Catálogo de publicaciones - libros

Segmenting cortical and sub-cortical structures from 3D brain images is of significant practical importance. However, various anatomical structures have similar intensity patterns in MRI, and the automatic segmentation of them is a challenging task. In this paper, we present a new brain segmentation algorithm using a hybrid model. (1) A multi-class classifier, PBT.M2, is proposed for learning/computing multi-class discriminative models. The PBT.M2 handles multi-class patterns more easily than the original probabilistic boosting tree (PBT) [11], and it facilitates the process, eventually, toward whole brain segmentation. (2) We use an edge field, by learning, to constraint the region boundaries. We show the improvements due to the two new aspects both numerically and visually, and also compare the results with those by FreeSurfer [2]. Our algorithm is general and easy to use, and the results obtained are encouraging.

Principal Component Analysis (PCA) has been widely used for dimensionality reduction in shape and appearance modeling. There have been several attempts of making PCA robust against outliers. However, there are cases in which a small subset of samples may appear as outliers and still correspond to plausible data. The example of shapes corresponding to fractures when building a vertebra shape model is addressed in this study. In this case, the modeling of “outliers” is important, and it might be desirable not only not to disregard them, but even to enhance their importance.

A variation on PCA that deals naturally with the importance of outliers is presented in this paper. The technique is utilized for building a shape model of a vertebra, aiming at segmenting the spine out of lateral X-ray images. The results show that the algorithm can implement both an outlier-enhancing and a robust PCA. The former improves the segmentation performance in fractured vertebrae, while the latter does so in the unfractured ones.

To perform cartilage quantitative analysis requires the accurate segmentation of each individual cartilage. In this paper we present a model based scheme that can automatically and accurately segment each individual cartilage in healthy knees from a clinical MR sequence (fat suppressed spoiled gradient recall). This scheme consists of three stages; the automatic segmentation of the bones, the extraction of the bone-cartilage interfaces (BCI) and segmentation of the cartilages. The bone segmentation is performed using three-dimensional active shape models. The BCI is extracted using image information and prior knowledge about the likelihood of each point belonging to the interface. A cartilage thickness model then provides constraints and regularizes the cartilage segmentation performed from the BCI. The accuracy and robustness of the approach was experimentally validated, with (patellar, tibial and femoral) cartilage segmentations having a median DSC of (0.870, 0.855, 0.870), performing significantly better than non-rigid registration (0.787, 0.814, 0.795). The total cartilage segmentation had an average DSC of (0.891), close to the (0.896) obtained using a semi-automatic watershed algorithm. The error in quantitative volume and thickness measures was (8.29, 4.94, 5.56)% and (0.19, 0.33, 0.10) respectively.

We present a new model-based approach for an automated labeling and segmentation of the rib cage in chest CT scans. A mean rib cage model including a complete vertebral column is created out of 29 data sets. We developed a ray search based procedure for rib cage detection and initial model pose. After positioning the model, it was adapted to 18 unseen CT data. In 16 out of 18 data sets, detection, labeling, and segmentation succeeded with a mean segmentation error of less than 1.3 mm between true and detected object surface. In one case the rib cage detection failed, in another case the automated labeling.

Radiotherapy planning needs accurate delineations of the critical structures. Atlas-based segmentation has been shown to be very efficient to delineate brain structures [1]. However, the construction of an atlas from a dataset of images [2], particularly for the head and neck region, is very difficult due to the high variability of the images and can generate over-segmented structures in the atlas. To overcome this drawback, we present in this paper an alternative method to select as a template the image in a database that is the most similar to the patient to be segmented. This similarity is based on a distance between transformations. A major contribution is that we do not compute every patient-to-sample registration to find the most similar template, but only the registration of the patient towards an average image. This method has therefore the advantage of being computationally very efficient. We present a qualitative and quantitative comparison between the proposed method and a classical atlas-based segmentation method. This evaluation is performed on a subset of 45 patients using a Leave-One-Out method and shows a great improvement of the specificity of the results.

This paper describes an algorithm for unbiased construction of white matter (WM) atlases using full information available to diffusion tensor (DT) images. The key component of the proposed algorithm is a novel DT image registration method that leverages metrics comparing tensors as a whole and optimizes tensor orientation explicitly. The problem of unbiased atlas construction is formulated using the approach proposed by Joshi et al., i.e., the unbiased WM atlas is determined by finding the mappings that best match the atlas to the images in the population and have the least amount of deformation. We show how the proposed registration algorithm can be adapted to approximately find the optimal atlas. The utility of the proposed approach is demonstrated by constructing a WM atlas of 13 subjects. The presented DT registration method is also compared to the approach of matching DT images by aligning their fractional anisotropy images using large-deformation image registration methods. Our results suggest that using full tensor information can better align the orientations of WM fiber bundles.

In this paper we present a technique for fully automatic, real-time 3D SPECT (Single Photon Emitting Computed Tomography) and 2D ultrasound image registration. We use this technique in the context of kidney lesion diagnosis. Our registration algorithm allows a physician to perform an ultrasound exam after a SPECT image has been acquired and see in real time the registration of both modalities. An automatic segmentation algorithm has been implemented in order to display in 3D the positions of the acquired US images with respect to the organs.

Abdominal Aortic Aneurysm is a potentially life-threatening disease if not treated adequately. Its pathogenesis is complex and multifactorial and is still not fully understood. Many biochemical and biomechanical mechanisms have been identified as playing a role in the formation of aneurysms but it is as yet unclear what triggers the process. We investigated the role of the relevant biomechanical factors, in particular the wall shear stress and the intramural wall stress by simulating fluid structure interaction between the blood flow and the deforming arterial wall in a healthy abdominal aortic bifurcation, the preferred location of the disease. We then extended this study by introducing a hypothetical weakening of the aortic wall. Intramural wall stress was considerably higher and wall shear stress considerably lower in this configuration, supporting the hypothesis that biomechanical aneurysmal growth factors are self-sustaining.

Acute rejection is the most common reason of graft failure after kidney transplantation and early detection is crucial to survive the transplanted kidney function. In this paper, we introduce a new approach for the automatic classification of normal and acute rejection transplants from Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI). The proposed algorithm consists of three main steps; the first step isolates the kidney from the surrounding anatomical structures. In the second step, new motion correction models are employed to account for both the global and local motion of the kidney due to patient moving and breathing. Finally, the perfusion curves that show the transportation of the contrast agent into the tissue are obtained from the kidney and used in the classification of normal and acute rejection transplants. .

An image registration-based elastography algorithm is presented for assessing the stiffness of tissue regions inside the prostate for the purpose of detecting tumors. A 3D finite-element model of the prostate is built from ultrasound images and used to simulate the deformation of the prostate induced by a TRUS probe. To reconstruct the stiffness of tissues, their Young’s moduli are varied using Powell’s method so that the mutual information between a simulated and deformed image volume is maximized. The algorithm was validated using a gelatin prostate phantom embedded with a cylindrical inclusion that simulated a tumor. Results from the phantom study showed that the technique could detect the increased stiffness of the simulated tumor with a reasonable accuracy.