Multi-scale label interaction to improve object detection and segmentation

The project aims at advancing the state of the art in the use of context information for classification of image and video data. The use of context in the classification of images has been showed of great importance to improve the performance of actual object recognition systems. In our project we proposed the concept of Multi-scale Feature Labels as a general and compact method to exploit the local and global context. The feature extraction from the discriminative probability or classification confidence label field is of great novelty. Moreover the use of a multi-scale representation of the feature labels lead to a compact and efficient description of the context. The goal of the project has been also to provide a general-purpose method and prove its suitability in different image/video analysis problem. The two-year project generated 5 journal publications (plus 2 under submission), 10 conference publications (plus 2 under submission) and one patent (plus 1 pending). Of these publications, a relevant number make use of the main result of this project to improve the results in detection and/or segmentation of objects.

MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field.

The large spatial inhomogeneity in transmit B(1) field (B(1)(+)) observable in human MR images at high static magnetic fields (B(0)) severely impairs image quality. To overcome this effect in brain T(1)-weighted images, the MPRAGE sequence was modified to generate two different images at different inversion times, MP2RAGE. By combining the two images in a novel fashion, it was possible to create T(1)-weighted images where the result image was free of proton density contrast, T(2) contrast, reception bias field, and, to first order, transmit field inhomogeneity. MP2RAGE sequence parameters were optimized using Bloch equations to maximize contrast-to-noise ratio per unit of time between brain tissues and minimize the effect of B(1)(+) variations through space. Images of high anatomical quality and excellent brain tissue differentiation suitable for applications such as segmentation and voxel-based morphometry were obtained at 3 and 7 T. From such T(1)-weighted images, acquired within 12 min, high-resolution 3D T(1) maps were routinely calculated at 7 T with sub-millimeter voxel resolution (0.65-0.85 mm isotropic). T(1) maps were validated in phantom experiments. In humans, the T(1) values obtained at 7 T were 1.15+/-0.06 s for white matter (WM) and 1.92+/-0.16 s for grey matter (GM), in good agreement with literature values obtained at lower spatial resolution. At 3 T, where whole-brain acquisitions with 1 mm isotropic voxels were acquired in 8 min, the T(1) values obtained (0.81+/-0.03 s for WM and 1.35+/-0.05 for GM) were once again found to be in very good agreement with values in the literature.

Landmark detection for fusion of fundus and MRI toward a patient-specific multimodal eye model.

Ophthalmologists typically acquire different image modalities to diagnose eye pathologies. They comprise, e.g., Fundus photography, optical coherence tomography, computed tomography, and magnetic resonance imaging (MRI). Yet, these images are often complementary and do express the same pathologies in a different way. Some pathologies are only visible in a particular modality. Thus, it is beneficial for the ophthalmologist to have these modalities fused into a single patient-specific model. The goal of this paper is a fusion of Fundus photography with segmented MRI volumes. This adds information to MRI that was not visible before like vessels and the macula. This paper contributions include automatic detection of the optic disc, the fovea, the optic axis, and an automatic segmentation of the vitreous humor of the eye.

Dense deformation field estimation for atlas-based segmentation of pathological MR brain images.

Atlas registration is a recognized paradigm for the automatic segmentation of normal MR brain images. Unfortunately, atlas-based segmentation has been of limited use in presence of large space-occupying lesions. In fact, brain deformations induced by such lesions are added to normal anatomical variability and they may dramatically shift and deform anatomically or functionally important brain structures. In this work, we chose to focus on the problem of inter-subject registration of MR images with large tumors, inducing a significant shift of surrounding anatomical structures. First, a brief survey of the existing methods that have been proposed to deal with this problem is presented. This introduces the discussion about the requirements and desirable properties that we consider necessary to be fulfilled by a registration method in this context: To have a dense and smooth deformation field and a model of lesion growth, to model different deformability for some structures, to introduce more prior knowledge, and to use voxel-based features with a similarity measure robust to intensity differences. In a second part of this work, we propose a new approach that overcomes some of the main limitations of the existing techniques while complying with most of the desired requirements above. Our algorithm combines the mathematical framework for computing a variational flow proposed by Hermosillo et al. [G. Hermosillo, C. Chefd'Hotel, O. Faugeras, A variational approach to multi-modal image matching, Tech. Rep., INRIA (February 2001).] with the radial lesion growth pattern presented by Bach et al. [M. Bach Cuadra, C. Pollo, A. Bardera, O. Cuisenaire, J.-G. Villemure, J.-Ph. Thiran, Atlas-based segmentation of pathological MR brain images using a model of lesion growth, IEEE Trans. Med. Imag. 23 (10) (2004) 1301-1314.]. Results on patients with a meningioma are visually assessed and compared to those obtained with the most similar method from the state-of-the-art.

Bayesian image reconstruction with space-variant noise suppression

In this paper we present a Bayesian image reconstruction algorithm with entropy prior (FMAPE) that uses a space-variant hyperparameter. The spatial variation of the hyperparameter allows different degrees of resolution in areas of different statistical characteristics, thus avoiding the large residuals resulting from algorithms that use a constant hyperparameter. In the first implementation of the algorithm, we begin by segmenting a Maximum Likelihood Estimator (MLE) reconstruction. The segmentation method is based on using a wavelet decomposition and a self-organizing neural network. The result is a predetermined number of extended regions plus a small region for each star or bright object. To assign a different value of the hyperparameter to each extended region and star, we use either feasibility tests or cross-validation methods. Once the set of hyperparameters is obtained, we carried out the final Bayesian reconstruction, leading to a reconstruction with decreased bias and excellent visual characteristics. The method has been applied to data from the non-refurbished Hubble Space Telescope. The method can be also applied to ground-based images.

An Efficient Segmentation Method for Ultrasound Images based on a Semi-supervised Approach and Patch-based Features

Segmenting ultrasound images is a challenging problemwhere standard unsupervised segmentation methods such asthe well-known Chan-Vese method fail. We propose in thispaper an efficient segmentation method for this class ofimages. Our proposed algorithm is based on asemi-supervised approach (user labels) and the use ofimage patches as data features. We also consider thePearson distance between patches, which has been shown tobe robust w.r.t speckle noise present in ultrasoundimages. Our results on phantom and clinical data show avery high similarity agreement with the ground truthprovided by a medical expert.

Brain Surface Segmentation of Magnetic Resonance Images of the Fetus

In this work we present a method for the image analysisof Magnetic Resonance Imaging (MRI) of fetuses. Our goalis to segment the brain surface from multiple volumes(axial, coronal and sagittal acquisitions) of a fetus. Tothis end we propose a two-step approach: first, a FiniteGaussian Mixture Model (FGMM) will segment the image into3 classes: brain, non-brain and mixture voxels. Second, aMarkov Random Field scheme will be applied tore-distribute mixture voxels into either brain ornon-brain tissue. Our main contributions are an adaptedenergy computation and an extended neighborhood frommultiple volumes in the MRF step. Preliminary results onfour fetuses of different gestational ages will be shown.

A post-processing pipeline to reconstruct the developing fetal brain using low-resolution MRI

Introduction. Development of the fetal brain surfacewith concomitant gyrification is one of the majormaturational processes of the human brain. Firstdelineated by postmortem studies or by ultrasound, MRIhas recently become a powerful tool for studying in vivothe structural correlates of brain maturation. However,the quantitative measurement of fetal brain developmentis a major challenge because of the movement of the fetusinside the amniotic cavity, the poor spatial resolution,the partial volume effect and the changing appearance ofthe developing brain. Today extensive efforts are made todeal with the âeurooepost-acquisitionâeuro reconstruction ofhigh-resolution 3D fetal volumes based on severalacquisitions with lower resolution (Rousseau, F., 2006;Jiang, S., 2007). We here propose a framework devoted tothe segmentation of the basal ganglia, the gray-whitetissue segmentation, and in turn the 3D corticalreconstruction of the fetal brain. Method. Prenatal MRimaging was performed with a 1-T system (GE MedicalSystems, Milwaukee) using single shot fast spin echo(ssFSE) sequences in fetuses aged from 29 to 32gestational weeks (slice thickness 5.4mm, in planespatial resolution 1.09mm). For each fetus, 6 axialvolumes shifted by 1 mm were acquired (about 1 min pervolume). First, each volume is manually segmented toextract fetal brain from surrounding fetal and maternaltissues. Inhomogeneity intensity correction and linearintensity normalization are then performed. A highspatial resolution image of isotropic voxel size of 1.09mm is created for each fetus as previously published byothers (Rousseau, F., 2006). B-splines are used for thescattered data interpolation (Lee, 1997). Then, basalganglia segmentation is performed on this superreconstructed volume using active contour framework witha Level Set implementation (Bach Cuadra, M., 2010). Oncebasal ganglia are removed from the image, brain tissuesegmentation is performed (Bach Cuadra, M., 2009). Theresulting white matter image is then binarized andfurther given as an input in the Freesurfer software(http://surfer.nmr.mgh.harvard.edu/) to provide accuratethree-dimensional reconstructions of the fetal brain.Results. High-resolution images of the cerebral fetalbrain, as obtained from the low-resolution acquired MRI,are presented for 4 subjects of age ranging from 29 to 32GA. An example is depicted in Figure 1. Accuracy in theautomated basal ganglia segmentation is compared withmanual segmentation using measurement of Dice similarity(DSI), with values above 0.7 considering to be a verygood agreement. In our sample we observed DSI valuesbetween 0.785 and 0.856. We further show the results ofgray-white matter segmentation overlaid on thehigh-resolution gray-scale images. The results arevisually checked for accuracy using the same principlesas commonly accepted in adult neuroimaging. Preliminary3D cortical reconstructions of the fetal brain are shownin Figure 2. Conclusion. We hereby present a completepipeline for the automated extraction of accuratethree-dimensional cortical surface of the fetal brain.These results are preliminary but promising, with theultimate goal to provide âeurooemovieâeuro of the normal gyraldevelopment. In turn, a precise knowledge of the normalfetal brain development will allow the quantification ofsubtle and early but clinically relevant deviations.Moreover, a precise understanding of the gyraldevelopment process may help to build hypotheses tounderstand the pathogenesis of several neurodevelopmentalconditions in which gyrification have been shown to bealtered (e.g. schizophrenia, autismâeuro¦). References.Rousseau, F. (2006), 'Registration-Based Approach forReconstruction of High-Resolution In Utero Fetal MR Brainimages', IEEE Transactions on Medical Imaging, vol. 13,no. 9, pp. 1072-1081. Jiang, S. (2007), 'MRI of MovingSubjects Using Multislice Snapshot Images With VolumeReconstruction (SVR): Application to Fetal, Neonatal, andAdult Brain Studies', IEEE Transactions on MedicalImaging, vol. 26, no. 7, pp. 967-980. Lee, S. (1997),'Scattered data interpolation with multilevel B-splines',IEEE Transactions on Visualization and Computer Graphics,vol. 3, no. 3, pp. 228-244. Bach Cuadra, M. (2010),'Central and Cortical Gray Mater Segmentation of MagneticResonance Images of the Fetal Brain', ISMRM Conference.Bach Cuadra, M. (2009), 'Brain tissue segmentation offetal MR images', MICCAI.

ZHARP: three-dimensional motion tracking from a single image plane.

Three-dimensional imaging and quantification of myocardial function are essential steps in the evaluation of cardiac disease. We propose a tagged magnetic resonance imaging methodology called zHARP that encodes and automatically tracks myocardial displacement in three dimensions. Unlike other motion encoding techniques, zHARP encodes both in-plane and through-plane motion in a single image plane without affecting the acquisition speed. Postprocessing unravels this encoding in order to directly track the 3-D displacement of every point within the image plane throughout an entire image sequence. Experimental results include a phantom validation experiment, which compares zHARP to phase contrast imaging, and an in vivo study of a normal human volunteer. Results demonstrate that the simultaneous extraction of in-plane and through-plane displacements from tagged images is feasible.

Influence of scatter reduction method and monochromatic beams on image quality and dose in mammography.

In mammography, the image contrast and dose delivered to the patient are determined by the x-ray spectrum and the scatter to primary ratio S/P. Thus the quality of the mammographic procedure is highly dependent on the choice of anode and filter material and on the method used to reduce the amount of scattered radiation reaching the detector. Synchrotron radiation is a useful tool to study the effect of beam energy on the optimization of the mammographic process because it delivers a high flux of monochromatic photons. Moreover, because the beam is naturally flat collimated in one direction, a slot can be used instead of a grid for scatter reduction. We have measured the ratio S/P and the transmission factors for grids and slots for monoenergetic synchrotron radiation. In this way the effect of beam energy and scatter rejection method were separated, and their respective importance for image quality and dose analyzed. Our results show that conventional mammographic spectra are not far from optimum and that the use of a slot instead of a grid has an important effect on the optimization of the mammographic process. We propose a simple numerical model to quantify this effect.

Segmentation of Head and Neck Lymph Node Regions for Radiotherapy Planning Using Active Contour-Based Atlas Registration

In this paper, we present the segmentation of the headand neck lymph node regions using a new active contourbased atlas registration model. We propose to segment thelymph node regions without directly including them in theatlas registration process; instead, they are segmentedusing the dense deformation field computed from theregistration of the atlas structures with distinctboundaries. This approach results in robust and accuratesegmentation of the lymph node regions even in thepresence of significant anatomical variations between theatlas-image and the patient's image to be segmented. Wealso present a quantitative evaluation of lymph noderegions segmentation using various statistical as well asgeometrical metrics: sensitivity, specificity, dicesimilarity coefficient and Hausdorff distance. Acomparison of the proposed method with two other state ofthe art methods is presented. The robustness of theproposed method to the atlas selection, in segmenting thelymph node regions, is also evaluated.

Landmark detection for fusion of fundus and MRI toward a patient-specific multimodal eye model.

Ophthalmologists typically acquire different image modalities to diagnose eye pathologies. They comprise, e.g., Fundus photography, optical coherence tomography, computed tomography, and magnetic resonance imaging (MRI). Yet, these images are often complementary and do express the same pathologies in a different way. Some pathologies are only visible in a particular modality. Thus, it is beneficial for the ophthalmologist to have these modalities fused into a single patient-specific model. The goal of this paper is a fusion of Fundus photography with segmented MRI volumes. This adds information to MRI that was not visible before like vessels and the macula. This paper contributions include automatic detection of the optic disc, the fovea, the optic axis, and an automatic segmentation of the vitreous humor of the eye.

Occupational and patient exposure as well as image quality for full spine examinations with the EOS imaging system.

PURPOSE: EOS (EOS imaging S.A, Paris, France) is an x-ray imaging system that uses slot-scanning technology in order to optimize the trade-off between image quality and dose. The goal of this study was to characterize the EOS system in terms of occupational exposure, organ doses to patients as well as image quality for full spine examinations. METHODS: Occupational exposure was determined by measuring the ambient dose equivalents in the radiological room during a standard full spine examination. The patient dosimetry was performed using anthropomorphic phantoms representing an adolescent and a five-year-old child. The organ doses were measured with thermoluminescent detectors and then used to calculate effective doses. Patient exposure with EOS was then compared to dose levels reported for conventional radiological systems. Image quality was assessed in terms of spatial resolution and different noise contributions to evaluate the detector's performances of the system. The spatial-frequency signal transfer efficiency of the imaging system was quantified by the detective quantum efficiency (DQE). RESULTS: The use of a protective apron when the medical staff or parents have to stand near to the cubicle in the radiological room is recommended. The estimated effective dose to patients undergoing a full spine examination with the EOS system was 290μSv for an adult and 200 μSv for a child. MTF and NPS are nonisotropic, with higher values in the scanning direction; they are in addition energy-dependent, but scanning speed independent. The system was shown to be quantum-limited, with a maximum DQE of 13%. The relevance of the DQE for slot-scanning system has been addressed. CONCLUSIONS: As a summary, the estimated effective dose was 290μSv for an adult; the image quality remains comparable to conventional systems.