Search

A method by which to register multimodality medical images accommodating soft tissue deformation is presented in the context of interventional therapy with a MR scanner. Accuracy testing with arbitrarily deformed MR images and application studies of a pig’s brain were undertaken to evaluate the feasibility of the method. When Mutual Information is employed as the voxel similarity measure in the matching energy function, the algorithm can accommodate multimodality images. Coupled with rigid registration, the deformable registration of pre- and intraoperative multi-modality images enables surgeons to precisely define critical anatomical structures, such as vessels and functional areas, and to localize and optimize trajectories. The method directly and automatically works on volumetric multimodality images. Thus the algorithm is suitable for intraoperative registration, where stability and simplicity are desirable.

We have initiated studies towards the design and implementation of an ophthalmic augmented reality environment in order to allow for a) more precise laser treatment for ophthalmic diseases, b) teaching, c) telemedicine, and d) real-time image measurement, analysis, and comparison. The proposed system is being designed around a standard slit-lamp biomicroscope. The microscope will be interfaced to a CCD camera, and the image sent to a video capture board. A single computer workstation will coordinate image capture, registration, and display. The captured image is registered with previously stored, montaged photographic and angiographic data, with superposition facilitated by funduslandmark-based fast registration algorithms. The computer then drives a high intensity, VGA resolution video display with adjustable brightness and contrast attached to one of the oculars of the slitlamp biomicroscope. Preliminary studies with a modified binocular operating microscope interfaced to a Sun Ultral Workstation and an IBM-compatible PC demonstrates proof-of-principle. Robust, accurate fundus image montaging is accomplished with Hausdorff-distance-based methods. For photographic and angiographic data where the vessel gray levels vary from light to dark, and intensity-based correlation methods fail, image-preprocessing with smoothing, edge-detection, and thresholding facilitates registration. Non-real-time registration (∼ 0.4–4.0 CPU seconds) is achieved by non-optimized simple template matching (translation only, Matrox Inspector) or Hausdorff-distance-based (translation, rotation, and scale) algorithms performed on edge-detected fundus photographic and angiographic images, and on images of a model eye. Successful registration and image overlay of color, monochromatic, and angiographic images is demonstrated. To our knowledge, these studies represent the first investigation towards design and implementation of an ophthalmic augmented reality environment.

Purpose: Definitive radiation therapy for locally advanced cervical cancer involves both external beam radiation therapy (EBRT) and high-dose-rate (HDR) brachytherapy. There remains controversy and practice pattern variation for the optimal radiation prescription dose to metastatic pelvic lymph nodes. This study investigates the contribution of the pelvic lymph node dose from HDR brachytherapy. We also examine the variation of the lymph node prescribed dose and actual delivered dose as determined from the EBRT and brachytherapy plans. Materials and Methods: From September 2007 to October 2011, 68 patients were treated with curative intent. Of these, 17 patients with 36 positive pelvic lymph nodes were included in this retrospective analysis. All patients were treated with EBRT to the pelvis with a supplemental boost to the involved pelvic node, plus HDR brachytherapy. Pathologically involved lymph nodes were contoured on the planning EBRT image as well as the 4 to 5 brachytherapy planning images. The mean received dose of each lymph node from the EBRT and brachytherapy plans was recorded. The Equivalent Dose in 2-Gray Fractions (EQD2) was calculated using the equation EQD2 5 D [(d þ a/b)/2 Gy þ a/b] where D is total dose, d is dose per fraction and a/bratio is the dose at which the linear and quadratic components are equal. A student t-test was performed to determine if the mean received dose was significantly different from the mean intended prescribed dose and the mean intended EQD2. Results: The average prescribed dose from the EBRT, including the initial pelvic fields and boost contribution, was 54.09Gy. The average prescribed HDR dose to International Commission on Radiation Units (ICRU) point A was 26.81Gy. The average dose delivered to the involved pelvic lymph nodes from EBRT and brachytherapy were 54.25Gy and 4.31Gy, respectively, with the corresponding EQD2 of 53.45Gy and 4.00Gy. Therefore, there was no statistically significant difference between the means of the received individual lymph node dose and intended prescribed pelvic lymph node dose for the EBRT and brachytherapy plans (p! 0.05). The similar results between the received dose and the EQD2 were observed. Conclusions: The equivalent dose contribution to the involved pelvic lymph nodes in locally advanced cervical cancer from HDR brachytherapy was 4.00Gy. Our study shows that is the HDR contribution is 7% of the total EQD2 (57.45Gy). The HDR contribution needs to be accounted for when prescribing the EBRT boost dose to pelvic lymph nodes for the optimal therapeutic dose.

Applying model-based vision techniques to SAR data is particularly challenging because of the inherent difficulty in generating accurate predictions of an electromagnetic signature and the variation of observed signatures to small changes in sensing conditions, imaging geometry, and object characteristics. In order to cope with these difficulties we are developing a robust feature matching model to be part of the moving and stationary target acquisition and recognition model-based automatic target recognition system. The goals of this matching module are: (1) generate correspondences between predicted features and features extracted from a SAR image, (2) evaluate the match based on the degree of uncertainty of the features and their degree of match, (3) refine the target position/orientation/articulation based on the feature correspondences, and (4) analyze residual mix- matches for cueing scene interpretations of unexplained image features. We are developing a probabilistic optimization matching approach based on a (1) Bayesian evaluation metric and (2) they dynamic solution of the best correspondences during the search of pose space. The system is designed to support a wide range of features (points, regions, and other composite features) in a wide range of situations, such as obscuration, attenuation, layover, and variable target articulations and configurations. Initial test results in these types of situations are presented.

A new information-theoretic approach is presented for finding the registration of volumetric medical images of differing modalities. Registration is achieved by adjustment of the relative position and orientation until the mutual information between the images is maximized. In our derivation of the registration procedure, few assumptions are made about the nature of the imaging process. As a result the algorithms are quite general and can foreseeably be used with a wide variety of imaging devices. This approach works directly with image data; no pre-processing or segmentation is required. This technique is, however, more flexible and robust than other intensity-based techniques like correlation. Additionally, it has an efficient implementation that is based on stochastic approximation. Experiments are presented that demonstrate the approach registering magnetic resonance (MR) images with computed tomography (CT) images, and with positron-emission tomography (PET) images. Surgical applications of the registration method are described.

There is a need for frameless guidance systems to help surgeons plan the exact location for incisions, to define the margins of tumors, and to precisely identify locations of neighboring critical structures. The authors have developed an automatic technique for registering clinical data, such as segmented magnetic resonance imaging (MRI) or computed tomography (CT) reconstructions, with any view of the patient on the operating table. The authors demonstrate on the specific example of neurosurgery. The method enables a visual mix of live video of the patient and the segmented three-dimensional (3-D) MRI or CT model. This supports enhanced reality techniques for planning and guiding neurosurgical procedures and allows us to interactively view extracranial or intracranial structures nonintrusively. Extensions of the method include image guided biopsies, focused therapeutic procedures, and clinical studies involving change detection over time sequences of images.

We have combined three-dimensional (3D) computer-reconstructed neuroimages with a novel video registration technique for virtual reality-based, image-guided surgery of the brain and spine. This technique allows the surgeon to localize cerebral and spinal lesions by superimposing a 3D-reconstructed MR or CT scan on a live video image of the patient. Once the patient''s scan has been segmented into the relevant components (e.g., tumor, edema, ventricles, arteries, brain and skin), the surgeon studies the 3D anatomy to determine the optimal surgical approach. The proposed intraoperative surgeon''s perspective is displayed in the operating room at the time of surgery using a portable workstation. The patient is then brought to the operating room and positioned according to the planned approach. A video camera is trained on the patient from the proposed intraoperative surgeon''s perspective. A video mixer merges the images from the video camera and the 3D computer reconstruction. This video mixer can vary the output intensity of the two input images between 100% of either and 50% of both. This visually superimposes the two images, not unlike a photographic double exposure. The patient''s position and the 3D reconstruction are then adjusted until the images on the video mixer''s output monitor are identical in terms of scale, position and rotation. This superimposition is facilitated by aligning various surface landmarks such as the external auditory canal, lateral canthus, and nasion. In some cases, such as with spinal tumors, capsules placed on the skin prior to scanning serve as fiducials. After alignment of the video and computer skin images, the computer image of the skin is selectively deleted leaving the 3D image of the underlying brain or spine superimposed on the video image of the patient''s skin. The borders of the tumor and important cortical sulci or spinal anatomy may then be outlined on the patient''s skin using indelible markers. These markings allow the surgeon to plan an adequate opening with minimal exposure of adjacent structures. So far we have used this technique to localize seven brain tumors and one intradural, extramedullary spinal tumor. In addition, the same technique was used to guide the repositioning of the bones in a reconstructive craniofacial surgery. In each case we found excellent correlation between the operative findings and the predicted anatomy. No patients suffered any permanent neurologic deficit.

Purpose: To evaluate the use of a non‐rigid registration method for improved intra‐operative target definition in 125‐I permanent prostate implants. Methods: The validation dataset was created from 10 MRI‐guided prostate biopsy patients with both diagnostic (with endorectal coil) and intra‐procedural (without endorectal coil) 3 Tesla MRI scans under an IRB approved protocol. A biomechanical‐based probabilistic non‐rigid registration method was adapted to register diagnostic to intra‐procedural images by matching the contoured prostate boundaries. The probabilistic framework provides a collection of prostate configurations under deformation, or a marginal probability map for a given location to be identified as inside the prostate by the registration. The marginal probability map was compared to the intra‐procedural prostate contour. 125‐I treatment plans were generated for the intra‐procedural scans. Results: In two cases we find that ∼5% of the prostate volume within the 50th percentile of the marginal deformed probability is not included in the intra‐procedural contour, and 3–4% of the intra‐procedural prostate volume is not included in any of the deformed configurations. Different regions of the prostate have varying uncertainties; preliminary results show margins of 1–2 mm near mid‐gland and 3–4 mm around the apex. We present geometric uncertainties and the resulting variation of dosimetric quantities in the base, mid‐gland, apex and peripheral zone of the prostate. Conclusions: We study the feasibility of using a probabilistic non‐rigid registration method for supplementing intra‐procedural images with diagnostic MR images. The method is evaluated geometrically and dosimetrically, which quantifies the method's ability to provide intra‐procedural estimation of the accuracy of the deformed configurations. This method will be applied to registration of diagnostic MR with intra‐procedural transrectal ultrasoundimages to improve visualization of the prostate apex and substructure, which could provide improved prostate boundary definitions with uncertainty margins to guide the development of intra‐operative brachytherapy treatment plans. This work was supported by NIH grants 1U41RR019703‐01A2, 1R01CA111288‐01A1,P01‐CA67165,andU01CA151261.

densities and the projection model is linear. Several linear projection and feature models are discussed. Evidence isprovided to show that Normal feature deviation models can be appropriate for computer vision matching problems.Relation to other work and possible extensions and application areas are discussed.

An efficient computation of the time-dependent forward solution for photon transport in a head model is a key capability for performing accurate inversion for functional diffuse optical imaging of the brain. The diffusion approximation to photon transport is much faster to simulate than the physically correct radiative transport equation (RTE); however, it is commonly assumed that scattering lengths must be much smaller than all system dimensions and all absorption lengths for the approximation to be accurate. Neither of these conditions is satisfied in the cerebrospinal fluid (CSF). Since line-of-sight distances in the CSF are small, of the order of a few millimeters, we explore the idea that the CSF scattering coefficient may be modeled by any value from zero up to the order of the typical inverse line-of-sight distance, or approximately 0.3 mm(-1), without significantly altering the calculated detector signals or the partial path lengths relevant for functional measurements. We demonstrate this in detail by using a Monte Carlo simulation of the RTE in a three-dimensional head model based on clinical magnetic resonance imaging data, with realistic optode geometries. Our findings lead us to expect that the diffusion approximation will be valid even in the presence of the CSF, with consequences for faster solution of the inverse problem.

It seems clear that small robots which take advantage of recent reductions in packaging size and costs of microelectronics can potentially be very useful; even more so if similar savings could be achieved in the actuation and power supply areas. Typically, the computational power required in a robotic system that connects perception to action is enormous, but if the organization of the sensors, actuators and computing elements is carefully laid out, the actual silicon area required for the intelligence system becomes quite small. A viable avenue of pursuit, then, is to aim towards scaling down the rest of the subsystems in a robot to the same scale as the control system, integrating motors, sensors, computation and power supplies onto a single piece of silicon; the advantages being mass productibility, lower costs and the avoidance of the usual connector problems encountered in combining discrete subsystems. By rethinking implementation strategies with this new form of robotic technology (i.e., the application of many very small robots), it may be possible to solve many problems more cost effectively, albeit in novel ways. As the completely integrated robot faces many technology hurdles, it seems necessary to focus on just one or two of the problem areas at a time. It turns out that many of the cost-saving benefits still accrue at small, but macroscopic scales. This paper describes an exercise of building a complete system, aimed at being as small as possible, but using off the shelf components exclusively. The result is an autonomous mobile robot slightly larger than one cubic inch, which incorporates sensing, actuation, onboard computation and on-board power supplies. Nicknamed Squirt, this robot acts as a ‘bug’, hiding in dark corners and venturing out in the direction of last heard noises, only moving after the noises are long gone.

This paper describes an exercise in building a complete robot aimed at being as small as possible but using off-the-shelf components exclusively. The result is an autonomous mobile robot slightly larger than one cubic inch which incorporates sensing, actuation, onboard computation, and onboard power supplies. Nicknamed Squirt, this robot acts as a ``bug,'''' hiding in dark corners and venturing out in the direction of last heard noises, only moving after the noises are long gone.

The analysis of the connectome of the human brain provides key insight into the brain's organisation and function, and its evolution in disease or ageing. Parcellation of the cortical surface into distinct regions in terms of structural connectivity is an essential step that can enable such analysis. The estimation of a stable connectome across a population of healthy subjects requires the estimation of a groupwise parcellation that can capture the variability of the connectome across the population. This problem has solely been addressed in the literature via averaging of connectivity profiles or finding correspondences between individual parcellations a posteriori. In this paper, we propose a groupwise parcellation method of the cortex based on diffusion MR images (dMRI). We borrow ideas from the area of cosegmentation in computer vision and directly estimate a consistent parcellation across different subjects and scales through a spectral clustering approach. The parcellation is driven by the tractography connectivity profiles, and information between subjects and across scales. Promising qualitative and quantitative results on a sizeable data-set demonstrate the strong potential of the method.

Segmentation of interstitial catheters from MRI needs to be addressed in order for MRI-based brachytherapy treatment planning to become part of the clinical practice of gynecologic cancer radiotherapy. This paper presents a validation study of a novel image-processing method for catheter segmentation. The method extends the distal catheter tip, interactively provided by the physician, to its proximal end, using knowledge of catheter geometry and appearance in MRI sequences. The validation study consisted of comparison of the algorithm results to expert manual segmentations, first on images of a phantom, and then on patient MRI images obtained during MRI-guided insertion of brachytherapy catheters for the treatment of gynecologic cancer. In the phantom experiment, the maximum disagreement between automatic and manual segmentation of the same MRI image, as computed using the Hausdorf distance, was 1.5 mm, which is of the same order as the MR image spatial resolution, while the disagreement between automatic segmentation of MR images and "ground truth", manual segmentation of CT images, was 3.5 mm. The segmentation method was applied to an IRB-approved retrospective database of 10 interstitial brachytherapy patients which included a total of 101 catheters. Compared with manual expert segmentations, the automatic method correctly segmented 93 out of 101 catheters, at an average rate of 0.3 seconds per catheter using a 3 GHz Intel Core i7 computer with 16 GB RAM and running Mac OS X 10.7. These results suggest that the proposed catheter segmentation is both technically and clinically feasible.

International audience; Functional Magnetic Resonance Imaging (fMRI) studies map the human brain by testing the response of groups of individuals to carefully-crafted and contrasted tasks in order to delineate specialized brain regions and networks. The number of functional networks extracted is limited by the number of subject-level contrasts and does not grow with the cohort. Here, we introduce a new group-level brain mapping strategy to differentiate many regions reflecting the variety of brain network configurations observed in the population. Based on the principle of functional segregation, our approach singles out functionally-specialized brain regions by learning group-level functional profiles on which the response of brain regions can be represented sparsely. We use a dictionary-learning formulation that can be solved efficiently with on-line algorithms, scaling to arbitrary large datasets. Importantly, we model inter-subject correspondence as structure imposed in the estimated functional profiles, integrating a structure-inducing regularization with no additional computational cost. On a large multi-subject study, our approach extracts a large number of brain networks with meaningful functional profiles.

Corpus Callosum Agenesis (CCA), one of the most common congenital anomalies, has uncertain neurodevelopmental outcome, especially when the disease is isolated. To provide parents with informed counselling, it is crucial to identify anatomical markers linked to a predicted outcome early in pregnancy. Quantitative exploration of fetal brains with CCA is rare and has been mostly limited to the study of specific brain structures. Here, we propose a pipeline to analyse fetal brain Magnetic Resonance Imaging (MRI) that is based on diffeomorphic transformation. It consists in two steps: a semi-automatic fetal MRI preprocessing procedure and a pipeline to quantify anatomical deviations from normal development. Following MRI preprocessing, each volumetric fetal brain is compared to an age-matched healthy template brain at a global scale using registration. Deformations are parallel transported to the same space to correct age differences between fetuses. Deformation modes specific to CCA are identified using Principal Component Analysis and classification. The pipeline is tested on retrospectively selected MRIs from 38 healthy fetuses and 73 fetuses with CCA. In accordance with more local analyses, the most relevant deformation mode for classification combines well-known alterations of brains with CCA. This preliminary work is promising for the quantitative exploration of abnormal fetal brains and will be used in the future to identify anatomical features correlated to poor clinical outcome.

International audience; This paper presents a new method for classifying surface datavia spectral representations of shapes. Our approach benefits classificationproblems that involve data living on surfaces, such as in cortical parcellation.For instance, current methods for labeling cortical points into surface parcelsoften involve a slow mesh deformation toward pre-labeled atlases, requiringas much as 4 hours with the established FreeSurfer. This may burden neurosciencestudies involving region-specific measurements. Learning techniquesoffer an attractive computational advantage, however, their representation ofspatial information, typically defined in a Euclidean domain, may be inadequatefor cortical parcellation. Indeed, cortical data resides on surfaces thatare highly variable in space and shape. Consequently, Euclidean representationsof surface data may be inconsistent across individuals. We proposeto fundamentally change the spatial representation of surface data, by exploitingspectral coordinates derived from the Laplacian eigenfunctions ofshapes. They have the advantage over Euclidean coordinates, to be geometryaware and to parameterize surfaces explicitly. This change of paradigm,from Euclidean to spectral representations, enables a classifier to be applieddirectly on surface data via spectral coordinates. In this paper, we decide tobuild upon the successful Random Decision Forests algorithm and improve itsspatial representation with spectral features. Our method, Spectral Forests,is shown to significantly improve the accuracy of cortical parcellations overstandard Random Decision Forests (74% versus 28% Dice overlaps), and produceaccuracy equivalent to FreeSurfer in a fraction of its time (23 secondsversus 3 to 4 hours).

ISBN 978-3-319-24552-2; International audience; Functional Arterial Spin Labeling (fASL) MRI can provide a quantitative measurement of changes of cerebral blood flow induced by stimulation or task performance. fASL data is commonly analysed using a general linear model (GLM) with regressors based on the canonical hemodynamic response function. In this work, we consider instead a joint detection-estimation (JDE) framework which has the advantage of allowing the extraction of both task-related perfusion and hemodynamic responses not restricted to canonical shapes. Previous JDE attempts for ASL have been based on computer intensive sampling (MCMC) methods. Our contribution is to provide a comparison with an alternative variational expectation-maximization (VEM) algorithm on synthetic and real data.

International audience; Recent work on brain tumor growth modeling for glioblas-toma using reaction-diffusion equations suggests that the diffusion coefficient and the proliferation rate can be related to clinically relevant information. However, estimating these parameters is difficult due to the lack of identifiability of the parameters, the uncertainty in the tumor segmen-tations, and the model approximation, which cannot perfectly capture the dynamics of the tumor. Therefore, we propose a method for conducting the Bayesian personalization of the tumor growth model parameters. Our approach estimates the posterior probability of the parameters, and allows the analysis of the parameters correlations and uncertainty. Moreover , this method provides a way to compute the evidence of a model, which is a mathematically sound way of assessing the validity of different model hypotheses. Our approach is based on a highly parallelized implementation of the reaction-diffusion equation, and the Gaussian Process Hamiltonian Monte Carlo (GPHMC), a high acceptance rate Monte Carlo technique. We demonstrate our method on synthetic data, and four glioblastoma patients. This promising approach shows that the infiltration is better captured by the model compared to the speed of growth.

International audience; In the context of forest-based segmentation of medical data, modeling the visual appearance around a voxel requires the choice of the scale at which contextual information is extracted, which is of crucial im- portance for the final segmentation performance. Building on Haar-like visual features, we introduce a simple yet effective modification of the for- est training which automatically infers the most informative scale at each stage of the procedure. Instead of the standard uniform sampling during node split optimization, our approach draws candidate features sequen- tially in a fine-to-coarse fashion. While being very easy to implement, this alternative is free of additional parameters, has the same computa- tional cost as a standard training and shows consistent improvements on three medical segmentation datasets with very different properties.

International audience; Cardiac cine MRI with 3D isotropic resolution is challenging as it requires efficient data acquisition and motion management. It is proposed to use a 2D balanced SSFP (steady-state free precession) sequence rather than its 3D version as it provides better contrast between blood and tissue. In order to obtain 3D isotropic images, 2D multi-slice datasets are acquired in different orientations (short axis, horizontal long axis and vertical long axis) while the patient is breathing freely. Image reconstruction is performed in two steps: (i) a motion-compensated reconstruction of each image stack corrects for nonrigid cardiac and respiratory motion; (ii) a super-resolution (SR) algorithm combines the three motion-corrected volumes (with low resolution in the slice direction) into a single volume with isotropic resolution. The SR reconstruction was implemented with two regularization schemes including a conventional one (Tikhonov) and a feature-preserving one (Beltrami). The method was validated in 8 volunteers and 10 patients with breathing difficulties. Image sharpness, as assessed by intensity profiles and by objective metrics based on the structure tensor, was improved with both SR techniques. The Beltrami constraint provided efficient denoising without altering the effective resolution.

International audience; Medical image segmentation is often a prerequisite for clinical applications. As an ill-posed problem, it leads to uncertain estimations of the region of interest which may have a significant impact on downstream applications, such as therapy planning. To quantify the uncertainty related to image segmentations, a classical approach is to measure the effect of using various plausible segmentations. In this paper, a method for producing such image segmentation samples from a single expert segmentation is introduced. A probability distribution of image segmentation boundaries is defined as a Gaussian process, which leads to segmentations that are spatially coherent and consistent with the presence of salient borders in the image. The proposed approach outperforms previous generative segmentation approaches, and segmentation samples can be generated efficiently. The sample variability is governed by a parameter which is correlated with a simple DICE score. We show how this approach can have multiple useful applications in the field of uncertainty quantification, and an illustration is provided in radiotherapy planning.