disease's symptoms. The subthalamic nucleus (STN) has turned out to be the optimal target for this kind of surgery. Unfortunately, the STN is in general not clearly distinguishable in common medical imaging modalities. Usual techniques to infer its location are the use of anatomical atlases and visible surrounding landmarks. Surgeons have to adjust the electrode intra-operatively using electrophysiological recordings and macrostimulation tests. We constructed a ground truth derived from specific patients whose STNs are clearly visible on MR T2-weighted images. A patient is chosen as atlas both for the right and left sides. Then, by registering each patient with the atlas using different methods, several estimations of the STN location are obtained. Two studies are driven using our proposed validation scheme. First, a comparison between different atlas-based and non-rigid registration algorithms with a evaluation of their performance and usability to locate the STN automatically. Second, a study of which visible surrounding structures influence the STN location. The two studies are cross-validated between them and against expert's variability. Using this scheme we evaluated the expert's ability against the estimation error provided by the tested algorithms and we demonstrated that automatic STN targeting is possible and as accurate as the expert-driven techniques currently used. We also show which structures have to be taken into account to accurately estimate the STN location.

2006

Hierarchical Image Registration with an Active Contour-Based Atlas Registration Model

Valérie Duay, Nawal Houhou, Jean-Philippe Thiran, Subrahmanyam Venkata Ravi Mohana Sai Gorthi

This paper proposes to apply the non parametric atlas registration framework we have recently developed in [6]. This technique derived from the optical flow model and the active contour framework allows to base the registration of an anatomical atlas on selected structures. A well-suited application of our model is the non rigid registration of medical images based on a hierarchical atlas. This hierarchical registration approach that we have previously introduced in [7], aims to better exploit the spatial dependencies that exist between anatomical structures in an image matching process. Its basic idea is to first register the structures the most relevant to estimate the deformation in order to help the registration of secondary structures. This aims to reduce the risks of mismatching. Here, we propose to test our novel simultaneous registration and segmentation model on different types of medical image registration problems. Results show the advantages to combine our active contour-based registration framework with the structure-based hierarchical approach and highlight the importance of the registration order of the anatomical structures.