Current methods in medical image segmentation

Atlas guided identification of brain structures by combining 3D segmentation and SVM classification

This study presents a novel automatic approach for the identification of anatomical brain structures in magnetic resonance images (MRI). The method combines a fast multiscale multi-channel three dimensional (3D) segmentation algorithm providing a rich feature vocabulary together with a support vector machine (SVM) based classifier. The segmentation produces a full hierarchy of segments, expressed by an irregular pyramid with only linear time complexity. The pyramid provides a rich, adaptive representation of the image, enabling detection of various anatomical structures at different scales. A key aspect of the approach is the thorough set of multiscale measures employed throughout the segmentation process which are also provided at its end for clinical analysis. These features include in particular the prior probability knowledge of anatomic structures due to the use of an MRI probabilistic atlas. An SVM classifier is trained based on this set of features to identify the brain structures. We validated the approach using a gold standard real brain MRI data set. Comparison of the results with existing algorithms displays the promise of our approach.

Segmenting magnetic resonance images via hierarchical mixture modelling

We present a statistically innovative as well as scientifically and practically relevant method for automatically segmenting magnetic resonance images using hierarchical mixture models. Our method is a general tool for automated cortical analysis which promises to contribute substantially to the science of neuropsychiatry. We demonstrate that our method has advantages over competing approaches on a magnetic resonance brain imagery segmentation task.

A multiresolution diffused expectation-maximization algorithm for medical image segmentation

In this paper a new method for segmenting medical images is presented, the multiresolution diffused expectation-maximization (MDEM) algorithm. The algorithm operates within a multiscale framework, thus taking advantage of the fact that objects/regions to be segmented usually reside at different scales. At each scale segmentation is carried out via the expectation-maximization algorithm, coupled with anisotropic diffusion on classes, in order to account for the spatial dependencies among pixels. This new approach is validated via experiments on a variety of medical images and its performance is compared with more standard methods.

Segmentation of subcortical brain structures using fuzzy templates

We propose a novel method to automatically segment subcortical structures of human brain in magnetic resonance images by using fuzzy templates. A set of fuzzy templates of the structures based on features such as intensity, spatial location, and relative spatial relationship among structures are first created from a set of training images by defining the fuzzy membership functions and by fusing the information of features. Segmentation is performed by registering the fuzzy templates of the structures on the test image and then by fusing them with the tissue maps of the test image. The final decision is taken in order to optimize the certainty in the intensity, location, relative position, and tissue content of the structure. Our method does not require specific expert definition of each structure or manual interactions during segmentation process. The technique is demonstrated with the segmentation of five structures: thalamus, putamen, caudate, hippocampus, and amygdala; the performance of the present method is comparable with previous techniques.

An integrated environment for plastic surgery support: building virtual patients, simulating interventions, and supporting intraoperative decisions

In the last decade a number of environments for Computer Supported Plastic Surgery have been presented. Nevertheless, an overall approach for training and intraoperative support is still missing or has not been widely exploited yet. We developed a fully integrated system which allows surgical simulation, planning, and support for computer-guided plastic surgery procedures starting from image acquisition to final intraoperative assistance. The system also provides the user with a radiological workstation able to analyse patient medical images and case studies, with advanced bidimensional and three dimensional image processing functionalities. We intend to demonstrate that such a platform can be built at an affordable cost. The radiological workstation is capable of supporting radiologists and surgeons in real patient case studies and the simulation workstation may be adopted by plastic surgeons in teaching and training of complex surgical planning. Moreover, results of simulation can be used in the operating room with a relatively high benefit in terms of improved accuracy, reduction of surgical risks, and decrease in training costs.

An automated method for analysis of flow characteristics of circulating particles from in vivo video microscopy

The behavior of white and red blood cells, platelets, and circulating injected particles is one of the most studied areas of physiology. Most methods used to analyze the circulatory patterns of cells are time consuming. We describe a system named CellTrack, designed for fully automated tracking of circulating cells and micro-particles and retrieval of their behavioral characteristics. The task of automated blood cell tracking in vessels from in vivo video is particularly challenging because of the blood cells' nonrigid shapes, the instability inherent in in vivo videos, the abundance of moving objects and their frequent superposition. To tackle this, the CellTrack system operates on two levels: first, a global processing module extracts vessel borders and center lines based on color and temporal patterns. This enables the computation of the approximate direction of the blood flow in each vessel. Second, a local processing module extracts the locations and velocities of circulating cells. This is performed by artificial neural network classifiers that are designed to detect specific types of blood cells and micro-particles. The motion correspondence problem is then resolved by a novel algorithm that incorporates both the local and the global information. The system has been tested on a series of in vivo color video recordings of rat mesentery. Our results show that the synergy between the global and local information enables CellTrack to overcome many of the difficulties inherent in tracking methods that rely solely on local information. A comparison was made between manual measurements and the automatically extracted measurements of leukocytes and fluorescent microspheres circulatory velocities. This comparison revealed an accuracy of 97%. CellTrack also enabled a much larger volume of sampling in a fraction of time compared to the manual measurements. All these results suggest that our method can in fact constitute a reliable replacement for manual extraction of blood flow characteristics from in vivo videos.

Adaptive model initialization and deformation for automatic segmentation of T1-weighted brain MRI data

A fully automatic, two-step, T1-weighted brain magnetic resonance imaging (MRI) segmentation method is presented. A preliminary mask of parenchyma is first estimated through adaptive image intensity analysis and mathematical morphological operations. It serves as the initial model and probability reference for a level-set algorithm in the second step, which finalizes the segmentation based on both image intensity and geometric information. The Dice coefficient and Euclidean distance between boundaries of automatic results and the corresponding references are reported for both phantom and clinical MR data. For the 28 patient scans acquired at our institution, the average Dice coefficient was 98.2% and the mean Euclidean surface distance measure was 0.074 mm. The entire segmentation for either a simulated or a clinical image volume finishes within 2 min on a modern PC system. The accuracy and speed of this technique allow us to automatically create patient-specific finite element models within the operating room on a timely basis for application in image-guided updating of preoperative scans.

Interplay between intensity standardization and inhomogeneity correction in MR image processing

Image intensity standardization is a postprocessing method designed for correcting acquisition-to-acquisition signal intensity variations (nonstandardness) inherent in magnetic resonance (MR) images. Inhomogeneity correction is a process used to suppress the low frequency background nonuniformities (inhomogeneities) of the image domain that exist in MR images. Both these procedures have important implications in MR image analysis. The effects of these postprocessing operations on improvement of image quality in isolation has been well documented. However, the combined effects of these two processes on MR images and how the processes influence each other have not been studied thus far. In this paper, we evaluate the effect of inhomogeneity correction followed by standardization and vice-versa on MR images in order to determine the best sequence to follow for enhancing image quality. We conducted experiments on several clinical and phantom data sets (nearly 4000 three-dimensional MR images were analyzed) corresponding to four different MRI protocols. Different levels of artificial nonstandardness, and different models and levels of artificial background inhomogeneity were used in these experiments. Our results indicate that improved standardization can be achieved by preceding it with inhomogeneity correction. There is no statistically significant difference in image quality obtained between the results of standardization followed by correction and that of correction followed by standardization from the perspective of inhomogeneity correction. The correction operation is found to bias the effect of standardization. We demonstrate this bias both qualitatively and quantitatively by using two different methods of inhomogeneity correction. We also show that this bias in standardization is independent of the specific inhomogeneity correction method used. The effect of this bias due to correction was also seen in magnetization transfer ratio (MTR) images, which are naturally endowed with the standardness property. Standardization, on the other hand, does not seem to influence the correction operation. It is also found that longer sequences of repeated correction and standardization operations do not considerably improve image quality. These results were found to hold for the clinical and the phantom data sets, for different MRI protocols, for different levels of artificial nonstandardness, for different models and levels of artificial inhomogeneity, for different correction methods, and for images that were endowed with inherent standardness as well as for those that were standardized by using the intensity standardization method. Overall, we conclude that inhomogeneity correction followed by intensity standardization is the best sequence to follow from the perspective of both image quality and computational efficiency.

Multivariate image analysis in biomedicine

In recent years, multivariate imaging techniques are developed and applied in biomedical research in an increasing degree. In research projects and in clinical studies as well m-dimensional multivariate images (MVI) are recorded and stored to databases for a subsequent analysis. The complexity of the m-dimensional data and the growing number of high throughput applications call for new strategies for the application of image processing and data mining to support the direct interactive analysis by human experts. This article provides an overview of proposed approaches for MVI analysis in biomedicine. After summarizing the biomedical MVI techniques the two level framework for MVI analysis is illustrated. Following this framework, the state-of-the-art solutions from the fields of image processing and data mining are reviewed and discussed. Motivations for MVI data mining in biology and medicine are characterized, followed by an overview of graphical and auditory approaches for interactive data exploration. The paper concludes with summarizing open problems in MVI analysis and remarks upon the future development of biomedical MVI analysis.

ANATOMICAL GUIDED SEGMENTATION WITH NON-STATIONARY TISSUE CLASS DISTRIBUTIONS IN AN EXPECTATION-MAXIMIZATION FRAMEWORK

High quality segmentation of brain MR images is a challenging task. To deal with this problem many automatic segmentation methods rely on atlas information of anatomical structures. We further investigate this line of research by introducing hierarchical representations of anatomical structures in an Expectation-Maximization framework. This new approach enables us to divide a complex segmentation scenario into less difficult sub-problems reducing the scenario's statistical complexity. We demonstrate the method's strength by segmenting a set of brain MR images into 31 different anatomical structures as well as comparing it to other methods.

Computational anatomy: shape, growth, and atrophy comparison via diffeomorphisms

Computational anatomy (CA) is the mathematical study of anatomy I in I = I(alpha) o G, an orbit under groups of diffeomorphisms (i.e., smooth invertible mappings) g in G of anatomical exemplars I(alpha) in I. The observable images are the output of medical imaging devices. There are three components that CA examines: (i) constructions of the anatomical submanifolds, (ii) comparison of the anatomical manifolds via estimation of the underlying diffeomorphisms g in G defining the shape or geometry of the anatomical manifolds, and (iii) generation of probability laws of anatomical variation P(.) on the images I for inference and disease testing within anatomical models. This paper reviews recent advances in these three areas applied to shape, growth, and atrophy.

Monte Carlo modelling of a voxel head phantom for in vivo measurement of bone-seeker nuclides

Whole-body counters (WBCs) are used for the assessment of the internal contamination of actinides in the human body. WBCs require adequate calibration procedures that rely on the use of suitable calibration phantoms. A previous study carried out at the ENEA-Radiation Protection Institute was aimed at designing a head calibration phantom in which a heterogeneous distribution of 241Am point sources could satisfactorily approximate an assumed homogeneous contamination throughout the head bones. Suitable correction factors for the WBC detection efficiencies were evaluated with Monte Carlo. The present paper summarises the main aspects and implications of an advanced modelling technique based on a VOXEL approach. The methodology could be extended to other bone-seeker radionuclides.

A novel kernelized fuzzy C-means algorithm with application in medical image segmentation

Image segmentation plays a crucial role in many medical imaging applications. In this paper, we present a novel algorithm for fuzzy segmentation of magnetic resonance imaging (MRI) data. The algorithm is realized by modifying the objective function in the conventional fuzzy C-means (FCM) algorithm using a kernel-induced distance metric and a spatial penalty on the membership functions. Firstly, the original Euclidean distance in the FCM is replaced by a kernel-induced distance, and thus the corresponding algorithm is derived and called as the kernelized fuzzy C-means (KFCM) algorithm, which is shown to be more robust than FCM. Then a spatial penalty is added to the objective function in KFCM to compensate for the intensity inhomogeneities of MR image and to allow the labeling of a pixel to be influenced by its neighbors in the image. The penalty term acts as a regularizer and has a coefficient ranging from zero to one. Experimental results on both synthetic and real MR images show that the proposed algorithms have better performance when noise and other artifacts are present than the standard algorithms.

Robust Image Segmentation Using FCM With Spatial Constraints Based on New Kernel-Induced Distance Measure

Fuzzy c-means clustering (FCM) with spatial constraints (FCM_S) is an effective algorithm suitable for image segmentation. Its effectiveness contributes not only to the introduction of fuzziness for belongingness of each pixel but also to exploitation of spatial contextual information. Although the contextual information can raise its insensitivity to noise to some extent, FCM_S still lacks enough robustness to noise and outliers and is not suitable for revealing non-Euclidean structure of the input data due to the use of Euclidean distance (L2 norm). In this paper, to overcome the above problems, we first propose two variants, FCM_S1 and FCM_S2, of FCM_S to aim at simplifying its computation and then extend them, including FCM_S, to corresponding robust kernelized versions KFCM_S, KFCM_S1 and KFCM_S2 by the kernel methods. Our main motives of using the kernel methods consist in: inducing a class of robust non-Euclidean distance measures for the original data space to derive new objective functions and thus clustering the non-Euclidean structures in data; enhancing robustness of the original clustering algorithms to noise and outliers, and still retaining computational simplicity. The experiments on the artificial and real-world datasets show that our proposed algorithms, especially with spatial constraints, are more effective.

Semiautomated thermal lesion segmentation for three-dimensional elastographic imaging

Several studies have demonstrated that lesion volumes computed from multiple planar slices through the region-of-interest (ROI) are more accurate than volumes estimated assuming simple shapes and incorporating single or orthogonal diameter estimates. However, manual delineation of boundaries on multiple planar 2-D images is tedious and labor-intensive. Automatic extraction of lesion boundaries is, therefore, attractive and imperative to remove subjectivity and reduce assessment time. This paper presents a semiautomated segmentation algorithm for thermal lesions on 3-D elastographic data to obtain both area and volume information. The semiautomated segmentation algorithm is based on thresholding and morphologic opening of both 2-D and 3-D elastographic data. Results obtained on 44 thermal lesions imaged in vitro using elastography were compared to manual delineation of both elastographic and pathology images. Results obtained using semiautomated segmentation demonstrate a close correspondence with manual delineation results. However, area and volume estimates obtained using both manual and semiautomated segmentation of lesions seen on elastograms slightly underestimate areas and volumes measured from pathology.

Three validation metrics for automated probabilistic image segmentation of brain tumours

The validity of brain tumour segmentation is an important issue in image processing because it has a direct impact on surgical planning. We examined the segmentation accuracy based on three two-sample validation metrics against the estimated composite latent gold standard, which was derived from several experts' manual segmentations by an EM algorithm. The distribution functions of the tumour and control pixel data were parametrically assumed to be a mixture of two beta distributions with different shape parameters. We estimated the corresponding receiver operating characteristic curve, Dice similarity coefficient, and mutual information, over all possible decision thresholds. Based on each validation metric, an optimal threshold was then computed via maximization. We illustrated these methods on MR imaging data from nine brain tumour cases of three different tumour types, each consisting of a large number of pixels. The automated segmentation yielded satisfactory accuracy with varied optimal thresholds. The performances of these validation metrics were also investigated via Monte Carlo simulation. Extensions of incorporating spatial correlation structures using a Markov random field model were considered.

Alterations in calretinin immunostaining in the ferret superior olivary complex after cochlear ablation

In this study, we used image analysis to assess changes in calretinin immunoreactivity in the lateral (LSO) and medial (MSO) superior olivary nuclei in ferrets 2 months after unilateral cochlear ablations at 30-40 days of age, soon after hearing onset. These two nuclei are the first significant sites of binaural convergence in the ascending auditory system, and both receive direct projections from the deafferented cochlear nucleus. Cochlear ablation results in a decrease in the overall level of calretinin immunostaining within the LSO ipsilaterally compared with the contralateral side and with control animals and within the MSO bilaterally compared with control ferrets. In addition, the level of calretinin immunostaining ipsilaterally within neurons in the LSO was significantly less in cochlear ablated than control animals. In contrast, there was no effect of cochlear ablation on the level of calretinin immunostaining within neurons either in the contralateral LSO or in the MSO. These results are consistent with a downregulation in calretinin within the neuropil of MSO bilaterally and LSO ipsilaterally, as well as a downregulation in calretinin within somata in the ipsilateral LSO as a result of unilateral cochlear ablation soon after hearing onset. Thus, cochlear-driven activity appears to affect calcium binding protein levels in both neuropil and neurons within the superior olivary complex.

Digitizing life at the level of the cell: high-performance laser-scanning microscopy and image analysis for in toto imaging of development

The field of biological imaging is progressing at an amazing rate. Advances in both laser-scanning microscopy and green fluorescent protein (GFP) technology are combining to make possible imaging-based approaches for studying developmental mechanisms that were previously impossible. Modern confocal and multi-photon microscopes are pushing the envelope of speed, sensitivity, spectral resolution, and depth resolution to allow in vivo imaging of whole, live embryos at cellular resolution over extended periods of time. In toto imaging, in which nearly every cell in an embryo or tissue can be tracked through space and time during development, may become a standard technique for small transparent embryos such as zebrafish and early stage chick and mouse embryos. GFP and its spectral variants can be used to mark a wide range of in vivo biological information for in toto imaging including gene expression patterns, mutant phenotypes, and protein subcellular localization patterns. Combining in toto imaging and GFP transgenic approaches on a large scale may usher in an explosion of in vivo, developmental data as has happened in the past several years with genomic data. There are significant challenges that must be met to reach these goals. This paper will discuss the current state-of-the-art, the challenges, and the prospects of in toto imaging in the areas of imaging, image analysis, and informatics.

Geometric atlas: modeling the cortex as an organized surface

Recent atlases of the cortical surface are based on a modelization of the cerebral cortex as a topological sphere. This captures effectively its organization as a regular bidimensional sheet of layers parallel to the surface and with perpendicular cortical columns. Yet, while in the vertical direction cortices are almost the same throughout phylia, in the sense of its surface the cerebral cortex is one of the most variable and distinctive parts of the nervous system. Indeed, gyri and sulci appear to have a crucial organizing role in an architectonic, connectional, and functional sense. This organization is not explicitly captured by the surface model of the cortex. We propose a geometric model of the cortical anatomy based on flat representations of principal sulci obtained from surface reconstructions of MRI data, and on neuroanatomical and theoretical considerations concerning the folding patterns of the cortex. The cortex is modeled by a sphere where primary sulci are included as axes. The arrangement of the axes is a simplification of the arrangement of principal sulci observed in flat stereographic representations of the whole cortical surface. The position of secondary and tertiary sulci is then defined by a field of orientations parallel and orthogonal to the axes. We consider the use of the geometric model as a synthetic reference cortex for addressing reconstructions of cortical surfaces. We present a method which establishes a bijection between the geometric model and a cortical surface reconstruction by using the axes of the model as boundary conditions for a set of partial differential equations solved over both surfaces. Using the geometric model as atlas provides a natural parameterization of the cortical surface that, unlike angular coordinates, allows for a localization based on the surface distance to its main organizing landmarks and folding patterns.

Fluorescent ligands, antibodies, and proteins for the study of receptors

Fluorescent molecules bound to receptors can show their location and, if binding is reversible, can provide pharmacological information such as affinity and proximity between interacting molecules. The spatial precision offered by visualisation transcends the diverse localisation and low molecular concentration of receptor molecules. Consequently, the relationships between receptor location and function and life cycles of receptors have become better understood as a result of fluorescent labeling. Each of these aspects contributes new insights to drug action and potential new targets. The relationships between spatial distribution of receptor and function are largely unknown. This is particularly apparent for native receptors expressed in their normal host tissues where communication between heterogeneous cell types influences receptor distribution and function. In cultured cell systems, particularly for G-protein-coupled receptors (GPCR), fluorescence-based methods have enabled the visualisation of the cycle of agonist-stimulated receptor clustering, endocytic internalisation to the perinuclear region, degradation of the receptor-ligand complex, and recycling back to the surface membrane. Using variant forms of green fluorescent protein (GFP), antibodies, or fluorescent ligands, it is possible to detect or visualise the formation of oligomeric receptor complexes. Careful selection of fluorescent molecules based on their spectral properties enables resonance energy transfer and multilabel visualisation with colocalisation studies. Fluorescent agonist and antagonist ligands are now being used in parallel with GFP to study receptor cycling in live cells. This review covers how labeling and visualisation technologies have been applied to the study of major pharmacologically important receptors and illustrates this by giving examples of recent techniques that have relied on GFP, antibodies, or fluorescent ligands alone or in combination for the purpose of studying GPCR.

Quantitative image analysis: software systems in drug development trials

Multi-dimensional image analysis is being used increasingly to arrive at surrogate end-points for drug development trials. Various imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET) and ultrasound are used to analyze treatments for diseases such as cancer, multiple sclerosis, osteoarthritis, and Alzheimer's disease. However, extracting information from images can be tedious and is prone to high user variability. The medical image analysis community is moving towards advanced software systems specifically designed for drug development trials. These systems can automatically identify the anatomy of interest in medical images (segmentation methods), can compare the anatomy over time or between patients (registration methods) and allow the quantitative extraction of anatomical features and the integration of the data and results into a database management system, automatically tracking the changes made to the data (audit trail generation). In this article, we present a case study using a prototype system that is used for quantifying multiple sclerosis lesions from multivariate MRI.

A new accurate and precise 3-D segmentation method for skeletal structures in volumetric CT data

We developed a highly automated three-dimensionally based method for the segmentation of bone in volumetric computed tomography (CT) datasets. The multistep approach starts with three-dimensional (3-D) region-growing using local adaptive thresholds followed by procedures to correct for remaining boundary discontinuities and a subsequent anatomically oriented boundary adjustment using local values of cortical bone density. We describe the details of our approach and show applications in the proximal femur, the knee, and the skull. The accuracy of the determination of geometrical parameters was analyzed using CT scans of the semi-anthropomorphic European spine phantom. Depending on the settings of the segmentation parameters cortical thickness could be determined with an accuracy corresponding to the side length of 1 to 2.5 voxels. The impact of noise on the segmentation was investigated by artificially adding noise to the CT data. An increase in noise by factors of two and five changed cortical thickness corresponding to the side length of one voxel. Intraoperator and interoperator precision was analyzed by repeated analysis of nine pelvic CT scans. Precision errors were smaller than 1% for trabecular and total volumes and smaller than 2% for cortical thickness. Intraoperator and interoperator precision errors were not significantly different. Our segmentation approach shows: 1) high accuracy and precision and is 2) robust to noise, 3) insensitive to user-defined thresholds, 4) highly automated and fast, and 5) easy to initialize.

MRI tissue characterization of experimental cerebral ischemia in rat

PURPOSE

To extend the ISODATA image segmentation method to characterize tissue damage in stroke, by generating an MRI score for each tissue that corresponds to its histological damage.

MATERIALS AND METHODS

After preprocessing and segmentation (using ISODATA clustering), the proposed method scores tissue regions between 1 and 100. Score 1 is assigned to normal brain matter (white or gray matter), and score 100 to cerebrospinal fluid (CSF). Lesion zones are assigned a score based on their relative levels of similarities to normal brain matter and CSF. To evaluate the method, 15 rats were imaged by a 7T MRI system at one of three time points (acute, subacute, chronic) after MCA occlusion. Then they were killed and their brains were sliced and prepared for histological studies. MRI of two or three slices of each rat brain (using two DWI (b = 400, b = 800), one PDWI, one T2WI, and one T1WI) was performed, and an MRI score between 1 and 100 was determined for each region. Segmented regions were mapped onto the histology images and scored on a scale of 1-10 by an experienced pathologist. The MRI scores were validated by comparison with histology scores. To this end, correlation coefficients between the two scores (MRI and histology) were determined.

RESULTS

Experimental results showed excellent correlations between MRI and histology scores at different time points. Depending on the reference tissue (gray matter or white matter) used in the standardization, the correlation coefficients ranged from 0.73 (P < 0.0001) to 0.78 (P < 0.0001) using the entire dataset, including acute, subacute, and chronic time points. This suggests that the proposed multiparametric approach accurately identified and characterized ischemic tissue in a rat model of cerebral ischemia at different stages of stroke evolution.

CONCLUSION

The proposed approach scores tissue regions and characterizes them using unsupervised clustering and multiparametric image analysis techniques. The method can be used for a variety of applications in the field of computer-aided diagnosis and treatment, including evaluation of response to treatment. For example, volume changes for different zones of the lesion over time (e.g., tissue recovery) can be evaluated.

Problems of interactive segmentation

Computer-assisted surgery makes a patient-individual treatment feasible, aiming at decreased surgical risk and reduced recovery time of patients. At present, in areas of application, e.g., heart surgery as well as craniofacial surgery, its use is still limited to complex cases due to the high effort. In surgical planning, it is caused by extensive medical image analysis, including tissue classification. Especially, the classification (or segmentation) requires a lot of manual intervention. For a long time research has been devoted solely to computational aspects of segmentation, where usability aspects has been out of scope. This article focuses on the major problems of interactive segmentation and provides consequences on the segmentation process towards a solution.

A review of medical imaging informatics

This review of medical imaging informatics is a survey of current developments in an exciting field. The focus is on informatics issues rather than traditional data processing and information systems, such as picture archiving and communications systems (PACS) and image processing and analysis systems. In this review, we address imaging informatics issues within the requirements of an informatics system defined by the American Medical Informatics Association. With these requirements as a framework, we review, in four sections: (1) Methods to present imaging and associated data without causing an overload, including image study summarization, content-based medical image retrieval, and natural language processing of text data. (2) Data modeling techniques to represent clinical data with focus on an image data model, including general-purpose time-based multimedia data models, health-care-specific data models, knowledge models, and problem-centric data models. (3) Methods to integrate medical data information from heterogeneous clinical data sources. Advances in centralized databases and mediated architectures are reviewed along with a discussion on our efforts at data integration based on peer-to-peer networking and shared file systems. (4) Visualization schemas to present imaging and clinical data: the large volume of medical data presents a daunting challenge for an efficient visualization paradigm. In this section we review current multimedia visualization methods including temporal modeling, problem-specific data organization, including our problem-centric, context and user-specific visualization interface.

A robust method for extraction and automatic segmentation of brain images

A new protocol is introduced for brain extraction and automatic tissue segmentation of MR images. For the brain extraction algorithm, proton density and T2-weighted images are used to generate a brain mask encompassing the full intracranial cavity. Segmentation of brain tissues into gray matter (GM), white matter (WM), and cerebral spinal fluid (CSF) is accomplished on a T1-weighted image after applying the brain mask. The fully automatic segmentation algorithm is histogram-based and uses the Expectation Maximization algorithm to model a four-Gaussian mixture for both global and local histograms. The means of the local Gaussians for GM, WM, and CSF are used to set local thresholds for tissue classification. Reproducibility of the extraction procedure was excellent, with average variation in intracranial capacity (TIC) of 0.13 and 0.66% TIC in 12 healthy normal and 33 Alzheimer brains, respectively. Repeatability of the segmentation algorithm, tested on healthy normal images, indicated scan-rescan differences in global tissue volumes of less than 0.30% TIC. Reproducibility at the regional level was established by comparing segmentation results within the 12 major Talairach subdivisions. Accuracy of the algorithm was tested on a digital brain phantom, and errors were less than 1% of the phantom volume. Maximal Type I and Type II classification errors were low, ranging between 2.2 and 4.3% of phantom volume. The algorithm was also insensitive to variation in parameter initialization values. The protocol is robust, fast, and its success in segmenting normal as well as diseased brains makes it an attractive clinical application.

On the metrics and euler-lagrange equations of computational anatomy

This paper reviews literature, current concepts and approaches in computational anatomy (CA). The model of CA is a Grenander deformable template, an orbit generated from a template under groups of diffeomorphisms. The metric space of all anatomical images is constructed from the geodesic connecting one anatomical structure to another in the orbit. The variational problems specifying these metrics are reviewed along with their associated Euler-Lagrange equations. The Euler equations of motion derived by Arnold for the geodesics in the group of divergence-free volume-preserving diffeomorphisms of incompressible fluids are generalized for the larger group of diffeomorphisms used in CA with nonconstant Jacobians. Metrics that accommodate photometric variation are described extending the anatomical model to incorporate the construction of neoplasm. Metrics on landmarked shapes are reviewed as well as Joshi's diffeomorphism metrics, Bookstein's thin-plate spline approximate-metrics, and Kendall's affine invariant metrics. We conclude by showing recent experimental results from the Toga & Thompson group in growth, the Van Essen group in macaque and human cortex mapping, and the Csernansky group in hippocampus mapping for neuropsychiatric studies in aging and schizophrenia.

Estimating cerebral atrophy in multiple sclerosis patients from various MR pulse sequences

PURPOSE

The purpose of this study was to determine how measures reflecting cerebral atrophy (CA) are influenced by pulse sequence (PS) and segmentation algorithm (SA) used in multiple sclerosis (MS) patients and healthy control (HC)s.

METHODS

Magnetic resonance imaging (MRI) scans from 10 relapsing-remitting MS (RRMS) patients and five HCs were used to determine the change in brain fractional volume (BFV) over a two-year period. T1-weighted, fluid-attenuated inversion recovery (FLAIR), and proton density (PD)/T2-weighted sequences were analysed Image segmentation to determine brain volume was performed using the following a histogram SA, an adaptive fuzzy c-means algorithm (AFCM), and an adaptive Bayesian segmentation with a K-means clustering.

RESULTS

Combinations of the SA and PS in MS patents demonstrated significant differences in the per cent change in BFV from baseline. For the combination of PS and SA the per cent change in BFV for year one and year two varied from +2.05% to - 1.6% and +0.79% to -3.11%, respectively. Analysis of the HCs data revealed fluctuations in BFV varying from +0.26% to -0.29%.

CONCLUSIONS

MRI estimates of CA are dependent on both the type of PS and SA; therefore, the choice of SA technique and PS should be consistent during an MS treatment trial. The progression of CA in MS should only be used as a secondary or tertiary outcome measure in treatment trials until a better understanding of how this measurement is affected by the disease, the image acquisition and analysis techniques.

Rapid method for correction of CSF partial volume in quantitative proton MR spectroscopic imaging

Partial volume effects with cerebrospinal fluid (CSF), if uncorrected, can lead to underestimation of metabolite concentrations in quantitative proton magnetic resonance spectroscopic imaging (MRSI) of the brain. A rapid method for the correction of CSF partial volume effects is described based on selective CSF imaging using long echo time (TE) fast spin echo (FSE) magnetic resonance imaging (MRI). In order to achieve maximum suppression of signal from brain parenchyma, the FSE sequence is coupled with an inversion recovery (IR) pulse. Scan time is minimized using single shot (SS) IR-FSE. The method is validated against a current "gold standard" for the determination of CSF volumes, namely, segmented 3D spoiled gradient-echo (SPGR) scans. Excellent agreement in CSF percentage determined by the two methods was found (linear regression analysis: slope = 0.99 +/- 0.02, intercept = 2.08 +/- 0.45; mean +/- standard errors, R = 0.93) in pooled data from four healthy subjects. An example of the use of SS-IR-FSE for partial volume correction in a leukodystrophy patient with T(2) hyperintense lesions is demonstrated. SS-IR-FSE is a simple and rapid method for applying partial volume corrections in quantitative proton MRSI, which may be of particular value in the clinical environment when time constraints do not allow longer, perhaps more accurate segmentation methods to be used.

Simultaneous quantification of cell motility and protein-membrane-association using active contours

We present a new method for the quantification of dynamic changes in fluorescence intensities at the cell membrane of moving cells. It is based on an active contour method for cell-edge detection, which allows tracking of changes in cell shape and position. Fluorescence intensities at specific cortical subregions can be followed in space and time and correlated with cell motility. The translocation of two GFP tagged proteins (CRAC and GRP1) from the cytosol to the membrane in response to stimulation with the chemoattractant cAMP during chemotaxis of Dictyostelium cells and studies of the spatio-temporal dynamics of this process exemplify the method: We show that the translocation can be correlated with motility parameters and that quantitative differences in the rate of association and dissociation from the membrane can be observed for the two PH domain containing proteins. The analysis of periodic CRAC translocation to the leading edge of a cell responding to natural cAMP waves in a mound demonstrates the power of this approach. It is not only capable of tracking the outline of cells within aggregates in front of a noisy background, but furthermore allows the construction of spatio-temporal polar plots, capturing the dynamics of the protein distribution at the cell membrane within the cells' moving co-ordinate system. Compilation of data by means of normalised polar plots is suggested as a future tool, which promises the so-far impossible practicability of extensive statistical studies and automated comparison of complex spatio-temporal protein distribution patterns.

A novel three-dimensional variant of the watershed transform for segmentation of electron density maps

Electron density maps at moderate resolution are often difficult to interpret due to the lack of recognizable features. This is especially true for electron tomograms that suffer in addition to the resolution limitation from low signal-to-noise ratios. Reliable segmentation of such maps into smaller, manageable units can greatly facilitate interpretation. Here, we present a segmentation approach targeting three-dimensional electron density maps derived by electron microscopy. The approach consists of a novel three-dimensional variant of the immersion-based watershed algorithm. We tested the algorithm on calculated data and applied it to a wide variety of electron density maps ranging from reconstructions of single macromolecules to tomograms of subcellular structures. The results indicate that the algorithm is reliable, efficient, accurate, and applicable to a wide variety of biological problems.

Imaging three-dimensional cardiac function

The three-dimensional (3-D) nature of myocardial deformations is dependent on ventricular geometry, muscle fiber architecture, wall stresses, and myocardial-material properties. The imaging modalities of X-ray angiography, echocardiography, computed tomography, and magnetic resonance (MR) imaging (MRI) are described in the context of visualizing and quantifying cardiac mechanical function. The quantification of ventricular anatomy and cavity volumes is then reviewed, and surface reconstructions in three dimensions are demonstrated. The imaging of myocardial wall motion is discussed, with an emphasis on current MRI and tissue Doppler imaging techniques and their potential clinical applications. Calculation of 3-D regional strains from motion maps is reviewed and illustrated with clinical MRI tagging results. We conclude by presenting a promising technique to assess myocardial-fiber architecture, and we outline its potential applications, in conjunction with quantification of anatomy and regional strains, for the determination of myocardial stress and work distributions. The quantification of multiple components of 3-D cardiac function has potential for both fundamental-science and clinical applications.