Catálogo de publicaciones - libros

This paper presents a novel approach that three-dimensionally visualizes and evaluates stenoses in human coronary arteries by using harmonic skeletons. A harmonic skeleton is the center line of a multi-branched tubular surface extracted based on a harmonic function, which is the solution of the Laplace equation. This skeletonization method guarantees smoothness and connectivity and provides a fast and straightforward way to calculate local cross-sectional areas of the arteries, and thus provides the possibility to localize and evaluate coronary artery stenosis, which is a commonly seen pathology in coronary artery disease.

Although significant effort has been spent over the past decades to develop innovative image processing algorithms and to improve existing methods in terms of precision, reproducibility and computational efficiency, relatively few research was undertaken to find out to what extent the validity of results obtained with these methods is limited by inherent imperfections of the input images. This observation is especially true for MRI based morphometry, which aims at precise and highly reproducible determination of geometrical properties of anatomical structures despite the fact that MR images are geometrically distorted. We here present (a) a method for characterization of site-specific geometrical distortions and (b) the results of a long term study designed to find out how precisely geometrical properties and morphological changes of brain structures can, in principle, be detected in images acquired with MRI scanners. Due to the long-term character of our study, our findings include effects resulting from limited hardware stability as well as from variations in patient positioning. Our results show that these effects can be strong enough to substantially confound MRI studies of small morphological changes.

In any medical domain, it is common to have more than one test (classifier) to diagnose a disease. In image analysis, for example, there is often more than one reader or more than one algorithm applied to a certain data set. Combining of classifiers is often helpful, but determining the way in which classifiers should be combined is not trivial. Standard strategies are based on learning classifier combination functions from data. We describe a simple strategy to combine results from classifiers that have not been applied to a common data set, and therefore can not undergo this type of joint training. The strategy, which assumes conditional independence of classifiers, is based on the calculation of a combined Receiver Operating Characteristic (ROC) curve, using maximum likelihood analysis to determine a combination rule for each ROC operating point. We offer some insights into the use of ROC analysis in the field of medical imaging.

In this paper, we present an evaluation of seven automatic brain tissue classifiers based on level of agreements. A number of agreement measures are explained, and we show how they can be used to compare different segmentation techniques. We use the Simultaneous Truth and Performance Level Estimation (STAPLE) of Warfield et al. but also introduce a novel evaluation technique based on the Williams’ index. The methods are evaluated using these two techniques on a population of forty subjects, each having an SPGR scan and a co-registered T2 weighted scan. We provide an interpretation of the results and show how similar the output of the STAPLE analysis and Williams’ index are. When no ground truth is required, we recommend the use of Williams’ index as it is easy and fast to compute.

We propose a novel method for the validation of vascular segmentations. Our technique combines morphological operators and the STAPLE algorithm to obtain ground truth of centerline extractions as well as a measure of accuracy of the methods to be compared. Moreover, our method can be extended to the validation of any open-curves. We also present a comparison study of three vascular segmentation methods: ridge traversal, statistical and curves level set. They are compared with manual segmentations from five experts.

In this work, we aim at validating some soft tissue deformation models using high resolution Micro Computed Tomography (Micro-CT) and medium resolution Cone-Beam CT (CBCT) images. These imaging techniques play a key role in detecting the tissue deformation details in the contact region between the tissue and the surgical tool (probe) even for small force loads, and provide good capabilities for creating accurate 3D models of tissues. Surgical simulations rely on accurate representation of the mechanical response of soft tissues subjected to surgical manipulations. Several finite element (F.E.) models have been suggested to characterize soft tissues. However, validating these models for specific tissues still remains a challenge. For our validation, ex vivo lamb liver is chosen to validate the linear elastic model (LEM), the linear viscoelastic model (LVEM), and the neo-Hooke hyperelastic model (NHM). We found that the LEM is more applicable to lamb liver than the LVEM for small force loads (<40) and that the NHM is closer to reality than the LVEM for this same range of force loads.

We provided in [14] an augmented reality guidance system for liver punctures, which has been validated on a static abdominal phantom [16]. In this paper, we report the first in vivo experiments.

We developed a strictly passive protocol to directly evaluate our system on patients. We show that the system algorithms work efficiently and we highlight the clinical constraints that we had to overcome (small operative field, weight and sterility of the tracked marker attached to the needle...). Finally, we investigate to what extent breathing motion can be neglected for free breathing patient. Results show that the guiding accuracy, close to 1 cm, is sufficient for large targets only (above 3 cm of diameter) when the breathing motion is neglected. In the near future, we aim at validating our system on smaller targets using a respiratory gating technique.

This paper describes a novel approach to forming high resolution MR images of the human fetal brain. It addresses the key problem of motion of the fetus by proposing a registration refined compounding of multiple sets of orthogonal fast 2D MRI slices, that are currently acquired for clinical studies, into a single high resolution MRI volume. A robust multi-resolution slice alignment is applied iteratively to the data to correct motion of the fetus that occurs between 2D acquisitions. This is combined with an intensity correction step and a super resolution reconstruction step, to form a single high isotropic resolution volume of the fetal brain. Experimental validation on synthetic image data with known motion types and underlying anatomy, together with retrospective application to sets of clinical acquisitions are included. Results indicate the method promises a unique route to acquiring high resolution MRI of the fetal brain in vivo allowing comparable quality to that of neonatal MRI. Such data is highly valuable in allowing a clinically applicable window into the process of normal and abnormal brain development.

Current methods of four-dimensional (4D) CT imaging of the thorax synchronise the acquisition with a respiratory signal to restrospectively sort acquired data. The quality of the 4D images relies on an accurate description of the position of the thorax in the respiratory cycle by the respiratory signal. Most of the methods used an external device for acquiring the respiratory signal. We propose to extract it directly from thorax cone-beam (CB) CT projections. This study implied two main steps: the simulation of a set of CBCT projections, and the extraction, selection and integration of motion information from the simulation output to obtain the respiratory signal. A real respiratory signal was used for simulating the CB acquisition of a breathing patient. We extracted from CB images a respiratory signal with 96.4% linear correlation with the reference signal, but we showed that other measures of the quality of the extracted respiratory signal were required.

Liver transplantation affords a unique opportunity to assess and improve radiological imaging of the liver, as the full explanted liver is available for review and comparison. Quantitative comparison between the explanted liver and in vivo images acquired prior to transplantation requires accurate registration of images of the explanted liver to the radiological images. However, this registration problem is challenging because the orientation change and the deformation magnitude between the two image sets exceed the level assumed for most registration algorithms. This paper suggests a two-step registration process to overcome the difficulty: to first align the orientation of 3D liver models built from two sets of image data using maximum volume overlap as their similarity measurement, and second to deform one model to match the other. The key contribution of this paper is that it utilizes the global volumetric information and the asymmetry property of the liver model to determinately provide a simple and reliable initial point for further deformable model based registration. Our experimental data demonstrate the effectiveness of this approach.