Neural-network-based segmentation of multi-modal medical images: a comparative and prospective study

Application of a Modified Combinational Approach to Brain Tumor Detection in MR Images

For many years, brain tumor detection has been one of the most essential and competitive issues for medical researchers. Many methods have been developed to detect normal and abnormal tissues in Magnetic Resonance (MR) images. In this work, we present a novel algorithm based on iterative Co-Clustering and K-Means (ICCK). After image pre-processing and enhancement, this algorithm recognizes the part of the image that contains the tumor and eliminates the unused parts using a modification of the Co-Clustering method. Finally, the K-Means clustering method is adopted to detect the tumor area. The Co-Clustering methods cannot be used directly for the detection of brain tumors because they manipulate the image matrix for the purpose of block clustering. Furthermore, they are incapable of detecting the tumor area correctly and accurately. Such issues are addressed by our proposed methodology. The latent block model (LBM) is applied as the Co-Clustering method in this work. We evaluate the performance of our method on the images that were collected from the BraTS2019 dataset. The sensitivity, specificity, accuracy, and dice similarity coefficient values for our method are 82.41%, 99.74%, 99.28%, and 84.87%, respectively, which shows that the proposed method outperforms the existing methods in the literature. Moreover, it performs much better on complex images.

Brain tissues have single-voxel signatures in multi-spectral MRI

Since the seminal works by Brodmann and contemporaries, it is well-known that different brain regions exhibit unique cytoarchitectonic and myeloarchitectonic features. Transferring the approach of classifying brain tissues - and other tissues - based on their intrinsic features to the realm of magnetic resonance (MR) is a longstanding endeavor. In the 1990s, atlas-based segmentation replaced earlier multi-spectral classification approaches because of the large overlap between the class distributions. Here, we explored the feasibility of performing global brain classification based on intrinsic MR features, and used several technological advances: ultra-high field MRI, q-space trajectory diffusion imaging revealing voxel-intrinsic diffusion properties, chemical exchange saturation transfer and semi-solid magnetization transfer imaging as a marker of myelination and neurochemistry, and current neural network architectures to analyze the data. In particular, we used the raw image data as well to increase the number of input features. We found that a global brain classification of roughly 97 brain regions was feasible with gross classification accuracy of 60%; and that mapping from voxel-intrinsic MR data to the brain region to which the data belongs is possible. This indicates the presence of unique MR signals of different brain regions, similar to their cytoarchitectonic and myeloarchitectonic fingerprints.

3D Visualization of Brain Tumors Using MR Images: A Survey

BACKGROUND

Three-Dimensional visualization of brain tumors is very useful in both diagnosis and treatment stages of brain cancer.

DISCUSSION

It helps the oncologist/neurosurgeon to take the best decision in Radiotherapy and/or surgical resection techniques. 3D visualization involves two main steps; tumor segmentation and 3D modeling.

CONCLUSION

In this article, we illustrate the most widely used segmentation and 3D modeling techniques for brain tumors visualization. We also survey the public databases available for evaluation of the mentioned techniques.

Brain abnormality detection using template matching

Magnetic resonance imaging (MRI) is a widely used imaging modality to evaluate brain disorders. MRI generates huge volumes of data, which consist of a sequence of scans taken at different instances of time. As the presence of brain disorders has to be evaluated on all magnetic resonance (MR) sequences, manual brain disorder detection becomes a tedious process and is prone to inter- and intra-rater errors. A technique for detecting abnormalities in brain MRI using template matching is proposed. Bias filed correction is performed on volumetric scans using N4ITK filter, followed by volumetric registration. Normalized cross-correlation template matching is used for image registration taking into account, the rotation and scaling operations. A template of abnormality is selected which is then matched in the volumetric scans, if found, the corresponding image is retrieved. Post-processing of the retrieved images is performed by the thresholding operation; the coordinates and area of the abnormality are reported. The experiments are carried out on the glioma dataset obtained from Brain Tumor Segmentation Challenge 2013 database (BRATS 2013). Glioma dataset consisted of MR scans of 30 real glioma patients and 50 simulated glioma patients. NVIDIA Compute Unified Device Architecture framework is employed in this paper, and it is found that the detection speed using graphics processing unit is almost four times faster than using only central processing unit. The average Dice and Jaccard coefficients for a wide range of trials are found to be 0.91 and 0.83, respectively.

Fully Automatic Brain Tumor Segmentation using End-To-End Incremental Deep Neural Networks in MRI images

BACKGROUND AND OBJECTIVE

Nowadays, getting an efficient Brain Tumor Segmentation in Multi-Sequence MR images as soon as possible, gives an early clinical diagnosis, treatment and follow-up. The aim of this study is to develop a new deep learning model for the segmentation of brain tumors. The proposed models are used to segment the brain tumors of Glioblastomas (with both high and low grade). Glioblastomas have four properties: different sizes, shapes, contrasts, in addition, Glioblastomas appear anywhere in the brain.

METHODS

In this paper, we propose three end-to-end Incremental Deep Convolutional Neural Networks models for fully automatic Brain Tumor Segmentation. Our proposed models are different from the other CNNs-based models that follow the technique of trial and error process which does not use any guided approach to get the suitable hyper-parameters. Moreover, we adopt the technique of Ensemble Learning to design a more efficient model. For solving the problem of training CNNs model, we propose a new training strategy which takes into account the most influencing hyper-parameters by bounding and setting a roof to these hyper-parameters to accelerate the training.

RESULTS

Our experiment results reported on BRATS-2017 dataset. The proposed deep learning models achieve the state-of-the-art performance without any post-processing operations. Indeed, our models achieve in average 0.88 Dice score over the complete region. Moreover, the efficient design with the advantage of GPU implementation, allows our three deep learning models to achieve brain segmentation results in average 20.87 s.

CONCLUSIONS

The proposed deep learning models are effective for the segmentation of brain tumors and allow to obtain high accurate results. Moreover, the proposed models could help the physician experts to reduce the time of diagnostic.

A simple and fast adaptive nonlocal multispectral filtering algorithm for efficient noise reduction in magnetic resonance imaging

PURPOSE

We recently introduced a multispectral (MS) nonlocal (NL) filter based on maximum likelihood estimation (MLE) of voxel intensities, termed MS-NLML. While MS-NLML provides excellent noise reduction and improved image feature preservation as compared to other NL or MS filters, it requires considerable processing time, limiting its application in routine analyses. In this work, we introduced a fast, simple, and robust filter, termed nonlocal estimation of multispectral magnitudes (NESMA), for noise reduction in multispectral (MS) magnetic resonance imaging (MRI).

METHODS

Through extensive simulation and in-vivo analyses, we compared the performance of NESMA and MS-NLML in terms of noise reduction and processing efficiency. Further, we introduce two simple adaptive methods that permit spatial variation of similar voxels, R, used in the filtering. The first method is semi-adaptive and permits variation of R across the image by using a relative Euclidean distance (RED) similarity threshold. The second method is fully adaptive and filters the raw data with several RED similarity thresholds to spatially determine the optimal threshold value using an unbiased criterion.

RESULTS

NESMA shows very similar filtering performance as compared to MS-NLML, however, with much simple implementation and very fast processing time. Further, for both filters, the adaptive methods were shown to further reduce noise in comparison with the conventional non-adaptive method in which R is set to a constant value throughout the image.

CONCLUSIONS

NESMA is fast, robust, and straightforward to implement filter. These features render it suitable for routine clinical use and analysis of large MRI datasets.

Intelligent Computing in Medical Imaging

Biomedical imaging is considered main procedure to acquire valuable physical information about the human body and some other biological species. It produces specialized images of different parts of the biological species for clinical analysis. It assimilates various specialized domains including nuclear medicine, radiological imaging, Positron emission tomography (PET), and microscopy. From the early discovery of X-rays, progress in biomedical imaging continued resulting in highly sophisticated medical imaging modalities, such as magnetic resonance imaging (MRI), ultrasound, Computed Tomography (CT), and lungs monitoring. These biomedical imaging techniques assist physicians for faster and accurate analysis and treatment. The present chapter discussed the impact of intelligent computing methods for biomedical image analysis and healthcare. Different Artificial Intelligence (AI) based automated biomedical image analysis are considered. Different approaches are discussed including the AI ability to resolve various medical imaging problems. It also introduced the popular AI procedures that employed to solve some special problems in medicine. Artificial Neural Network (ANN) and support vector machine (SVM) are active to classify different types of images from various imaging modalities. Different diagnostic analysis, such as mammogram analysis, MRI brain image analysis, CT images, PET images, and bone/retinal analysis using ANN, feed-forward back propagation ANN, probabilistic ANN, and extreme learning machine continuously. Various optimization techniques of ant colony optimization (ACO), genetic algorithm (GA), particle swarm optimization (PSO) and other bio-inspired procedures are also frequently conducted for feature extraction/selection and classification. The advantages and disadvantages of some AI approaches are discussed in the present chapter along with some suggested future research perspectives.

A deep convolutional neural network-based automatic delineation strategy for multiple brain metastases stereotactic radiosurgery

Accurate and automatic brain metastases target delineation is a key step for efficient and effective stereotactic radiosurgery (SRS) treatment planning. In this work, we developed a deep learning convolutional neural network (CNN) algorithm for segmenting brain metastases on contrast-enhanced T1-weighted magnetic resonance imaging (MRI) datasets. We integrated the CNN-based algorithm into an automatic brain metastases segmentation workflow and validated on both Multimodal Brain Tumor Image Segmentation challenge (BRATS) data and clinical patients' data. Validation on BRATS data yielded average DICE coefficients (DCs) of 0.75±0.07 in the tumor core and 0.81±0.04 in the enhancing tumor, which outperformed most techniques in the 2015 BRATS challenge. Segmentation results of patient cases showed an average of DCs 0.67±0.03 and achieved an area under the receiver operating characteristic curve of 0.98±0.01. The developed automatic segmentation strategy surpasses current benchmark levels and offers a promising tool for SRS treatment planning for multiple brain metastases.

State of the art survey on MRI brain tumor segmentation

Brain tumor segmentation consists of separating the different tumor tissues (solid or active tumor, edema, and necrosis) from normal brain tissues: gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). In brain tumor studies, the existence of abnormal tissues may be easily detectable most of the time. However, accurate and reproducible segmentation and characterization of abnormalities are not straightforward. In the past, many researchers in the field of medical imaging and soft computing have made significant survey in the field of brain tumor segmentation. Both semiautomatic and fully automatic methods have been proposed. Clinical acceptance of segmentation techniques has depended on the simplicity of the segmentation, and the degree of user supervision. Interactive or semiautomatic methods are likely to remain dominant in practice for some time, especially in these applications where erroneous interpretations are unacceptable. This article presents an overview of the most relevant brain tumor segmentation methods, conducted after the acquisition of the image. Given the advantages of magnetic resonance imaging over other diagnostic imaging, this survey is focused on MRI brain tumor segmentation. Semiautomatic and fully automatic techniques are emphasized.

Swallow segmentation with artificial neural networks and multi-sensor fusion

Swallow segmentation is a critical precursory step to the analysis of swallowing signal characteristics. In an effort to automatically segment swallows, we investigated artificial neural networks (ANN) with information from cervical dual-axis accelerometry, submental MMG, and nasal airflow. Our objectives were (1) to investigate the relationship between segmentation performance and the number of signal sources and (2) to identify the signals or signal combinations most useful for swallow segmentation. Signals were acquired from 17 healthy adults in both discrete and continuous swallowing tasks using five stimuli. Training and test feature vectors were constructed with variances from single or multiple signals, estimated within 200 ms moving windows with 50% overlap. Corresponding binary target labels (swallow or non-swallow) were derived by manual segmentation. A separate 3-layer ANN was trained for each participant-signal combination, and all possible signal combinations were investigated. As more signal sources were included, segmentation performance improved in terms of sensitivity, specificity, accuracy, and adjusted accuracy. The combination of all four signal sources achieved the highest mean accuracy and adjusted accuracy of 88.5% and 89.6%, respectively. A-P accelerometry proved to be the most discriminatory source, while the inclusion of MMG or nasal airflow resulted in the least performance improvement. These findings suggest that an ANN, multi-sensor fusion approach to segmentation is worthy of further investigation in swallowing studies.

Segmentation of age-related white matter changes in a clinical multi-center study

Age-related white matter changes (WMC) are thought to be a marker of vascular pathology, and have been associated with motor and cognitive deficits. In the present study, an optimized artificial neural network was used as an automatic segmentation method to produce probabilistic maps of WMC in a clinical multi-center study. The neural network uses information from T1- and T2-weighted and fluid attenuation inversion recovery (FLAIR) magnetic resonance (MR) scans, neighboring voxels and spatial location. Generalizability of the neural network was optimized by including the Optimal Brain Damage (OBD) pruning method in the training stage. Six optimized neural networks were produced to investigate the impact of different input information on WMC segmentation. The automatic segmentation method was applied to MR scans of 362 non-demented elderly subjects from 11 centers in the European multi-center study Leukoaraiosis And Disability (LADIS). Semi-manually delineated WMC were used for validating the segmentation produced by the neural networks. The neural network segmentation demonstrated high consistency between subjects and centers, making it a promising technique for large studies. For WMC volumes less than 10 ml, an increasing discrepancy between semi-manual and neural network segmentation was observed using the similarity index (SI) measure. The use of all three image modalities significantly improved cross-center generalizability compared to neural networks using the FLAIR image only. Expert knowledge not available to the neural networks was a minor source of discrepancy, while variation in MR scan quality constituted the largest source of error.

Elimination of RF inhomogeneity effects in segmentation

There are various methods proposed for the segmentation and analysis of MR images. However the efficiency of these techniques is effected by various artifacts that occur in the imaging system. One of the most encountered problems is the intensity variation across an image. To overcome this problem different methods are used. In this paper we propose a method for the elimination of intensity artifacts in segmentation of MRI images. Inter imager variations are also minimized to produce the same tissue segmentation for the same patient. A well-known multivariate classification algorithm, maximum likelihood is employed to illustrate the enhancement in segmentation.

Automated segmentation and classification of high throughput yeast assay spots

Several technologies for characterizing genes and proteins from humans and other organisms use yeast growth or color development as read outs. The yeast two-hybrid assay, for example, detects protein-protein interactions by measuring the growth of yeast on a specific solid medium, or the ability of the yeast to change color when grown on a medium containing a chromogenic substrate. Current systems for analyzing the results of these types of assays rely on subjective and inefficient scoring of growth or color by human experts. Here, an image analysis system is described for scoring yeast growth and color development in high throughput biological assays. The goal is to locate the spots and score them in color images of two types of plates named "X-Gal" and "growth assay" plates, with uniformly placed spots (cell areas) on each plate (both plates in one image). The scoring system relies on color for the X-Gal spots, and texture properties for the growth assay spots. A maximum likelihood projection-based segmentation is developed to automatically locate spots of yeast on each plate. Then color histogram and wavelet texture features are extracted for scoring using an optimal linear transformation. Finally, an artificial neural network is used to score the X-Gal and growth assay spots using the extracted features. The performance of the system is evaluated using spots of 60 images. After training the networks using training and validation sets, the system was assessed on the test set. The overall accuracies of 95.4% and 88.2% are achieved, respectively, for scoring the X-Gal and growth assay spots.

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.

Magnetic resonance imaging segmentation techniques using batch-type learning vector quantization algorithms

In this article, we propose batch-type learning vector quantization (LVQ) segmentation techniques for the magnetic resonance (MR) images. Magnetic resonance imaging (MRI) segmentation is an important technique to differentiate abnormal and normal tissues in MR image data. The proposed LVQ segmentation techniques are compared with the generalized Kohonen's competitive learning (GKCL) methods, which were proposed by Lin et al. [Magn Reson Imaging 21 (2003) 863-870]. Three MRI data sets of real cases are used in this article. The first case is from a 2-year-old girl who was diagnosed with retinoblastoma in her left eye. The second case is from a 55-year-old woman who developed complete left side oculomotor palsy immediately after a motor vehicle accident. The third case is from an 84-year-old man who was diagnosed with Alzheimer disease (AD). Our comparisons are based on sensitivity of algorithm parameters, the quality of MRI segmentation with the contrast-to-noise ratio and the accuracy of the region of interest tissue. Overall, the segmentation results from batch-type LVQ algorithms present good accuracy and quality of the segmentation images, and also flexibility of algorithm parameters in all the comparison consequences. The results support that the proposed batch-type LVQ algorithms are better than the previous GKCL algorithms. Specifically, the proposed fuzzy-soft LVQ algorithm works well in segmenting AD MRI data set to accurately measure the hippocampus volume in AD MR images.

Segmentation of magnetic resonance images using a combination of neural networks and active contour models

Segmentation of medical images is very important for clinical research and diagnosis, leading to a requirement for robust automatic methods. This paper reports on the combined use of a neural network (a multilayer perceptron, MLP) and active contour model ('snake') to segment structures in magnetic resonance (MR) images. The perceptron is trained to produce a binary classification of each pixel as either a boundary or a non-boundary point. Subsequently, the resulting binary (edge-point) image forms the external energy function for a snake, used to link the candidate boundary points into a continuous, closed contour. We report here on the segmentation of the lungs from multiple MR slices of the torso; lung-specific constraints have been avoided to keep the technique as general as possible. In initial investigations, the inputs to the MLP were limited to normalised intensity values of the pixels from an (7 x 7) window scanned across the image. The use of spatial coordinates as additional inputs to the MLP is then shown to provide an improvement in segmentation performance as quantified using the effectiveness measure (a weighted product of precision and recall). Training sets were first developed using a lengthy iterative process. Thereafter, a novel cost function based on effectiveness is proposed for training that allows us to achieve dramatic improvements in segmentation performance, as well as faster, non-iterative selection of training examples. The classifications produced using this cost function were sufficiently good that the binary image produced by the MLP could be post-processed using an active contour model to provide an accurate segmentation of the lungs from the multiple slices in almost all cases, including unseen slices and subjects.

Coupling of radial-basis network and active contour model for multispectral brain MRI segmentation

Magnetic resonance (MR) has been accepted as the reference image study in the clinical environment. The development of new sequences has allowed obtaining diverse images with high clinical importance and whose interpretation requires their joint analysis (multispectral MRI). Recent approaches to segment MRI point toward the definition of hybrid models, where the advantages of region and contour-based methods can be exploited to look for the integration or fusion of information, thus enhancing the performance of the individual approaches. Following this perspective, a hybrid model for multispectral brain MRI segmentation is presented. The model couples a segmenter, based on a radial basis network (RBFNNcc), and an active contour model, based on a cubic spline active contour (CSAC) interpolation. Both static and dynamic coupling of RBFNNcc and CSAC are proposed; the RBFNNcc stage provides an initial contour to the CSAC; the initial contour is optimally sampled with respect to its curvature variations; multispectral information and a restriction term are included into the CSAC energy equation. Segmentations were compared to a reference stack, indicating high-quality performance as measured by Tanimoto indexes of 0.74 +/- 0.08.

Generalized Kohonen’s competitive learning algorithms for ophthalmological MR image segmentation

Kohonen's self-organizing map is a two-layer feedforward competitive learning network. It has been used as a competitive learning clustering algorithm. In this paper, we generalize Kohonen's competitive learning (KCL) algorithm with fuzzy and fuzzy-soft types called fuzzy KCL (FKCL) and fuzzy-soft KCL (FSKCL). These generalized KCL algorithms fuse the competitive learning with soft competition and fuzzy c-means (FCM) membership functions. We then apply these generalized KCLs to MRI and MRA ophthalmological segmentations. These KCL-based MRI segmentation techniques are useful in reducing medical image noise effects using a learning mechanism. They may be particularly helpful in clinical diagnosis. Two real cases with MR image data recommended by an ophthalmologist are examined. First case is a patient with Retinoblastoma in her left eye, an inborn malignant neoplasm of the retina frequently metastasis beyond the lacrimal cribrosa. The second case is a patient with complete left side oculomotor palsy immediately after a motor vehicle accident. Her brain MRI with MRA, skull routine, orbital CT, and cerebral angiography did not reveal brainstem lesions, skull fractures, or vascular anomalies. These generalized KCL algorithms were used in segmenting the ophthalmological MRIs. KCL, FKCL and FSKCL comparisons are made. Overall, the FSKCL algorithm is recommended for use in MR image segmentation as an aid to small lesion diagnosis.

Statistical mapping analysis of lesion location and neurological disability in multiple sclerosis: application to 452 patient data sets

In multiple sclerosis (MS), the correlation between disability and the volume of white matter lesions on magnetic resonance imaging (MRI) is usually weak. This may be because lesion location also influences the extent and type of functional disability. We applied an automatic lesion-detection algorithm to 452 MRI scans of patients with relapsing-remitting MS to identify the regions preferentially responsible for different types of clinical deficits. Statistical parametric maps were generated by performing voxel-wise linear regressions between lesion probability and different clinical disability scores. There was a clear distinction between lesion locations causing physical and cognitive disability. Lesion likelihood correlated with the Expanded Disability Status Scale (EDSS) in the left internal capsule and in periventricular white matter mostly in the left hemisphere. Pyramidal deficits correlated with only one area in the left internal capsule that was also present in the EDSS correlation. Cognitive dysfunction correlated with lesion location at the grey-white junction of associative, limbic, and prefrontal cortex. Coordination impairment correlated with areas in interhemispheric and pyramidal periventricular white matter tracts, and in the inferior and superior longitudinal fascicles. Bowel and bladder scores correlated with lesions in the medial frontal lobes, cerebellum, insula, dorsal midbrain, and pons, areas known to be involved in the control of micturition. This study demonstrates for the first time a relationship between the site of lesions and the type of disability in large scale MRI data set in MS.

Deformation-based surface morphometry applied to gray matter deformation

We present a unified statistical approach to deformation-based morphometry applied to the cortical surface. The cerebral cortex has the topology of a 2D highly convoluted sheet. As the brain develops over time, the cortical surface area, thickness, curvature, and total gray matter volume change. It is highly likely that such age-related surface changes are not uniform. By measuring how such surface metrics change over time, the regions of the most rapid structural changes can be localized. We avoided using surface flattening, which distorts the inherent geometry of the cortex in our analysis and it is only used in visualization. To increase the signal to noise ratio, diffusion smoothing, which generalizes Gaussian kernel smoothing to an arbitrary curved cortical surface, has been developed and applied to surface data. Afterward, statistical inference on the cortical surface will be performed via random fields theory. As an illustration, we demonstrate how this new surface-based morphometry can be applied in localizing the cortical regions of the gray matter tissue growth and loss in the brain images longitudinally collected in the group of children and adolescents.

Survey on the use of smart and adaptive engineering systems in medicine

In this paper, the current published knowledge about smart and adaptive engineering systems in medicine is reviewed. The achievements of frontier research in this particular field within medical engineering are described. A multi-disciplinary approach to the applications of adaptive systems is observed from the literature surveyed. The three modalities of diagnosis, imaging and therapy are considered to be an appropriate classification method for the analysis of smart systems being applied to specified medical sub-disciplines. It is expected that future research in biomedicine should identify subject areas where more advanced intelligent systems could be applied than is currently evident. The literature provides evidence of hybridisation of different types of adaptive and smart systems with applications in different areas of medical specifications.

Automatic "pipeline" analysis of 3-D MRI data for clinical trials: application to multiple sclerosis

The quantitative analysis of magnetic resonance imaging (MRI) data has become increasingly important in both research and clinical studies aiming at human brain development, function, and pathology. Inevitably, the role of quantitative image analysis in the evaluation of drug therapy will increase, driven in part by requirements imposed by regulatory agencies. However, the prohibitive length of time involved and the significant intraand inter-rater variability of the measurements obtained from manual analysis of large MRI databases represent major obstacles to the wider application of quantitative MRI analysis. We have developed a fully automatic "pipeline" image analysis framework and have successfully applied it to a number of large-scale, multicenter studies (more than 1,000 MRI scans). This pipeline system is based on robust image processing algorithms, executed in a parallel, distributed fashion. This paper describes the application of this system to the automatic quantification of multiple sclerosis lesion load in MRI, in the context of a phase III clinical trial. The pipeline results were evaluated through an extensive validation study, revealing that the obtained lesion measurements are statistically indistinguishable from those obtained by trained human observers. Given that intra- and inter-rater measurement variability is eliminated by automatic analysis, this system enhances the ability to detect small treatment effects not readily detectable through conventional analysis techniques. While useful for clinical trial analysis in multiple sclerosis, this system holds widespread potential for applications in other neurological disorders, as well as for the study of neurobiology in general.

Improved high resolution MR imaging for surface coils using automated intensity non-uniformity correction: feasibility study in the orbit

This study examined the effects of a recently developed automated intensity non-uniformity correction on surface coil images using the orbit as an exemplar. Images were obtained using a standard head coil and a range of surface coils. Slices through the optic nerve head and cavernous sinus were subjected to the correction algorithm. Blind forced-choice rankings of the subjective image quality were performed. Quantitative measurements were taken of the similarity between vitreous humor at two depths from the coil, and of the conspicuity between orbital fat and temporalis muscle intensities. The combined qualitative ranks for corrected surface coil images were higher than for the equivalent uncorrected images in all cases. Intensity non-uniformity correction produced statistically significant improvements in orbital surface coil images, bringing their intensity uniformity in homogeneous tissue to the level of head coil images. The subjective quality of the corrected surface coil images was superior to head coil images, due to increased spatial resolution combined with improved signal to noise ratio across the image.

Novel approaches to imaging brain tumors

Brain imaging techniques are assuming a greater range of roles in neuro-oncology. New techniques promise earlier recognition of the spread of tumors to the brain, which is useful in staging of disseminated disease, as well as better definition of small lesions associated with presentations of epilepsy. There is the promise that entirely noninvasive, specific diagnosis of brain tumors may become possible. Imaging methods are being used increasingly to direct and monitor therapy. Preoperative and intraoperative imaging are being used for guiding tumor surgery. An exciting potential goal for greater use of imaging is in the individualization of medical therapies either by analysis of in vitro responses or by visualization of drug responses on the tumor in situ. An important focus for technical development is in the robust integration of complementary information to allow optimization of the sensitivity and specificity of multimodal examinations.

Sexual Dimorphism in the Corpus Callosum: Methodological Considerations in MRI Morphometry

Studies of sexual dimorphism in the corpus callosum (CC) have employed a variety of methodologies for measurement and normalization but have yielded disparate results. The present work demonstrates how in some cases different manipulations of the same raw data, corresponding to different commonly used methodologies, produce discordant results. Midsagittal CC area was measured from magnetic resonance images (MRIs) of 137 young normal volunteers. Three strategies intended to normalize for average differences in brain size between the sexes, as well as five different normalization variables, were contrasted and evaluated. The stereotaxic method normalizes for intersubject differences in overall brain size by scaling MRIs into a standardized space. The ratio method uses one of five different indices of brain size and divides it into CC area. The covariate method uses one of the indices as a covariate in statistical analyses. Male subjects show significantly larger absolute total area, as well as anterior third and posterior midbody. However, in two of three normalization strategies, namely the stereotaxic and ratio methods, females show relatively larger total area, anterior midbody, and splenium. The covariate method did not show any significant differences at the 0.05 level. Results suggest that different approaches to normalization and analysis are not necessarily equivalent and interchangeable.

Magnetic resonance image tissue classification using a partial volume model

We describe a sequence of low-level operations to isolate and classify brain tissue within T1-weighted magnetic resonance images (MRI). Our method first removes nonbrain tissue using a combination of anisotropic diffusion filtering, edge detection, and mathematical morphology. We compensate for image nonuniformities due to magnetic field inhomogeneities by fitting a tricubic B-spline gain field to local estimates of the image nonuniformity spaced throughout the MRI volume. The local estimates are computed by fitting a partial volume tissue measurement model to histograms of neighborhoods about each estimate point. The measurement model uses mean tissue intensity and noise variance values computed from the global image and a multiplicative bias parameter that is estimated for each region during the histogram fit. Voxels in the intensity-normalized image are then classified into six tissue types using a maximum a posteriori classifier. This classifier combines the partial volume tissue measurement model with a Gibbs prior that models the spatial properties of the brain. We validate each stage of our algorithm on real and phantom data. Using data from the 20 normal MRI brain data sets of the Internet Brain Segmentation Repository, our method achieved average kappa indices of kappa = 0.746 +/- 0.114 for gray matter (GM) and kappa = 0.798 +/- 0.089 for white matter (WM) compared to expert labeled data. Our method achieved average kappa indices kappa = 0.893 +/- 0.041 for GM and kappa = 0.928 +/- 0.039 for WM compared to the ground truth labeling on 12 volumes from the Montreal Neurological Institute's BrainWeb phantom.

Three dimensional reconstruction of coronary arteries from two views

Geometric representation and measurements of localized lumen stenosis of coronary arteries are important considerations in the diagnosis of cardiovascular diseases. This discrete narrowing of the arteries typically impairs blood flow in regions of the heart, and can be present along the entire length of the artery. Three-dimensional (3-D) reconstruction of coronary arterial tree allows clinician to visualize vascular geometry. Three-dimensional representation of tree topology facilitates calculation of hemodynamic measurements to study myocardial infarction and stenosis. The 3-D arterial tree, computed from two views, can provide more information about the tree geometry than individual views. In this paper, a 3-step algorithm for 3-D reconstruction of arterial tree using two standard views is presented. The first step is a multi-resolution segmentation of the coronary vessels followed by medial-axis detection along the entire arterial tree for both views. In the second step, arterial trees from the two views are registered using medial-axis representation at the coarsest resolution level to obtain an initial 3-D reconstruction. This initial reconstruction at the coarsest level is then modified using 3-D geometrical a priori information. In the third step, the modified reconstruction is projected on the next higher-resolution segmented medial-axis representation and an updated reconstruction is obtained at the higher resolution. The process is iterated until the final 3-D reconstruction is obtained at the finest resolution level. Linear programming based constrained optimization method is used for registering two views at the coarse resolution. This is followed by a Tree-Search method for registering detailed branches at higher resolutions. The automated 3-D reconstruction method was evaluated on computer-simulated as well as human angiogram data. Results show that the automated 3-D reconstruction method provided good registration of computer-simulated data. On human angiogram data, the computed 3-D reconstruction matched well with manual registration.

A fast model independent method for automatic correction of intensity nonuniformity in MRI data

A novel nonparametric approach for correcting intensity nonuniformity in magnetic resonance (MR) images is described. This approach is based solely on the assumption that the various sources of nonuniformity in MR imaging give rise to smooth variations in image intensity, and that these variations can be extracted and corrected for. The advantage of this computationally fast method is that it can be applied early in quantitative analysis while being independent of pulse sequence and is insensitive to pathological processes. This algorithm has been tested on both simulated and real data. Application to tissue segmentation and functional MR imaging has shown a marked improvement in quantitative analysis. J. Magn. Reson. Imaging 1999;10:550-562.

Abdominal organ segmentation using texture transforms and a Hopfield neural network

Abdominal organ segmentation is highly desirable but difficult, due to large differences between patients and to overlapping grey-scale values of the various tissue types. The first step in automating this process is to cluster together the pixels within each organ or tissue type. We propose to form images based on second-order statistical texture transforms (Haralick transforms) of a CT or MRI scan. The original scan plus the suite of texture transforms are then input into a Hopfield neural network (HNN). The network is constructed to solve an optimization problem, where the best solution is the minima of a Lyapunov energy function. On a sample abdominal CT scan, this process successfully clustered 79-100% of the pixels of seven abdominal organs. It is envisioned that this is the first step to automate segmentation. Active contouring (e.g., SNAKE's) or a back-propagation neural network can then be used to assign names to the clusters and fill in the incorrectly clustered pixels.

Structural maturation of neural pathways in children and adolescents: in vivo study

Structural maturation of fiber tracts in the human brain, including an increase in the diameter and myelination of axons, may play a role in cognitive development during childhood and adolescence. A computational analysis of structural magnetic resonance images obtained in 111 children and adolescents revealed age-related increases in white matter density in fiber tracts constituting putative corticospinal and frontotemporal pathways. The maturation of the corticospinal tract was bilateral, whereas that of the frontotemporal pathway was found predominantly in the left (speech-dominant) hemisphere. These findings provide evidence for a gradual maturation, during late childhood and adolescence, of fiber pathways presumably supporting motor and speech functions.

Unsupervised connectivity-based thresholding segmentation of midsagittal brain MR images

In this paper, we propose an algorithm for automated segmentation of midsagittal brain MR images. First, we apply thresholding to obtain binary images. From the binary images, we locate some landmarks. Based on the landmarks and anatomical information, we preprocess the binary images, which substantially simplifies the subsequent operations. To separate regions what are incorrectly merged after this initial segmentation, a new connectivity-based threshold algorithm is proposed. Assuming that some prior information about the general shape and location of objects is available, the algorithm finds a boundary between two regions using the path connection algorithm and changing the threshold adaptively. In order to test the robustness of the proposed algorithm we applied the algorithm to 120 midsagittal brain images and obtained satisfactory results.

A hybrid neural network analysis of subtle brain volume differences in children surviving brain tumors

In the treatment of children with brain tumors, balancing the efficacy of treatment against commonly observed side effects is difficult because of a lack of quantitative measures of brain damage that can be correlated with the intensity of treatment. We quantitatively assessed volumes of brain parenchyma on magnetic resonance (MR) images using a hybrid combination of the Kohonen self-organizing map for segmentation and a multilayer backpropagation neural network for tissue classification. Initially, we analyzed the relationship between volumetric differences and radiologists' grading of atrophy in 80 subjects. This investigation revealed that brain parenchyma and white matter volumes significantly decreased as atrophy increased, whereas gray matter volumes had no relationship with atrophy. Next, we compared 37 medulloblastoma patients treated with surgery, irradiation, and chemotherapy to 19 patients treated with surgery and irradiation alone. This study demonstrated that, in these patients, chemotherapy had no significant effect on brain parenchyma, white matter, or gray matter volumes. We then investigated volumetric differences due to cranial irradiation in 15 medulloblastoma patients treated with surgery and radiation therapy, and compared these with a group of 15 age-matched patients with low-grade astrocytoma treated with surgery alone. With a minimum follow-up of one year after irradiation, all radiation-treated patients demonstrated significantly reduced white matter volumes, whereas gray matter volumes were relatively unchanged compared with those of age-matched patients treated with surgery alone. These results indicate that reductions in cerebral white matter: 1) are correlated significantly with atrophy; 2) are not related to chemotherapy; and 3) are correlated significantly with irradiation. This hybrid neural network analysis of subtle brain volume differences with magnetic resonance may constitute a direct measure of treatment-induced brain damage.

Automatic tumor segmentation using knowledge-based techniques

A system that automatically segments and labels glioblastoma-multiforme tumors in magnetic resonance images (MRI's) of the human brain is presented. The MRI's consist of T1-weighted, proton density, and T2-weighted feature images and are processed by a system which integrates knowledge-based (KB) techniques with multispectral analysis. Initial segmentation is performed by an unsupervised clustering algorithm. The segmented image, along with cluster centers for each class are provided to a rule-based expert system which extracts the intracranial region. Multispectral histogram analysis separates suspected tumor from the rest of the intracranial region, with region analysis used in performing the final tumor labeling. This system has been trained on three volume data sets and tested on thirteen unseen volume data sets acquired from a single MRI system. The KB tumor segmentation was compared with supervised, radiologist-labeled "ground truth" tumor volumes and supervised k-nearest neighbors tumor segmentations. The results of this system generally correspond well to ground truth, both on a per slice basis and more importantly in tracking total tumor volume during treatment over time.

A nonparametric method for automatic correction of intensity nonuniformity in MRI data

A novel approach to correcting for intensity nonuniformity in magnetic resonance (MR) data is described that achieves high performance without requiring a model of the tissue classes present. The method has the advantage that it can be applied at an early stage in an automated data analysis, before a tissue model is available. Described as nonparametric nonuniform intensity normalization (N3), the method is independent of pulse sequence and insensitive to pathological data that might otherwise violate model assumptions. To eliminate the dependence of the field estimate on anatomy, an iterative approach is employed to estimate both the multiplicative bias field and the distribution of the true tissue intensities. The performance of this method is evaluated using both real and simulated MR data.

Statistical approach to segmentation of single-channel cerebral MR images

A statistical model is presented that represents the distributions of major tissue classes in single-channel magnetic resonance (MR) cerebral images. Using the model, cerebral images are segmented into gray matter, white matter, and cerebrospinal fluid (CSF). The model accounts for random noise, magnetic field inhomogeneities, and biological variations of the tissues. Intensity measurements are modeled by a finite Gaussian mixture. Smoothness and piecewise contiguous nature of the tissue regions are modeled by a three-dimensional (3-D) Markov random field (MRF). A segmentation algorithm, based on the statistical model, approximately finds the maximum a posteriori (MAP) estimation of the segmentation and estimates the model parameters from the image data. The proposed scheme for segmentation is based on the iterative conditional modes (ICM) algorithm in which measurement model parameters are estimated using local information at each site, and the prior model parameters are estimated using the segmentation after each cycle of iterations. Application of the algorithm to a sample of clinical MR brain scans, comparisons of the algorithm with other statistical methods, and a validation study with a phantom are presented. The algorithm constitutes a significant step toward a complete data driven unsupervised approach to segmentation of MR images in the presence of the random noise and intensity inhomogeneities.

Using neural networks to automatically detect brain tumours in MR images

Computer vision has been applied to many medical imaging problems with the aim of providing clinical tools to aid medical professionals. We present work being carried out to develop one such system to automatically detect a specific type of brain tumour from head MR images. The tumour under consideration is an acoustic neuroma, which is a benign tumour occurring in the acoustic canals. The hybrid system developed integrates neural networks with more conventional techniques used for computer vision tasks. A database of MR images from 50 patients has been assembled and the acoustic neuromas present in the images have been labelled by hand. Using this data, neural networks (MLPs) have been developed to classify the images at the pixel level to achieve a targeted segmentation. The data used to train and test the MLPs developed, consists of the grey levels of a square of pixels, the pixel to be classified being the centre pixel, together with its global position in the image. The initial pixel level segmentation is refined by a series of conventional techniques. It is combined with an edge-region based segmentation and a morphological operation is applied to the result. This processing produces clusters of adjacent regions, which are considered to be candidate tumour regions. For each possible combination of these regions, features are measured and presented to neural networks which have been trained to identify structures corresponding to acoustic neuromas. Using this approach, all the acoustic neuromas are identified together with three false positive errors.

A dual approach for minimizing false lesion classifications on magnetic resonance images

Segmentation methods based on dual-echo MR images are generally prone to significant false lesion classifications. We have minimized these false classifications by (1) improving the lesion-to-tissue contrast on MR images by developing a fast spin-echo sequence that incorporates both cerebrospinal fluid signal attenuation and magnetization transfer contrast and (2) including information from MR flow images. Studies on patients with multiple sclerosis indicate that this dual approach to tissue segmentation reduces the volume of false lesion classifications by an average of 87%.

Multispectral magnetic resonance images segmentation using fuzzy Hopfield neural network

This paper demonstrates a fuzzy Hopfield neural network for segmenting multispectral MR brain images. The proposed approach is a new unsupervised 2-D Hopfield neural network based upon the fuzzy clustering technique. Its implementation consists of the combination of 2-D Hopfield neural network and fuzzy c-means clustering algorithm in order to make parallel implementation for segmenting multispectral MR brain images feasible. For generating feasible results, a fuzzy c-means clustering strategy is included in the Hopfield neural network to eliminate the need for finding weighting factors in the energy function which is formulated and based on a basic concept commonly used in pattern classification, called the 'within-class scatter matrix' principle. The suggested fuzzy c-means clustering strategy has also been proven to be convergent and to allow the network to learn more effectively than the conventional Hopfield neural network. The experimental results show that a near optimal solution can be obtained using the fuzzy Hopfield neural network based on the within-class scatter matrix.

Segmentation of multivariate medical images via unsupervised clustering with "adaptive resolution"

The need for quantitative information is becoming increasingly important in the clinical field. In this paper we present an interactive X11 based system, devoted to segmentation of multivariate medical images, including an unsupervised neural network approach to clustering. The following steps are considered in the analysis sequence: feature extraction, reduction of dimensionality, unsupervised data clustering, voxel classification, interactive post-processing refinement. The environment turns out to be extremely interactive, thus making the user able to display and modify data during processing, to set parameters, to choose different methods and different tools for each step, and to define online the whole analysis sequence.

Knowledge-based interpretation of MR brain images

The authors have developed a method for fully automated segmentation and labeling of 17 neuroanatomic structures such as thalamus, caudate nucleus, ventricular system, etc. in magnetic resonance (MR) brain images. The authors' method is based on a hypothesize-and-verify principle and uses a genetic algorithm (GA) optimization technique to generate and evaluate image interpretation hypotheses in a feedback loop. The authors' method was trained in 20 individual T1-weighted MR images. Observer-defined contours of neuroanatomic structures were used as a priori knowledge. The method's performance was validated in eight MR images by comparison to observer-defined independent standards. The GA-based image interpretation method correctly interpreted neuroanatomic structures in all images from the test set. Computer-identified and observer-defined neuroanatomic structure areas correlated very well (r=0.99, y=0,95x-2.1). Border positioning errors were small, with a root mean square (rms) border positioning error of 1.5+/-0.6 pixels. The authors' GA-based image interpretation method represents a novel approach to image interpretation and has been shown to produce accurate labeling of neuroanatomic structures in a set of MR brain images.