Realistic Brain Phantom Research Articles

Non-rigid image registration of brain magnetic resonance images using graph-cuts

We present a graph-cuts based method for non-rigid medical image registration on brain magnetic resonance images. In this paper, the non-rigid medical image registration problem is reformulated as a discrete labeling problem. Based on a voxel-to-voxel intensity similarity measure, each voxel in the source image is assigned a displacement label, which represents a displacement vector indicating which position in the floating image it is spatially corresponding to. In the proposed method, a smoothness constraint based on the first derivative is used to penalize sharp changes in the adjacent displacement labels across voxels. The image registration problem is therefore modeled by two energy terms based on intensity similarity and smoothness of the displacement field. These energy terms are submodular and can be optimized by using the graph-cuts method via α‐expansions, which is a powerful combinatorial optimization tool and capable of yielding either a global minimum or a local minimum in a strong sense. Using the realistic brain phantoms obtained from the Simulated Brain Database, we compare the registration results of the proposed method with two state-of-the-art medical image registration approaches: free-form deformation based method and demons method. In addition, the registration results are also compared with that of the linear programming based image registration method. It is found that the proposed method is more robust against different challenging non-rigid registration cases with consistently higher registration accuracy than those three methods, and gives realistic recovered deformation fields.

Development of a positron probe for localization and excision of brain tumours during surgery

The survival outcome of patients suffering from gliomas is directly linked to the complete surgical resection of the tumour. To help the surgeons to delineate precisely the boundaries of the tumour, we developed an intraoperative positron probe with background noise rejection capability. The probe was designed to be directly coupled to the excision tool such that detection and removal of the radiolabelled tumours could be simultaneous. The device consists of two exchangeable detection heads composed of clear and plastic scintillating fibres. Each head is coupled to an optic fibre bundle that exports the scintillating light to a photodetection and processing electronic module placed outside the operative wound. The background rejection method is based on a real-time subtraction technique. The measured probe sensitivity for 18F was 1.1 cps kBq−1 ml−1 for the small head and 3.4 cps kBq−1 ml−1 for the large head. The mean spatial resolution was 1.6 mm FWHM on the detector surface. The γ-ray rejection efficiency measured by realistic brain phantom modelling of the surgical cavity was 99.4%. This phantom also demonstrated the ability of the probe to detect tumour discs as small as 5 mm in diameter (20 mg) for tumour-to-background ratios higher than 3:1 and with an acquisition time around 4 s at each scanning step. These results indicate that our detector could be a useful complement to existing techniques for the accurate excision of brain tumour tissue and more generally to improve the efficiency of radio-guided cancer surgery.

Physical Performance of an Intraoperative Beta Probe Dedicated to Glioma Radioguided Surgery

The precise delineation and excision of brain tumor extent allows to improve survival outcome and quality of life of surgically treated patients. In order to refine the resection of gliomas, we are developing a novel intraoperative probe specifically dedicated to the localization of residual tumor after the bulk has been excised. The probe, built around clear and plastic scintillating fibers, was designed to detect positrons emitted from radiolabeled brain tissue in order to discriminate more specifically neoplastic from normal tissues. The probe was also built to be directly coupled to the excision tool leading to simultaneous detection and removal of tumor. We report here performances of the first radio-isotopic configuration of the intraoperative probe which consists of a detection head composed of eight detection elements held around the excision tool in a closed packed annular arrangement. This head is coupled to an optic fiber bundle that exports the scintillating light to a multi-channel photomultiplier tube. The gamma ray background generated by the annihilation of beta+ in tissues is eliminated by a real-time subtraction method. The detector exhibits a beta sensitivity of 139 cps/kBq and a gamma ray rejection efficiency of 99.5%. The ability of the probe to detect residual lesions was evaluated with a realistic brain phantom representing the surgical cavity and the boundaries of the tumor. We showed that lesions as small as 5 mm in diameter can be detected for tumor to normal tissue uptake ratios of fluorinated tracers greater than 3.5. This ratio is achieved with radiopharmaceuticals like 18F-FET or 18F-choline. These promising results suggest that the features of our system are compatible with in situ localization of residual radiolabeled tumors.

Twenty New Digital Brain Phantoms for Creation of Validation Image Data Bases

Simulations provide a way of generating data where ground truth is known, enabling quantitative testing of image processing methods. In this paper, we present the construction of 20 realistic digital brain phantoms that can be used to simulate medical imaging data. The phantoms are made from 20 normal adults to take into account intersubject anatomical variabilities. Each digital brain phantom was created by registering and averaging four T1, T2, and proton density (PD)-weighted magnetic resonance imaging (MRI) scans from each subject. A fuzzy minimum distance classification was used to classify voxel intensities from T1, T2, and PD average volumes into grey-matter, white matter, cerebro-spinal fluid, and fat. Automatically generated mask volumes were required to separate brain from nonbrain structures and ten fuzzy tissue volumes were created: grey matter, white matter, cerebro-spinal fluid, skull, marrow within the bone, dura, fat, tissue around the fat, muscles, and skin/muscles. A fuzzy vessel class was also obtained from the segmentation of the magnetic resonance angiography scan of the subject. These eleven fuzzy volumes that describe the spatial distribution of anatomical tissues define the digital phantom, where voxel intensity is proportional to the fraction of tissue within the voxel. These fuzzy volumes can be used to drive simulators for different modalities including MRI, PET, or SPECT. These phantoms were used to construct 20 simulated T1-weighted MR scans. To evaluate the realism of these simulations, we propose two approaches to compare them to real data acquired with the same acquisition parameters. The first approach consists of comparing the intensities within the segmented classes in both real and simulated data. In the second approach, a whole brain voxel-wise comparison between simulations and real T1-weighted data is performed. The first comparison underlines that segmented classes appear to properly represent the anatomy on average, and that inside these classes, the simulated and real intensity values are quite similar. The second comparison enables the study of the regional variations with no a priori class. The experiments demonstrate that these variations are small when real data are corrected for intensity nonuniformity.

Quantitative simultaneous /sup 99m/Tc//sup 123/I SPECT: design study and validation with Monte Carlo simulations and physical acquisitions

Simultaneous dual isotope imaging (/sup 99m/Tc//sup 123/I) has potential clinical applications but has not been implemented in the clinic. The aim of this work was to design an artificial neural network (ANN) for crosstalk and scatter correction using a smaller number of energy windows (8) than we had previously proposed (26) to allow implementation on some clinical cameras, and to validate our approach using realistic Monte Carlo simulations and anthropomorphic brain phantom acquisitions. Monte Carlo simulations of dual isotope SPECT studies of a digital brain phantom and physical acquisitions of the striatal brain phantom were used to validate our approach. Corrected projections were reconstructed using an iterative ordered subsets expectation maximization (OSEM) algorithm that modeled nonuniform attenuation and variable collimator response in the projector/backprojector. Results: In Monte Carlo simulations, ANN26 and ANN8 yielded similarly accurate quantitation of /sup 123/I activity (bias <7%) in all brain structures. An asymmetric windowing method (AW) yielded accurate estimation in the striata (bias <7%) but not in other brain structures. The estimation bias of /sup 99m/Tc primary activity was <10% in all brain structures with ANN26 and ANN8.

A cooperative framework for segmentation of MRI brain scans

Automatic segmentation of MRI brain scans is a complex task for two main reasons: the large variability of the human brain anatomy, which limits the use of general knowledge and, inherent to MRI acquisition, the artifacts present in the images that are difficult to process. To tackle these difficulties, we propose to mix, in a cooperative framework, several types of information and knowledge provided and used by complementary individual systems: presently, a multi-agent system, a deformable model and an edge detector. The outcome is a cooperative segmentation performed by a set of region and edge agents constrained automatically and dynamically by both, the specific gray levels in the considered image, statistical models of the brain structures and general knowledge about MRI brain scans. Interactions between the individual systems follow three modes of cooperation: integrative, augmentative and confrontational cooperation, combined during the three steps of the segmentation process namely, the specialization of the seeded-region-growing agents, the fusion of heterogeneous information and the retroaction over slices. The described cooperative framework allows the dynamic adaptation of the segmentation process to the own characteristics of each MRI brain scan. Its evaluation using realistic brain phantoms is reported.

Registration of dynamic dopamine D2 receptor images using principal component analysis.

This paper describes a novel technique for registering a dynamic sequence of single-photon emission tomography (SPET) dopamine D2 receptor images, using principal component analysis (PCA). Conventional methods for registering images, such as count difference and correlation coefficient algorithms, fail to take into account the dynamic nature of the data, resulting in large systematic errors when registering time-varying images. However, by using principal component analysis to extract the temporal structure of the image sequence, misregistration can be quantified by examining the distribution of eigenvalues. The registration procedures were tested using a computer-generated dynamic phantom derived from a high-resolution magnetic resonance image of a realistic brain phantom. Each method was also applied to clinical SPET images of dopamine D2 receptors, using the ligands iodine-123 iodobenzamide and iodine-123 epidepride, to investigate the influence of misregistration on kinetic modelling parameters and the binding potential. The PCA technique gave highly significant (P<0.001) improvements in image registration, leading to alignment errors in x and y of about 25% of the alternative methods, with reductions in autocorrelations over time. It could also be applied to align image sequences which the other methods failed completely to register, particularly 123I-epidepride scans. The PCA method produced data of much greater quality for subsequent kinetic modelling, with an improvement of nearly 50% in the chi2 of the fit to the compartmental model, and provided superior quality registration of particularly difficult dynamic sequences.

Region of interest evaluation of SPECT image reconstruction methods using a realistic brain phantom

A realistic numerical brain phantom, developed by Zubal et al. (1994), was used for a region-of-interest evaluation of the accuracy and noise variance of the following SPECT reconstruction methods: (1) maximum-likelihood reconstruction using the expectation-maximization (ML-EM) algorithm; (2) an EM algorithm using ordered-subsets (OS-EM); (3) a re-scaled block iterative EM algorithm (RBI-EM); and (4) a filtered backprojection algorithm that uses a combination of the Bellini method for attenuation compensation and an iterative spatial blurring correction method using the frequency-distance principle (FDP). The Zubal phantom was made from segmented MRI slices of the brain, so that neuro-anatomical structures are well defined and indexed. Small regions-of-interest (ROIs) from the white matter, grey matter in the center of the brain and grey matter from the peripheral area of the brain were selected for the evaluation. Photon attenuation and distance-dependent collimator blurring were modeled. Multiple independent noise realizations were generated for two different count levels. The simulation study showed that the ROI bias measured for the EM-based algorithms decreased as the iteration number increased, and that the OS-EM and RBI-EM algorithms (16 and 64 subsets were used) achieved the equivalent accuracy of the ML-EM algorithm at about the same noise variance, with much fewer number of iterations. The Bellini-FDP restoration algorithm converged fast and required less computation per iteration. The ML-EM algorithm had a slightly better ROI bias vs. variance trade-off than the other algorithms.

A Bayesian approach to PET reconstruction using image-modeling Gibbs priors: implementation and comparison

We demonstrate that (i) classical methods of image reconstruction from projections can be improved upon by considering the output of such a method as a distorted version of the original image and applying a Bayesian approach to estimate from it the original image (based on a model of distortion and on a Gibbs distribution as the prior) and (ii) by selecting an image-modeling prior distribution (i.e., one which is such that it is likely that a random sample from it shares important characteristics of the images of the application area) one can improve over another Gibbs prior formulated using only pairwise interactions. We illustrate our approach using simulated positron emission tomography (PET) data from realistic brain phantoms. Since algorithm performance ultimately depends on the diagnostic task being performed. We examine a number of different medically relevant figures of merit to give a fair comparison. Based on a training-and-testing evaluation strategy, we demonstrate that statistically significant improvements can be obtained using the proposed approach. We also present a statistical verification of the normality condition required for the above statistical claim.

Automatic 3‐D model‐based neuroanatomical segmentation

AbstractExplicit segmentation is required for many forms of quantitative neuroanatomic analysis. However, manual methods are time‐consuming and subject to errors in both accuracy and reproducibility (precision). A 3‐D model‐based segmentation method is presented in this paper for the completely automatic identification and delineation of gross anatomical structures of the human brain based on their appearance in magnetic resonance images (MRI).The approach depends on a general, iterative, hierarchical non‐linear registration procedure and a 3‐D digital model of human brain anatomy that contains both volumetric intensity‐based data and a geometric atlas. Here, the traditional segmentation strategy is inverted: instead of matching geometric contours from and idealized atlas directly to the MRI data, segmentation is achieved by identifying the non‐linear spatial transformation that best maps corresponding intensity‐based features between a model image and a new MRI brain volume. When completed, atlas contours defined on the model image are mapped through the same transformation to segment and label individual structures in the new data set.Using manually segmented sturcture boundaries for comparison, measures of volumetric difference and volumetric overlap were less than 2% and better than 97% for realistic brain phantom data, and less than 10% and better than 85%, respectively, for human MRI data. This compares favorably to intra‐observer variability estimates of 4.9% and 87%, respectively. The procedure performs well, is objective and its implementation robust. The procedure requires no manual intervention, and is thus applicable to studies of large numbers of subjects. The general method for non‐linear image matching is also useful for non‐linear mapping of brain data sets into stereotaxic space if the target volume is already in stereotaxic space. © 1995 Wiley‐Liss, Inc.

3-D computer simulations of resolution effects of multidetector point focusing SPECT imaging

The authors present an efficient algorithm and the results of its application in simulating the three-dimensional (3-D) projection data resulting from a 3-D distribution of radioactivity. The algorithm was applied to a series of geometrical mathematical phantoms and to a realistic mathematical brain phantom. The authors simulated the projection data from a multidetector single-photon emission computed tomography (SPECT) system with point focusing collimators. The simulated projection data were then reconstructed using the manufacturer's software. The objects simulated included simple geometrical solids such as spheres and sheets, as well as the distribution of muscarinic cholinergic receptors in a realistic brain slice. Spheres were chosen as a model for brain structures such as caudate nucleus, thalamus, and cerebellum; sheets were selected as representing lateral cortical gray matter regions. The results of these simulations indicate the existence of significant qualitative and quantitative artifacts in reconstructed human brain images.

Artifacts in camera based single photon emission tomography due to time activity variation.

Image quality in single photon emission computed tomography (SPECT) using a rotating gamma camera is dependent on the time course of the tracer in the field of view. If acquisition times are slow compared to the tracer turnover, artifacts may occur in the reconstructed images. The properties of such artifacts were studied by computer simulation. Experimental projection data of point sources, cylindrical phantoms, and an anatomically realistic brain phantom were altered by sequentially weighting the projections with a function that varied exponentially or linearly with time. The observed distortion in the reconstructed images could be related to the ratio between the object activity variation and the camera rotation time. If the tracer concentration changed less than a factor of two during one camera rotation then little image distortion was visible although quantitatively the resolution was degraded. If the object's activity variation with time is fast enough to produce noticeable distortion, the artifacts can be reduced by performing multiple, rapid, camera rotations, instead of one rotation, for the same total acquisition time. The proposed procedure is computer storage space intensive and takes longer to produce the transaxial images, but improves image quality.

A Realistic Computer-Simulated Brain Phantom for Evaluation of PET Charactenstics

To evaluate accurately the imaging characteristics of positron emission tomography (PET), a realistic computer-simulated brain phantom was developed. A cross-sectional slice from a human cadaver brain was chosen for its combination of gray matter, white matter, and cerebrospinal fluid (CSF) regions. The slice was photographed and digitized into a gray-level image with a video digitizer, boundary edges were located around cerebral structures in the digitized image, and each structural region was assigned a uniform pixel value dependent on both the cerebral parameter (e.g., blood flow, oxygen uptake, metabolic rate) under investigation and the type of structure (gray matter, white matter, CSF). Line integrals through the regions were generated at various angular and transverse positions according to specific physical characteristics (such as detector line-spread function) of a tomographic scanner configuration to create a set of simulated but realistic projection measurements. The set of projection measurements can be processed with any standard reconstruction program to create a tomographic image to reveal the effects of various PET characteristics. Investigations with this computer-simulated brain phantom have demonstrated its usefulness for examining the interrelations among neuroanatomical structure volume, tomographic spatial resolution, partial volume effect, and nonlinear parameter estimation. Transportability of the simulated phantom and the procedure to other medical imaging environments is described, and limitations of this simulation procedure are discussed.