MRI simulation-based evaluation of image-processing and classification methods

Bias artifact suppression on MR volumes

RF-inhomogeneity correction is a relevant research topic in the field of magnetic resonance imaging (MRI). A volume corrupted by this artifact exhibits nonuniform illumination both inside a single slice and between adjacent ones. In this work a bias correction technique is presented, which suppresses this artifact on MR volumes scanned from different body parts without any a priori hypothesis on the artifact model. Theoretical foundations of the method are reported together with experimental results and a comparison is presented with both the 2D version of the algorithm and other techniques that are widely used in MRI literature.

Estimating the joint statistics of images using nonparametric windows with application to registration using mutual information

Recently, the Nonparametric (NP) Windows has been proposed to estimate the statistics of real 1D and 2D signals. NP Windows is accurate, because it is equivalent to sampling images at a high (infinite) resolution for an assumed interpolation model. This paper extends the proposed approach to consider joint distributions of image-pairs. Second, Green's Theorem is used to simplify the previous NP Windows algorithm. Finally, a resolution-aware NP Windows algorithm is proposed to improve robustness to relative scaling between an image pair. Comparative testing of 2D image registration was performed using translation-only and affine transformations. Although more expensive than other methods, NP Windows frequently demonstrated superior performance for bias (distance between ground truth and global maximum) and frequency of convergence. Unlike other methods, the number of samples and the number of bins have little effect on NP Windows and the prior selection of a kernel is not required.

Brain MR images segmentation using statistical ratio: mapping between watershed and competitive Hopfield clustering network algorithms

Conventional watershed segmentation methods invariably produce over-segmented images due to the presence of noise or local irregularities in the source images. In this paper, a robust medical image segmentation technique is proposed, which combines watershed segmentation and the competitive Hopfield clustering network (CHCN) algorithm to minimize undesirable over-segmentation. In the proposed method, a region merging method is presented, which is based on employing the region adjacency graph (RAG) to improve the quality of watershed segmentation. The relation of inter-region similarities is then investigated using image mapping in the watershed and CHCN images to determine more appropriate region merging. The performance of the proposed technique is presented through quantitative and qualitative validation experiments on benchmark images. Significant and promising segmentation results were achieved on brain phantom simulated data.

Robust MRI brain tissue parameter estimation by multistage outlier rejection

This article addresses the problem of the tissue type parameter estimation in brain MRI in the presence of partial volume effects. Automatic MRI brain tissue classification is hampered by partial volume effects that are caused by the finite resolution of the acquisition process. Due to this effect intensity distributions in brain MRI cannot be well modeled by a simple mixture of Gaussians and therefore more complex models have been developed. Unfortunately, these models do not seem to be robust enough for clinical conditions, as the quality of the tissue classification decreases rapidly with the image quality. Also, the application of these methods for pathological images with unmodeled intensities (e.g. MS plaques, tumors, etc.) remains uncertain. In the present work a new robust method for brain tissue characterization is presented, treating the partial volume affected voxels as outliers of the pure tissue distributions. The proposed method estimates the tissue characteristics from a reduced set of intensities belonging to a particular pure tissue class. This reduced set is selected by using a trimming procedure based on local gradient information and distributional data. This feature makes the method highly tolerant of a large amount of unexpected intensities without degrading its performance. The proposed method has been evaluated using both synthetic and real MR data and compared with state-of-the-art methods showing the best results in the comparative.

A modified fuzzy C-means classification method using a multiscale diffusion filtering scheme

A fully automatic, multiscale fuzzy C-means (MsFCM) classification method for MR images is presented in this paper. We use a diffusion filter to process MR images and to construct a multiscale image series. A multiscale fuzzy C-means classification method is applied along the scales from the coarse to fine levels. The objective function of the conventional fuzzy C-means (FCM) method is modified to allow multiscale classification processing where the result from a coarse scale supervises the classification in the next fine scale. The method is robust for noise and low-contrast MR images because of its multiscale diffusion filtering scheme. The new method was compared with the conventional FCM method and a modified FCM (MFCM) method. Validation studies were performed on synthesized images with various contrasts and on the McGill brain MR image database. Our MsFCM method consistently performed better than the conventional FCM and MFCM methods. The MsFCM method achieved an overlap ratio of greater than 90% as validated by the ground truth. Experiments results on real MR images were given to demonstrate the effectiveness of the proposed method. Our multiscale fuzzy C-means classification method is accurate and robust for various MR images. It can provide a quantitative tool for neuroimaging and other applications.

Reconstruction of central cortical surface from brain MRI images: method and application

Reconstruction of the central surface representation of the cerebral cortex is an important means to study the structure and function of the human brain. In this paper, we propose a novel method based on an elastic transform vector field to drive a deformable model for the reconstruction of the central cortical surface. Both simulated brain cortexes and real brain images are used to evaluate this approach. We applied the surface reconstruction method and a hybrid volumetric and surface registration algorithm to detect simulated brain atrophy. Experimental results show that the central cortical surface representation has better performance in detecting simulated atrophy than the traditionally used inner or outer cortical surface representations.

MRI-based automated computer classification of probable AD versus normal controls

Automated computer classification (ACC) techniques are needed to facilitate physician's diagnosis of complex diseases in individual patients. We provide an example of ACC using computational techniques within the context of cross-sectional analysis of magnetic resonance images (MRI) in neurodegenerative diseases, namely Alzheimer's dementia (AD). In this paper, the accuracy of our ACC methodology is assessed when presented with real life, imperfect data, i.e., cohorts of MRI with varying acquisition parameters and imaging quality. The comparative methodology uses the Jacobian determinants derived from dense deformation fields and scaled grey-level intensity from a selected volume of interest centered on the medial temporal lobe. The ACC performance is assessed in a series of leave-one-out experiments aimed at separating 75 probable AD and 75 age-matched normal controls. The resulting accuracy is 92% using a support vector machine classifier based on least squares optimization. Finally, it is shown in the Appendix that determinants and scaled grey-level intensity are appreciably more robust to varying parameters in validation studies using simulated data, when compared to raw intensities or grey/white matter volumes. The ability of cross-sectional MRI at detecting probable AD with high accuracy could have profound implications in the management of suspected AD candidates.

Validation of non-rigid registration between functional and anatomical magnetic resonance brain images

This paper presents a set of validation procedures for nonrigid registration of functional EPI to anatomical MRI brain images. Although various registration techniques have been developed and validated for high-resolution anatomical MRI images, due to a lack of quantitative and qualitative validation procedures, the use of nonrigid registration between functional EPI and anatomical MRI images has not yet been deployed in neuroimaging studies. In this paper, the performance of a robust formulation of a nonrigid registration technique is evaluated in a quantitative manner based on simulated data and is further evaluated in a quantitative and qualitative manner based on in vivo data as compared to the commonly used rigid and affine registration techniques in the neuroimaging software packages. The nonrigid registration technique is formulated as a second-order constrained optimization problem using a free-form deformation model and mutual information similarity measure. Bound constraints, resolution level and cross-validation issues have been discussed to show the degree of accuracy and effectiveness of the nonrigid registration technique. The analyses performed reveal that the nonrigid approach provides a more accurate registration, in particular when the functional regions of interest lie in regions distorted by susceptibility artifacts.

MR brain tissue classification using an edge-preserving spatially variant Bayesian mixture model

In this paper, a spatially constrained mixture model for the segmentation of MR brain images is presented. The novelty of this work is an edge-preserving smoothness prior which is imposed on the probabilities of the voxel labels. This prior incorporates a line process, which is modeled as a Bernoulli random variable, in order to preserve edges between tissues. The main difference with other, state of the art methods imposing priors, is that the constraint is imposed on the probabilities of the voxel labels and not onto the labels themselves. Inference of the proposed Bayesian model is obtained using variational methodology and the model parameters are computed in closed form. Numerical experiments are presented where the proposed model is favorably compared to state of the art brain segmentation methods as well as to a spatially varying Gaussian mixture model.

Composite index for the quantitative evaluation of image segmentation results

Medical image segmentation is one of the most productive research areas in medical image processing. The goal of most new image segmentation algorithms is to achieve higher segmentation accuracy than existing algorithms. But the issue of quantitative, reproducible validation of segmentation results, and the questions: What is segmentation accuracy?, and: What segmentation accuracy can a segmentation algorithm achieve? remain wide open. The creation of a validation framework is relevant and necessary for consistent and realistic comparisons of existing, new and future segmentation algorithms. An important component of a reproducible and quantitative validation framework for segmentation algorithms is a composite index that will measure segmentation performance at a variety of levels. We present a prototype composite index that includes the measurement of seven metrics on segmented image sets. We explain how the composite index is a more complete and robust representation of algorithmic performance than currently used indices that rate segmentation results using a single metric. Our proposed index can be read as an averaged global metric or as a series of algorithmic ratings that will allow the user to compare how an algorithm performs under many categories.

Computer-generated fMRI phantoms with motion-distortion interaction

As an extension of previous work on computer-generated phantoms, more accurate, realistic phantoms are generated by integrating image distortion and signal loss caused by susceptibility variations. With the addition of real motions and activations determined from actual functional MRI studies, these phantoms can be used by the fMRI community to assess with higher fidelity pre-processing algorithms such as motion correction, distortion correction and signal-loss compensation. These phantoms were validated by comparison to real echo-planar images. Specifically, studies have shown the effects of motion-distortion interactions on fMRI. We performed motion correction and activation analysis on these phantoms based on a block paradigm design using SPM2, and the results demonstrate that interactions between motion and distortion affect both motion correction and activation detection and thus represent a critical component of phantom generation.

Genetic algorithms for finite mixture model based voxel classification in neuroimaging

Finite mixture models (FMMs) are an indispensable tool for unsupervised classification in brain imaging. Fitting an FMM to the data leads to a complex optimization problem. This optimization problem is difficult to solve by standard local optimization methods, such as the expectation-maximization (EM) algorithm, if a principled initialization is not available. In this paper, we propose a new global optimization algorithm for the FMM parameter estimation problem, which is based on real coded genetic algorithms. Our specific contributions are two-fold: 1) we propose to use blended crossover in order to reduce the premature convergence problem to its minimum and 2) we introduce a completely new permutation operator specifically meant for the FMM parameter estimation. In addition to improving the optimization results, the permutation operator allows for imposing biologically meaningful constraints to the FMM parameter values. We also introduce a hybrid of the genetic algorithm and the EM algorithm for efficient solution of multidimensional FMM fitting problems. We compare our algorithm to the self-annealing EM-algorithm and a standard real coded genetic algorithm with the voxel classification tasks within the brain imaging. The algorithms are tested on synthetic data as well as real three-dimensional image data from human magnetic resonance imaging, positron emission tomography, and mouse brain MRI. The tissue classification results by our method are shown to be consistently more reliable and accurate than with the competing parameter estimation methods.

Brain functional localization: a survey of image registration techniques

Functional localization is a concept which involves the application of a sequence of geometrical and statistical image processing operations in order to define the location of brain activity or to produce functional/parametric maps with respect to the brain structure or anatomy. Considering that functional brain images do not normally convey detailed structural information and, thus, do not present an anatomically specific localization of functional activity, various image registration techniques are introduced in the literature for the purpose of mapping functional activity into an anatomical image or a brain atlas. The problems addressed by these techniques differ depending on the application and the type of analysis, i.e., single-subject versus group analysis. Functional to anatomical brain image registration is the core part of functional localization in most applications and is accompanied by intersubject and subject-to-atlas registration for group analysis studies. Cortical surface registration and automatic brain labeling are some of the other tools towards establishing a fully automatic functional localization procedure. While several previous survey papers have reviewed and classified general-purpose medical image registration techniques, this paper provides an overview of brain functional localization along with a survey and classification of the image registration techniques related to this problem.

Intensity Non-uniformity Correction of Magnetic Resonance Images Using a Fuzzy Segmentation Algorithm

Artifacts in magnetic resonance images can make conventional intensity-based segmentation methods very difficult, especially for the spatial intensity non-uniformity induced by the radio frequency (RF) coil. The non-uniformity introduces a slow-varying shading artifact across the images. Many advanced techniques, such as nonparametric, multi-channel methods, cannot solve the problem. In this paper, the extension of an improved fuzzy segmentation method, based on the traditional fuzzy c-means (FCM) algorithm and neighborhood attraction, is proposed to correct the intensity non-uniformity. Experimental results on both synthetic non-MR and MR images are given demonstrate the superiority of the algorithm.

Change Detection & Characterization: A New Tool for Imaging Informatics and Cancer Research

Modern imaging systems are able to produce a rich and diverse array of information, regarding various facets of anatomy and function. The quantity of information produced by these systems is so bountiful, however, as to have the potential to become a hindrance to clinical assessment. In the context of serial image evaluation, computer-based change detection and characterization is one important mechanism to process the information produced by imaging systems, so as to reduce the quantity of data, direct the attention of the physician to regions of the data which are the most informative for their purposes, and present the data in the form in which it will be the most useful. Change detection and characterization algorithms may serve as a basis for the creation of an objective definition of progression, which will reduce inter and intra-observer variability, and facilitate earlier detection of disease and recurrence, which in turn may lead to improved outcomes. Decreased observer variability combined with increased acuity should make it easier to discover promising therapies. Quantitative measures of the response to these therapies should provide a means to compare the effectiveness of treatments under investigation. Change detection may be applicable to a broad range of cancers, in essentially all anatomical regions. The source of information upon which change detection comparisons may be based is likewise broad. Validation of algorithms for the longitudinal assessment of cancer patients is expected to be challenging, though not insurmountable, as the many facets of the problem mean that validation will likely need to be approached from a variety of vantage points. Change detection and characterization is quickly becoming a very active field of investigation, and it is expected that this burgeoning field will help to facilitate cancer care both in the clinic and research.

Strategies for attenuation compensation in neurological PET studies

Molecular brain imaging using positron emission tomography (PET) has evolved into a vigorous academic field and is progressively gaining importance in the clinical arena. Significant progress has been made in the design of high-resolution three-dimensional (3-D) PET units dedicated to brain research and the development of quantitative imaging protocols incorporating accurate image correction techniques and sophisticated image reconstruction algorithms. However, emerging clinical and research applications of molecular brain imaging demand even greater levels of accuracy and precision and therefore impose more constraints with respect to the quantitative capability of PET. It has long been recognized that photon attenuation in tissues is the most important physical factor degrading PET image quality and quantitative accuracy. Quantitative PET image reconstruction requires an accurate attenuation map of the object under study for the purpose of attenuation compensation. Several methods have been devised to correct for photon attenuation in neurological PET studies. Significant attention has been devoted to optimizing computational performance and to balancing conflicting requirements. Approximate methods suitable for clinical routine applications and more complicated approaches for research applications, where there is greater emphasis on accurate quantitative measurements, have been proposed. The number of scientific contributions related to this subject has been increasing steadily, which motivated the writing of this review as a snapshot of the dynamically changing field of attenuation correction in cerebral 3D PET. This paper presents the physical and methodological basis of photon attenuation and summarizes state of the art developments in algorithms used to derive the attenuation map aiming at accurate attenuation compensation of brain PET data. Future prospects, research trends and challenges are identified and directions for future research are discussed.

Sequential anisotropic Wiener filtering applied to 3D MRI data

We present three different sequential Wiener filters, namely, isotropic, orientation and anisotropic. The first one is similar to the classical Wiener filter in the sense that it uses an isotropic neighborhood to estimate its parameters. Here we present a sequential version of it. The orientation Wiener filter uses oriented neighborhoods to estimate the structure orientation present at each voxel, giving rise to a modified estimator of the parameters. Finally, the anisotropic Wiener filter combines both approaches adaptively so that the appropriate approach is locally selected. Several synthetic experiments are presented showing the performance of the filters with respect to their parameters. A mean square error analysis is performed using a publicly available magnetic resonance imaging (MRI) brain phantom and a comparison with other filtering approaches is carried out. In addition, results from filtering real MRI data are presented.

Development of a functional magnetic resonance imaging simulator for modeling realistic rigid-body motion artifacts

Functional magnetic resonance imaging (FMRI) is a noninvasive method of imaging brain function in vivo. However, images produced in FMRI experiments are imperfect and contain several artifacts that contaminate the data. These artifacts include rigid-body motion effects, B0-field inhomogeneities, chemical shift, and eddy currents. To investigate these artifacts, with the eventual aim of minimizing or removing them completely, a computational model of the FMR image acquisition process was built that can simulate all of the above-mentioned artifacts. This paper gives an overview of the development of the FMRI simulator. The simulator uses the Bloch equations together with a geometric definition of the object (brain) and a varying T2* model for the BOLD activations. Furthermore, it simulates rigid-body motion of the object by solving Bloch equations for given motion parameters that are defined for an object moving continuously in time, including during the read-out period, which is a novel approach in the area of MRI computer simulations. With this approach it is possible, in a controlled and precise way, to simulate the full effects of various rigid-body motion artifacts in FMRI data (e.g. spin-history effects, B0-motion interaction, and within-scan motion blurring) and therefore formulate and test algorithms for their reduction.

Novel multistage three-dimensional medical image segmentation: methodology and validation

In this paper, we propose a novel multistage method for three-dimensional (3-D) segmentation of medical images and a new radial distance-based segmentation validation approach. For the 3-D segmentation method, we first employ a morphological recursive erosion operation to reduce the connectivity between the region of interest and its surrounding neighborhood; then we design a hybrid segmentation method to achieve an initial result. The hybrid approach integrates an improved fast marching method and a morphological reconstruction algorithm. Finally, a morphological recursive dilation is employed to recover any lost structure from the first stage of the multistage method. This approach is tested on 12 CT and 3 MRI images of the brain, heart, and kidney, to demonstrate the effectiveness and accuracy of this technique across a variety of imaging modalities and organ systems. In order to validate the multistage segmentation method, a novel radial distance-based validation method is proposed that uses a global accuracy (GA) measure. The GA is calculated based on local radial distance errors (LRDE), where LRDE are calculated on the radii emitted from points along the skeleton of the object rather than the centroid, in order to accommodate more complicated organ structures. The experimental results demonstrate that the proposed multistage segmentation method is fast and accurate, with comparable performance to existing segmentation methods, but with a significantly higher execution speed.

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.

Generalized overlap measures for evaluation and validation in medical image analysis

Measures of overlap of labelled regions of images, such as the Dice and Tanimoto coefficients, have been extensively used to evaluate image registration and segmentation algorithms. Modern studies can include multiple labels defined on multiple images yet most evaluation schemes report one overlap per labelled region, simply averaged over multiple images. In this paper, common overlap measures are generalized to measure the total overlap of ensembles of labels defined on multiple test images and account for fractional labels using fuzzy set theory. This framework allows a single "figure-of-merit" to be reported which summarises the results of a complex experiment by image pair, by label or overall. A complementary measure of error, the overlap distance, is defined which captures the spatial extent of the nonoverlapping part and is related to the Hausdorff distance computed on grey level images. The generalized overlap measures are validated on synthetic images for which the overlap can be computed analytically and used as similarity measures in nonrigid registration of three-dimensional magnetic resonance imaging (MRI) brain images. Finally, a pragmatic segmentation ground truth is constructed by registering a magnetic resonance atlas brain to 20 individual scans, and used with the overlap measures to evaluate publicly available brain segmentation algorithms.

A new improved version of the realistic digital brain phantom

Image analysis methods must be tested and evaluated within a controlled environment. Simulations can be an extremely helpful tool for validation because ground truth is known. We created the digital brain phantom that is at the heart of our publicly available database of realistic simulated magnetic resonance image (MRI) volumes known as BrainWeb. Even though the digital phantom had l mm(3) isotropic voxel size and a small number of tissue classes, the BrainWeb database has been used in more than one hundred peer-reviewed publications validating different image processing methods. In this paper, we describe the next step in the natural evolution of BrainWeb: the creation of digital brain phantom II that includes three major improvements over the original phantom. First, the realism of the phantom, and the resulting simulations, was improved by modeling more tissue classes to include blood vessels, bone marrow and dura mater classes. In addition. a more realistic skull class was created. The latter is particularly useful for SPECT, PET and CT simulations for which bone attenuation has an important effect. Second, the phantom was improved by an eight-fold reduction in voxel volume to 0.125 mm(3). Third, the method used to create the new phantom was modified not only to take into account the segmentation of these new structures, but also to take advantage of many more automated procedures now available. The overall process has reduced subjectivity and manual intervention when compared to the original phantom, and the process may be easily applied to create phantoms from other subjects. MRI simulations are shown to illustrate the difference between the previous and the new improved digital brain phantom II. Example PET and SPECT simulations are also presented.

Comparative assessment of statistical brain MR image segmentation algorithms and their impact on partial volume correction in PET

Magnetic resonance imaging (MRI)-guided partial volume effect correction (PVC) in brain positron emission tomography (PET) is now a well-established approach to compensate the large bias in the estimate of regional radioactivity concentration, especially for small structures. The accuracy of the algorithms developed so far is, however, largely dependent on the performance of segmentation methods partitioning MRI brain data into its main classes, namely gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). A comparative evaluation of three brain MRI segmentation algorithms using simulated and clinical brain MR data was performed, and subsequently their impact on PVC in 18F-FDG and 18F-DOPA brain PET imaging was assessed. Two algorithms, the first is bundled in the Statistical Parametric Mapping (SPM2) package while the other is the Expectation Maximization Segmentation (EMS) algorithm, incorporate a priori probability images derived from MR images of a large number of subjects. The third, here referred to as the HBSA algorithm, is a histogram-based segmentation algorithm incorporating an Expectation Maximization approach to model a four-Gaussian mixture for both global and local histograms. Simulated under different combinations of noise and intensity non-uniformity, MR brain phantoms with known true volumes for the different brain classes were generated. The algorithms' performance was checked by calculating the kappa index assessing similarities with the "ground truth" as well as multiclass type I and type II errors including misclassification rates. The impact of image segmentation algorithms on PVC was then quantified using clinical data. The segmented tissues of patients' brain MRI were given as input to the region of interest (RoI)-based geometric transfer matrix (GTM) PVC algorithm, and quantitative comparisons were made. The results of digital MRI phantom studies suggest that the use of HBSA produces the best performance for WM classification. For GM classification, it is suggested to use the EMS. Segmentation performed on clinical MRI data show quite substantial differences, especially when lesions are present. For the particular case of PVC, SPM2 and EMS algorithms show very similar results and may be used interchangeably. The use of HBSA is not recommended for PVC. The partial volume corrected activities in some regions of the brain show quite large relative differences when performing paired analysis on 2 algorithms, implying a careful choice of the segmentation algorithm for GTM-based PVC.

Robust analysis of short echo time (1)H MRSI of human brain

Short echo time proton MR Spectroscopic Imaging (MRSI) suffers from low signal-to-noise ratio (SNR), limiting accuracy to estimate metabolite intensities. A method to coherently sum spectra in a region of interest of the human brain by appropriate peak alignment was developed to yield a mean spectrum with increased SNR. Furthermore, principal component (PC) spectra were calculated to estimate the variance of the mean spectrum. The mean or alternatively the first PC (PC(1)) spectrum from the same region can be used for quantitation of peak areas of metabolites in the human brain at increased SNR. Monte Carlo simulations showed that both mean and PC(1) spectra were more accurate in estimating regional metabolite concentrations than solutions that regress individual spectra against the tissue compositions of MRSI voxels. Back-to-back MRSI studies on 10 healthy volunteers showed that mean spectra markedly improved reliability of brain metabolite measurements, most notably for myo-inositol, as compared to regression methods.

Efficient simulation of magnetic resonance imaging with Bloch-Torrey equations using intra-voxel magnetization gradients

The process of image formation in magnetic resonance imaging (MRI) can be simulated by means of an iterative solution of Bloch-Torrey equations. This is a useful accessory to analyze the influence of sample properties, sequence parameters and hardware specifications on the MRI signal. In this paper, a computer algorithm is presented which is based on calculating partial derivatives of the magnetization vector. This technique allows more efficient simulation than summation of isochromats (the latter being commonly employed for this purpose) and, as a result, the effect of diffusion on the MRI signal can be calculated iteratively. A detailed description of the algorithm is given, and its feasibility for different applications is studied. It is shown that the algorithm is most applicable to simulating the effect of field perturbations, i.e. intra-voxel dephasing, but is also useful for other typical imaging experiments and the simulation of diffusion weighting.

A hybrid region-boundary model for cerebral cortical segmentation in MRI

Automatic segmentation of cerebral cortex in magnetic resonance imaging (MRI) is a challenging problem in understanding brain anatomy and functions. The difficulty is mainly due to variable brain structures, various MRI artifacts and restrictive body scanning methods. This paper describes a hybrid model-based method for obtaining an accurate and topologically-preserving segmentation of the brain cortex. The approach is based on defining region and boundary information using, respectively, level set and Bayesian techniques, and fusing these two types of information to achieve cerebral cortex segmentation. It is automatic and robust to noise, intensity inhomogeneities, and partial volume effect. Another particularity of the proposed approach is that bias field is corrected during segmentation process and that the central layer of the cortex is accurately obtained through a topology correction step. The proposed method is evaluated on both simulated and real data, and compared with existing segmentation techniques. The obtained results have demonstrated its improved performance and robustness.

High-Performance Medical Image Registration Using New Optimization Techniques

Optimization of a similarity metric is an essential component in intensity-based medical image registration. The increasing availability of parallel computers makes parallelizing some registration tasks an attractive option to increase speed. In this paper, two new deterministic, derivative-free, and intrinsically parallel optimization methods are adapted for image registration. DIviding RECTangles (DIRECT) is a global technique for linearly bounded problems, and multidirectional search (MDS) is a recent local method. The performance of DIRECT, MDS, and hybrid methods using a parallel implementation of Powell's method for local refinement, are compared. Experimental results demonstrate that DIRECT and MDS are robust, accurate, and substantially reduce computation time in parallel implementations.

A hybrid tissue segmentation approach for brain MR images

A novel hybrid algorithm for the tissue segmentation of brain magnetic resonance images is proposed. The core of the algorithm is a probabilistic neural network (PNN) in which weighting factors are added to the summation layer, such that partial volume effects can be taken into account in the modeling process. The mean vectors for the probability density function estimation and the corresponding weighting factors are generated by a hierarchical scheme involving a self-organizing map neural network and an expectation maximization algorithm. Unlike conventional PNN, this approach circumvents the need for training sets. Tissue segmentation results from various algorithms are compared and the effectiveness and robustness of the proposed approach are demonstrated.

76-space analysis of grey matter diffusivity: methods and applications

Diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI) allow in vivo investigation of molecular motion of tissue water at a microscopic level in cerebral gray matter (GM) and white matter (WM). DWI/DTI measure of water diffusion has been proven to be invaluable for the study of many neurodegenerative diseases (e.g., Alzheimer's disease and Creutzfeldt-Jakob disease) that predominantly involve GM. Thus, quantitative analysis of GM diffusivity is of scientific interest and is promised to have a clinical impact on the investigation of normal brain aging and neuropathology. In this paper, we propose an automated framework for analysis of GM diffusivity in 76 standard anatomic subdivisions of gray matter to facilitate studies of neurodegenerative and other gray matter neurological diseases. The computational framework includes three enabling technologies: (1) automatic parcellation of structural MRI GM into 76 precisely defined neuroanatomic subregions ("76-space"), (2) automated segmentation of GM, WM and CSF based on DTI data, and (3) automatic measurement of the average apparent diffusion coefficient (ADC) in each segmented GM subregion. We evaluate and validate this computational framework for 76-space GM diffusivity analysis using data from normal volunteers and from patients with Creutzfeldt-Jakob disease.

Realistic Simulated MRI and SPECT Databases

This paper describes the construction of simulated SPECT and MRI databases that account for realistic anatomical and functional variability. The data is used as a gold-standard to evaluate four SPECT/MRI similarity-based registration methods. Simulation realism was accounted for using accurate physical models of data generation and acquisition. MRI and SPECT simulations were generated from three subjects to take into account inter-subject anatomical variability. Functional SPECT data were computed from six functional models of brain perfusion. Previous models of normal perfusion and ictal perfusion observed in Mesial Temporal Lobe Epilepsy (MTLE) were considered to generate functional variability. We studied the impact noise and intensity non-uniformity in MRI simulations and SPECT scatter correction may have on registration accuracy. We quantified the amount of registration error caused by anatomical and functional variability. Registration involving ictal data was less accurate than registration involving normal data. MR intensity nonuniformity was the main factor decreasing registration accuracy. The proposed simulated database is promising to evaluate many functional neuroimaging methods, involving MRI and SPECT data.

Comparison and validation of tissue modelization and statistical classification methods in T1-weighted MR brain images

This paper presents a validation study on statistical nonsupervised brain tissue classification techniques in magnetic resonance (MR) images. Several image models assuming different hypotheses regarding the intensity distribution model, the spatial model and the number of classes are assessed. The methods are tested on simulated data for which the classification ground truth is known. Different noise and intensity nonuniformities are added to simulate real imaging conditions. No enhancement of the image quality is considered either before or during the classification process. This way, the accuracy of the methods and their robustness against image artifacts are tested. Classification is also performed on real data where a quantitative validation compares the methods' results with an estimated ground truth from manual segmentations by experts. Validity of the various classification methods in the labeling of the image as well as in the tissue volume is estimated with different local and global measures. Results demonstrate that methods relying on both intensity and spatial information are more robust to noise and field inhomogeneities. We also demonstrate that partial volume is not perfectly modeled, even though methods that account for mixture classes outperform methods that only consider pure Gaussian classes. Finally, we show that simulated data results can also be extended to real data.

Inhomogeneity correction for brain magnetic resonance images by rank leveling

OBJECTIVE

A postprocessing method of rank filtering inhomogeneity correction using nonlinear rank filtering of magnetic resonance imaging (MRI) scans is described. The method addresses some of the problems of homomorphic unsharp masking (HUM) using mean or median filtering.

METHODS

Maximum rank filtering was used to estimate the bias image, which was then smoothed and used to normalize the original image. The coefficient of variation within and between tissue classes before and after inhomogeneity correction was calculated in simulated brain phantom images and clinical T1-weighted MRI images. Comparison was made with mean filter-based and median filter-based HUM.

RESULTS

Maximum rank filtering reduced within and between class coefficients of variation. Performance of median filtering was inferior to that of mean filtering, and both were inferior to performance of maximum rank filtering.

CONCLUSION

The method is easy to implement and is effective against different bias types. It is less prone to edge effects than mean and median filtering.

MRI Fuzzy Segmentation of Brain Tissue Using Neighborhood Attraction With Neural-Network Optimization

Image segmentation is an indispensable process in the visualization of human tissues, particularly during clinical analysis of magnetic resonance (MR) images. Unfortunately, MR images always contain a significant amount of noise caused by operator performance, equipment, and the environment, which can lead to serious inaccuracies with segmentation. A robust segmentation technique based on an extension to the traditional fuzzy c-means (FCM) clustering algorithm is proposed in this paper. A neighborhood attraction, which is dependent on the relative location and features of neighboring pixels, is shown to improve the segmentation performance dramatically. The degree of attraction is optimized by a neural-network model. Simulated and real brain MR images with different noise levels are segmented to demonstrate the superiority of the proposed technique compared to other FCM-based methods. This segmentation method is a key component of an MR image-based classification system for brain tumors, currently being developed. Index Terms-Improved fuzzy c-means clustering (IFCM), magnetic resonance imaging (MRI), neighborhood attraction, segmentation.

N-acetylaspartate reductions in the mediodorsal and anterior thalamus in men with schizophrenia verified by tissue volume corrected proton MRSI

OBJECTIVE

Deficits in the mediodorsal and anterior nuclei of the thalamus may contribute to the psychopathological symptoms of schizophrenia. These thalamic nuclei have been found to be abnormal in schizophrenia and have close connections with other brain structures implicated in the disorder. We therefore examined schizophrenia-related alterations in brain metabolite levels specifically in the mediodorsal and anterior thalamic subregions.

METHOD

We used in vivo proton magnetic resonance spectroscopic imaging ((1)H MRSI) to measure N-acetylaspartate (NAA), choline-containing compounds (Cho), and creatine+phosphocreatine (Cr) in the mediodorsal and anterior thalamus in 22 male patients with schizophrenia and 22 male controls. Magnetic resonance imaging (MRI) tissue segmentation and thalamic volume mask techniques were performed to distinguish the thalamus, extrathalamic gray and white matter, and CSF within the spectroscopic voxels.

RESULTS

Compared to healthy subjects, patients with schizophrenia had significantly lower NAA in the mediodorsal and anterior thalamus bilaterally. No significant differences in Cho or Cr levels were seen. NAA was significantly higher in the left thalamus relative to the right in both groups. We found a strong negative correlation between left thalamic NAA and duration of illness, even after partialling out the effect of age. Tissue segmentation and thalamic volume mask techniques detected no group or lateralized differences in tissue type or CSF percentages, demonstrating that the metabolite reductions were not an artifact of spectroscopic voxel heterogeneity.

CONCLUSIONS

These findings suggest diminished function and/or structure in the mediodorsal and anterior thalamus in male patients with schizophrenia and support earlier research demonstrating schizophrenia-related abnormalities in the thalamus and its circuitry.

Mapping the dimensionality, density and topology of data: the growing adaptive neural gas

Self-organized maps are commonly applied for tasks of cluster analysis, vector quantization or interpolation. The artificial neural network model introduced in this paper is a hybrid model of the growing neural gas model introduced by Fritzke (Fritzke, in Advances in Neural Information Processing Systems 7, MIT Press, Cambridge MA, 1995) and the adaptive resolution clustering modification for self-organized maps proposed by Firenze (Firenze et al., in International Conference on Artificial Neural Networks, Springer-Verlag, London, 1994). The hybrid model is capable of mapping the distribution, dimensionality and topology of the input data. It has a local performance measure that enables the network to terminate growing in areas of the input space that is mapped by units reaching a performance goal. Therefore the network can accurately map clusters of data appearing on different scales of density. The capabilities of the algorithm are tested using simulated datasets with similar spatial spread but different local density distributions, and a simulated multivariate MR dataset of an anatomical human brain phantom with mild multiple sclerosis lesions. These tests demonstrate the advantages of the model compared to the growing neural gas algorithm when adaptive mapping of areas with low sample density is desirable.

The SIMRI project: a versatile and interactive MRI simulator

This paper gives an overview of SIMRI, a new 3D MRI simulator based on the Bloch equation. This simulator proposes an efficient management of the T2* effect, and in a unique simulator integrates most of the simulation features that are offered in different simulators. It takes into account the main static field value and enables realistic simulations of the chemical shift artifact, including off-resonance phenomena. It also simulates the artifacts linked to the static field inhomogeneity like those induced by susceptibility variation within an object. It is implemented in the C language and the MRI sequence programming is done using high level C functions with a simple programming interface. To manage large simulations, the magnetization kernel is implemented in a parallelized way that enables simulation on PC grid architecture. Furthermore, this simulator includes a 1D interactive interface for pedagogic purpose illustrating the magnetization vector motion as well as the MRI contrasts.

A statistically based flow for image segmentation

In this paper we present a new algorithm for 3D medical image segmentation. The algorithm is versatile, fast, relatively simple to implement, and semi-automatic. It is based on minimizing a global energy defined from a learned non-parametric estimation of the statistics of the region to be segmented. Implementation details are discussed and source code is freely available as part of the 3D Slicer project. In addition, a new unified set of validation metrics is proposed. Results on artificial and real MRI images show that the algorithm performs well on large brain structures both in terms of accuracy and robustness to noise.

Fast and robust parameter estimation for statistical partial volume models in brain MRI

Due to the finite spatial resolution of imaging devices, a single voxel in a medical image may be composed of mixture of tissue types, an effect known as partial volume effect (PVE). Partial volume estimation, that is, the estimation of the amount of each tissue type within each voxel, has received considerable interest in recent years. Much of this work has been focused on the mixel model, a statistical model of PVE. We propose a novel trimmed minimum covariance determinant (TMCD) method for the estimation of the parameters of the mixel PVE model. In this method, each voxel is first labeled according to the most dominant tissue type. Voxels that are prone to PVE are removed from this labeled set, following which robust location estimators with high breakdown points are used to estimate the mean and the covariance of each tissue class. Comparisons between different methods for parameter estimation based on classified images as well as expectation--maximization-like (EM-like) procedure for simultaneous parameter and partial volume estimation are reported. The robust estimators based on a pruned classification as presented here are shown to perform well even if the initial classification is of poor quality. The results obtained are comparable to those obtained using the EM-like procedure, but require considerably less computation time. Segmentation results of real data based on partial volume estimation are also reported. In addition to considering the parameter estimation problem, we discuss differences between different approximations to the complete mixel model. In summary, the proposed TMCD method allows for the accurate, robust, and efficient estimation of partial volume model parameters, which is crucial to a variety of brain MRI data analysis procedures such as the accurate estimation of tissue volumes and the accurate delineation of the cortical surface.

Information theoretic similarity measures in non-rigid registration

Mutual Information (MI) and Normalised Mutual Information (NMI) have enjoyed success as image similarity measures in medical image registration. More recently, they have been used for non-rigid registration, most often evaluated empirically as functions of changing registration parameter. In this paper we present expressions derived from intensity histogram representations of these measures, for their change in response to a local perturbation of a deformation field linking two images. These expressions give some insight into the operation of NMI in registration and are implemented as driving forces within a fluid registration framework. The performance of the measures is tested on publicly available simulated multi-spectral MR brain images.

Classification of Anatomical Structures in MR Brain Images Using Fuzzy Parameters

We present an algorithm that automatically segments and classifies the brain structures in a set of magnetic resonance (MR) brain images using expert information contained in a small subset of the image set. The algorithm is intended to do the segmentation and classification tasks mimicking the way a human expert would reason. The algorithm uses a knowledge base taken from a small subset of semiautomatically classified images that is combined with a set of fuzzy indexes that capture the experience and expectation a human expert uses during recognition tasks. The fuzzy indexes are tissue specific and spatial specific, in order to consider the biological variations in the tissues and the acquisition inhomogeneities through the image set. The brain structures are segmented and classified one at a time. For each brain structure the algorithm needs one semiautomatically classified image and makes one pass through the image set. The algorithm uses low-level image processing techniques on a pixel basis for the segmentations, then validates or corrects the segmentations, and makes the final classification decision using higher level criteria measured by the set of fuzzy indexes. We use single-echo MR images because of their high volumetric resolution; but even though we are working with only one image per brain slice, we have multiple sources of information on each pixel: absolute and relative positions in the image, gray level value, statistics of the pixel and its three-dimensional neighborhood and relation to its counterpart pixels in adjacent images. We have validated our algorithm for ease of use and precision both with clinical experts and with measurable error indexes over a Brainweb simulated MR set.

MRI simulator with object-specific field map calculations

A new MRI simulator has been developed that generates images of realistic objects for arbitrary pulse sequences executed in the presence of static field inhomogeneities, including those due to magnetic susceptibility, variations in the applied field, and chemical shift. In contrast to previous simulators, this system generates object-specific inhomogeneity patterns from first principles and propagates the consequent frequency offsets and intravoxel dephasing through the acquisition protocols to produce images with realistic artifacts. The simulator consists of two parts. The input to part 1 is a set of "susceptibility voxels" that describe the magnetic properties of the object being imaged. It calculates a frequency offset for each voxel by computing the size of the static field offset at each voxel in the image based on the magnetic susceptibility of each tissue type within all voxels. The method of calculation is a three-dimensional convolution of the susceptibility-voxels with a kernel derived from a previously published method and takes advantage of the superposition principle to include voxels with mixtures of substances of differing susceptibilities. Part 2 produces both a signal and a reconstructed image. Its inputs include a voxel-based description of the object, frequency offsets computed by part 1, applied static field errors, chemical shift values, and a description of the imaging protocol. Intravoxel variations in both static field and time-dependent phase are calculated for each voxel. Validations of part 1 are presented for a known analytic solution and for experimental data from two phantoms. Part 2 was validated with comparisons to an independent simulation provided by the Montreal Neurological Institute and experimental data from a phantom.

An automated algorithm for the computation of brain volume change from sequential MRIs using an iterative principal component analysis and its evaluation for the assessment of whole-brain atrophy rates in patients with probable Alzheimer's disease

This article introduces an automated method for the computation of changes in brain volume from sequential magnetic resonance images (MRIs) using an iterative principal component analysis (IPCA) and demonstrates its ability to characterize whole-brain atrophy rates in patients with Alzheimer's disease (AD). The IPCA considers the voxel intensity pairs from coregistered MRIs and identifies those pairs a sufficiently large distance away from the iteratively determined PCA major axis. Analyses of simulated and real MRI data support the underlying assumption of a linear relationship in paired voxel intensities, identify an outlier distance threshold that optimizes the trade-off between sensitivity and specificity in the detection of small volume changes while accounting for global intensity changes, and demonstrate an ability to detect changes as small as 0.04% of brain volume without confounding effects of between-scan shifts in voxel intensity. In eight patients with probable AD and eight age-matched normal control subjects, the IPCA was comparable to the established but partly manual digital subtraction (DS) method in characterizing annual rates of whole-brain atrophy: resulting rates were correlated (Spearman rank correlation = 0.94, P < 0.0005) and comparable in distinguishing probable AD from normal aging (IPCA-detected atrophy rates: 2.17 +/- 0.52% per year in the patients vs. 0.41 +/- 0.22% per year in the controls [Wilcoxon-Mann-Whitney test P = 7.8 x 10(-4)]; DS-detected atrophy rates: 3.51 +/- 1.31% per year in the patients vs. 0.48 +/- 0.29% per year in the controls [P = 7.8 x 10(-4)]). The IPCA could be used in tracking the progression of AD, evaluating the disease-modifying effects of putative treatments, and investigating the course of other normal and pathological changes in brain morphology.