A nonrigid registration algorithm for longitudinal breast MR images and the analysis of breast tumor response

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) can estimate parameters relating to blood flow and tissue volume fractions and therefore may be used to characterize the response of breast tumors to treatment. To assess treatment response, values of these DCE-MRI parameters are observed at different time points during the course of treatment. We propose a method whereby DCE-MRI data sets obtained in separate imaging sessions can be co-registered to a common image space, thereby retaining spatial information so that serial DCE-MRI parameter maps can be compared on a voxel-by-voxel basis. In performing inter-session breast registration, one must account for patient repositioning and breast deformation, as well as changes in tumor shape and volume relative to other imaging sessions. One challenge is to optimally register the normal tissues while simultaneously preventing tumor distortion. We accomplish this by extending the adaptive bases algorithm through adding a tumor-volume preserving constraint in the cost function. We also propose a novel method to generate the simulated breast magnetic resonance (MR) images, which can be used to evaluate the proposed registration algorithm quantitatively. The proposed nonrigid registration algorithm is applied to both simulated and real longitudinal 3D high resolution MR images and the obtained transformations are then applied to lower resolution physiological parameter maps obtained via DCE-MRI. The registration results demonstrate the proposed algorithm can successfully register breast MR images acquired at different time points and allow for analysis of the registered parameter maps.     

MR image-based measurement of rates of change in volumes of brain structures. Part I: method and validation

A detailed analysis procedure is described for evaluating rates of volumetric change in brain structures based on structural magnetic resonance (MR) images. In this procedure, a series of image processing tools have been employed to address the problems encountered in measuring rates of change based on structural MR images. These tools include an algorithm for intensity non-uniformity correction, a robust algorithm for three-dimensional image registration with sub-voxel precision and an algorithm for brain tissue segmentation. However, a unique feature in the procedure is the use of a fractional volume model that has been developed to provide a quantitative measure for the partial volume effect. With this model, the fractional constituent tissue volumes are evaluated for voxels at the tissue boundary that manifest partial volume effect, thus allowing tissue boundaries be defined at a sub-voxel level and in an automated fashion. Validation studies are presented on key algorithms including segmentation and registration. An overall assessment of the method is provided through the evaluation of the rates of brain atrophy in a group of normal elderly subjects for which the rate of brain atrophy due to normal aging is predictably small. An application of the method is given in Part II where the rates of brain atrophy in various brain regions are studied in relation to normal aging and Alzheimer's disease.     

A nonlinear mesh-warping technique for correcting brain deformation after stroke

This article presents a warping technique for correcting brain tissue distortion on magnetic resonance imaging (MRI) scans due to stroke lesion growth and for mapping MRI scans to histological sections. Meshes are imposed upon the images for feature specification, and these features are exactly matched in the different images to be mapped, while the other voxels are matched by interpolation. This technique was tested on serial MR images and histological sections that were acquired in a nonhuman primate model of stroke. This technique was able to deliver satisfactory warping results. It is simple and robust and can be utilized in many applications for comparison of multimodality medical images and histological sections.     

On-line range images registration with GPGPU

This paper concerns implementation of algorithms in the two important aspects of modern 3D data processing: data registration and segmentation. Solution proposed for the first topic is based on the 3D space decomposition, while the latter on image processing and local neighbourhood search. Data processing is implemented by using NVIDIA compute unified device architecture (NIVIDIA CUDA) parallel computation. The result of the segmentation is a coloured map where different colours correspond to different objects, such as walls, floor and stairs. The research is related to the problem of collecting 3D data with a RGB-D camera mounted on a rotated head, to be used in mobile robot applications. Performance of the data registration algorithm is aimed for on-line processing. The iterative closest point (ICP) approach is chosen as a registration method. Computations are based on the parallel fast nearest neighbour search. This procedure decomposes 3D space into cubic buckets and, therefore, the time of the matching is deterministic. First technique of the data segmentation uses accele-rometers integrated with a RGB-D sensor to obtain rotation compensation and image processing method for defining pre-requisites of the known categories. The second technique uses the adapted nearest neighbour search procedure for obtaining normal vectors for each range point.     

Image registration framework for large-scale longitudinal MRI data sets: strategy and validation

Advanced magnetic resonance imaging (MRI) studies often require the transformation of large numbers of images into a common space. Calculating transformations that relate each image to every other and applying them to the images on demand are theoretically possible; however, these can be computationally prohibitive. Therefore, relating each image to only one other image, then linking those transforms together to relate any two images in the database, may be an efficient alternative. Evaluated were the feasibility and validity of image registration to bring intraindividual MR images into mutual correspondence for longitudinal analysis through the concatenation of precomputed transforms. A longitudinal data set of 10 multiple sclerosis patients with nine serial dual-echo spin-echo, 1.5-T MRI scans was used. Intrasubject registrations were performed stepwise between consecutive images and direct from each time point to the baseline. Consecutive transforms were concatenated and evaluated against direct registrations by comparing the resulting transformed images (using Pearson correlation coefficient). Confounding variables such as time between scans, brain atrophy, and change in lesion load were evaluated. We found the images resampled with the direct and the concatenated transforms to be highly correlated, and there was no significant difference between methods. Differences in brain parenchymal fraction (a measure of brain atrophy) showed significant inverse correlation with the correspondence of the resampled images. Results indicate that concatenating multiple transforms that link two images together produces near-identical results to that of direct registration; thus, this method is both useful and valid.     

Enhancement of histological volumes through averaging and their use for the analysis of magnetic resonance images

Magnetic resonance imaging (MRI) of small animals is routinely performed in research centers. But despite its many advantages, MR still suffers from limited spatial resolution which makes the interpretation and quantitative analysis of the images difficult, particularly for small structures of interest within areas of significant heterogeneity. One possibility to address this issue is to complement the MR images with histological data, which requires reconstructing 3D volumes from a series of 2D images. A number of methods have been proposed recently in the literature to address this issue, but deformation or tearing during the slicing process often produces reconstructed volumes with visible artifacts and imperfections. In this paper, we show that a possible solution to this problem is to work with several histological volumes, reconstruct each of these separately and then compute an average. The resulting histological atlas shows structures and substructures more clearly than any individual volume. We also propose an original approach to normalize intensity values across slices, a required preprocessing step when reconstructing histological volumes. We show that the histological atlas we have created can be used to localize structures and substructures, which cannot be seen easily in MR images. We also create an MR atlas that is associated with the histological atlas. We show that using the histological volumes to create the MR atlas is better than using the MR volumes only. Finally, we validate our approach quantitatively on MR image volumes by comparing volumetric measurements obtained manually and obtained automatically with our atlases.     

Automatic registration of serial sections of mouse lymph node by using Image-Reg

The first step towards the three-dimensional (3D) reconstruction of histological structures from serial sectioned tissue blocks is the proper alignment of microscope image sequences. We have accomplished an automatic rigid registration program, named Image-Reg, to align serial sections from mouse lymph node and Peyer's patch. Our approach is based on the calculation of the pixel-correlation of objects in adjacent images. The registration process is mainly divided into two steps. Once the foreground images have been segmented from the original images, the first step (primary alignment) is performed on the binary images of segmented objects; this process includes rotation by using the moments and translation through the X, Y axes by using the centroid. In the second step, the matching error of two binary images is calculated and the registration results are refined through multi-scale iterations. In order to test the registration performance, Image-Reg has been applied to an image and its transformed (rotated) version and subsequently to an image sequence of three serial sections of mouse lymph node. In addition, to compare our algorithm with other registration methods, three other approaches, viz. manual registration with Reconstruct, semi-automatic landmark registration with Image-Pro Plus and the automatic phase-correlation method with Image-Pro Plus, have also been applied to these three sections. The performance of our program has been also tested on other two-image data sets. These include: (a) two light microscopic images acquired by the automatic microscope (stitched with other software); (b) two images fluorescent images acquired by confocal microscopy (tiled with other software). Our proposed approach provides a fast and accurate linear alignment of serial image sequences for the 3D reconstruction of tissues and organs.     

Registering spherical navigators with spherical harmonic expansions to measure three-dimensional rotations in magnetic resonance imaging

Subject motion remains a challenging problem to overcome in clinical and research applications of magnetic resonance imaging (MRI). Subject motion degrades the quality of MR images and the integrity of experimental data. A promising method to correct for subject motion in MRI is the spherical navigator (SNAV) echo. Spherical navigators acquire k-space data on the surface of a sphere in order to measure three-dimensional (3D) rigid-body motion. Analysis begins by registering the magnitude of two SNAVs to determine the 3D rotation between them. Several different methods to register SNAV data exist, each with specific capabilities and limitations. In this study, we assessed the accuracy, precision and computational requirements of measuring rotations about all three coordinate axes by correlating the spherical harmonic expansions of SNAV data. We compare the results of this technique to previous SNAV studies and show that, although computationally expensive, the spherical harmonic technique is a highly accurate, precise and robust method to register SNAVs and detect 3D rotations in MRI. A key advantage to the spherical harmonic technique is the ability to optimize the accuracy, precision, processing time and memory requirements by adjusting parameters used in the registration. While present developments are aimed at improving the programming efficiency and memory handling of the algorithm, this registration technique is currently well suited for retrospective motion correction applications, such as removing motion-related image artifacts and aligning slices within a high-resolution 3D volume.     

Multicontrast multiecho FLASH MRI for targeting the subthalamic nucleus

The subthalamic nucleus (STN) is one of the most common stimulation targets for treating Parkinson's disease using deep brain stimulation (DBS). This procedure requires precise placement of the stimulating electrode. Common practice of DBS implantation utilizes microelectrode recording to locate the sites with the correct electrical response after an initial location estimate based on a universal human brain atlas that is linearly scaled to the patient's anatomy as seen on the preoperative images. However, this often results in prolonged surgical time and possible surgical complications since the small-sized STN is difficult to visualize on conventional magnetic resonance (MR) images and its intersubject variability is not sufficiently considered in the atlas customization. This paper proposes a multicontrast, multiecho MR imaging (MRI) method that directly delineates the STN and other basal ganglia structures through five co-registered image contrasts (T1-weighted navigation image, R2 map, susceptibility-weighted imaging (phase, magnitude and fusion image)) obtained within a clinically acceptable time. The image protocol was optimized through both simulation and in vivo experiments to obtain the best image quality. Taking advantage of the multiple echoes and high readout bandwidths, no interimage registration is required since all images are produced in one acquisition, and image distortion and chemical shift are reduced. This MRI protocol is expected to mitigate some of the shortcomings of the state-of-the-art DBS implantation methods.     

基于大脑模板的MRI扫描自动定位方法

在大脑磁共振成像(MRI)影像学的数据采集中,通常先扫描一幅定位图像,并根据解剖学先验知识手动调整合适的扫描定位参数,再进行后续的正式扫描.该文实现了一种直接以大脑模板为参照的自动定位的方法:首先采集一幅中等分辨率的快速三维定位图像,然后通过与模板的配准确定定位参数,并应用到后续序列的扫描,以保证不同被试在图像采集时采用与模板一致的空间定位.该方法一方面便于不同被试的图像数据之间进行系统性比较与参照,帮助诊断者快速定位病灶,也可在后续常用的基于体素分析过程最大化数据的利用效率.另一方面,针对单个体多次扫描之间的自动定位,该文进一步使用迭代方法,通过多次"扫描、配准、自动定位"步骤,逐步减小图像配准算法的误差.实验证明,该文基于大脑模板的自动定位方法能够确保不同被试之间和同一被试之内在图像数据采集时的空间定位高度一致性,其中同一被试内多次扫描的空间定位误差1.0 mm和1.0o.     

3D reconstruction of the brain from magnetic resonance images using a connectivity algorithm

We present high resolution three dimensional (3D) connectivity, surface construction and display algorithms that detect, extract, and display the surface of a brain from contiguous magnetic resonance (MR) images. The algorithms identify the external brain surface and create a 3D image, showing the fissures and surface convolutions of the cerebral hemispheres, cerebellum, and brain stem. Images produced by these algorithms also show the morphology of other soft tissue boundaries such as the cerebral ventricular system and the skin of the patient. For the purposes of 3D reconstruction, our experiments show that T1 weighted images give better contrast between the surface of the brain and the cerebral spinal fluid than T2 weighted images. 3D reconstruction of MR data provides a non-invasive procedure for examination of the brain surface and other anatomical features.     

Using nine degrees-of-freedom registration to correct for changes in voxel size in serial MRI studies

Quantitative longitudinal brain magnetic resonance (MR) studies may be confounded by scanner-related drifts in voxel sizes. Total intracranial volume (TIV) normalisation is commonly used to correct serial cerebral volumetric measurements for these drifts. We hypothesised that automated rigid-body registration of whole brain incorporating automatic scaling correction might also correct for such fluctuations, and might be a more practical alternative. Twenty-three subjects (12 patients with Alzheimer's disease [AD] and 11 controls) had at least two serial T1-weighted volumetric brain MR scans. Ten scans from the control subjects were artificially scaled (stretched) by 1.5, 3.0, 4.6 and 6.1%. A 9-degrees-of-freedom (9dof) registration was used to register the scaled scans back onto the original scans and corresponding scaling factors compared to TIV measurements. A further nine 1-year repeat scans from the AD subjects were artificially scaled and registered (9dof) to baseline. The two correction methods were further assessed using multiple serial scans for each of the 23 subjects (resulting in 49 scan pairs). All serial scans were registered (9dof) to baseline. TIV was measured on all scans. It was found that the 9dof registration successfully recovered the artificially generated scaling changes. Scaling correction using 9dof registration did not alter the amount of brain atrophy measured over the 1-year period in the AD subjects. The 9dof volume scaling factors were very similar to the TIV ratios (repeat TIV over baseline TIV), but less variable (p < 0.001), in both artificial and 'real' scenarios. In the latter, the volume scaling factors allowed identification of two time-points in which a 3% change in voxel size had occurred. Both the 9dof brain registration and TIV correction were successfully able to correct for these fluctuations. Significant shifts in voxel size are a problem in longitudinal brain imaging studies. It is important that such changes are adjusted for: 9dof registration, which is automated and computationally inexpensive, may be superior to the more labour-intensive TIV correction for this purpose.     

Increased differentiation of intracranial white matter lesions by multispectral 3D-tissue segmentation: preliminary results

MRI is a very sensitive imaging modality, however with relatively low specificity. The aim of this work was to determine the potential of image post-processing using 3D-tissue segmentation technique for identification and quantitative characterization of intracranial lesions primarily in the white matter. Forty subjects participated in this study: 28 patients with brain multiple sclerosis (MS), 6 patients with subcortical ischemic vascular dementia (SIVD), and 6 patients with lacunar white matter infarcts (LI). In routine MR imaging these pathologies may be almost indistinguishable. The 3D-tissue segmentation technique used in this study was based on three input MR images (T(1), T(2)-weighted, and proton density). A modified k-Nearest-Neighbor (k-NN) algorithm optimized for maximum computation speed and high quality segmentation was utilized. In MS lesions, two very distinct subsets were classified using this procedure. Based on the results of segmentation one subset probably represent gliosis, and the other edema and demyelination. In SIVD, the segmented images demonstrated homogeneity, which differentiates SIVD from the heterogeneity observed in MS. This homogeneity was in agreement with the general histological findings. The LI changes pathophysiologically from subacute to chronic. The segmented images closely correlated with these changes, showing a central area of necrosis with cyst formation surrounded by an area that appears like reactive gliosis. In the chronic state, the cyst intensity was similar to that of CSF, while in the subacute stage, the peripheral rim was more prominent. Regional brain lesion load were also obtained on one MS patient to demonstrate the potential use of this technique for lesion load measurements. The majority of lesions were identified in the parietal and occipital lobes. The follow-up study showed qualitatively and quantitatively that the calculated MS load increase was associated with brain atrophy represented by an increase in CSF volume as well as decrease in "normal" brain tissue volumes. Importantly, these results were consistent with the patient's clinical evolution of the disease after a six-month period. In conclusion, these results show there is a potential application for a 3D tissue segmentation technique to characterize white matter lesions with similar intensities on T(2)-weighted MR images. The proposed methodology warrants further clinical investigation and evaluation in a large patient population.     

The registration of MR images using multiscale robust methods

Acquisition of MR images involves their registration against some prechosen reference image. Motion artifacts and misregistration can seriously flaw their interpretation and analysis. This article provides a global registration method that is robust in the presence of noise and local distortions between pairs of images. It uses a two-stage approach, comprising an optional Fourier phase-matching method to carry out preregistration, followed by an iterative procedure. The iterative stage uses a prescribed set of registration points, defined on the reference image, at which a robust nonlinear regression is computed from the squared residuals at these points. The method can readily accommodate general linear, or even nonlinear, registration transformations on the images. The algorithm was tested by recovering the registration transformation parameters when a 256 × 256 pixel T₂¹-weighted human brain image was scaled, rotated, and translated by prescribed amounts, and to which different amounts of Gaussian noise had been added. The results show subpixel accuracy of recovery when no noise is present, and graceful degradation of accuracy as noise is added. When 40% noise is added to images undergoing small shifts, the recovery errors are less than 3 pixels. The same tests applied to the Woods algorithm gave slightly inferior accuracy for these images, but failed to converge to the correct parameters in some cases of large-scale-shifted images with 10% added noise.     

3-D in vivo brain tumor geometry study by scaling analysis

A new method, based on scaling analysis, is used to calculate fractal dimension and local roughness exponents to characterize in vivo 3-D tumor growth in the brain. Image acquisition was made according to the standard protocol used for brain radiotherapy and radiosurgery, i.e., axial, coronal and sagittal magnetic resonance T₁-weighted images, and comprising the brain volume for image registration. Image segmentation was performed by the application of the k-means procedure upon contrasted images. We analyzed glioblastomas, astrocytomas, metastases and benign brain tumors. The results show significant variations of the parameters depending on the tumor stage and histological origin.     

BRAHMA: Population specific T1, T2, and FLAIR weighted brain templates and their impact in structural and functional imaging studies

Differences in brain morphology across population groups necessitate creation of population-specific Magnetic Resonance Imaging (MRI) brain templates for interpretation of neuroimaging data. Variations in the neuroanatomy in a genetically heterogeneous population make the development of a population-specific brain template for the Indian subcontinent imperative. A dataset of high-resolution 3D T1, T2-weighted, and FLAIR images acquired from a group of 113 volunteers (M/F - 56/57, mean age-28.96 ± 7.80 years) are used to construct T1, T2-weighted, and FLAIR templates, collectively referred to as Indian Brain Template, “BRAHMA”. A processing pipeline is developed and implemented in a MATLAB based toolbox for template construction and generation of tissue probability maps and segmentation atlases, with additional labels for deep brain regions such as the Substantia Nigra generated from the T2-weighted and FLAIR templates. The use of BRAHMA template for analysis of structural and functional neuroimaging data obtained from Indian participants, provides improved accuracy with statistically significant results over that obtained using the ICBM-152 (International Consortium for Brain Mapping) template. Our results indicate that segmentations generated on structural images are closer in volume to those obtained from registration to the BRAHMA template than to the ICBM-152. Furthermore, functional MRI data obtained for Working Memory and Finger Tapping paradigms processed using the BRAHMA template show a significantly higher percentage of the activation area than ICBM-152 in relevant brain regions, i.e. the left middle frontal gyrus, and the left and right precentral gyri, respectively. The availability of different image contrasts, tissue maps, and segmentation atlases makes the BRAHMA template a comprehensive tool for multi-modal image analysis in laboratory and clinical settings.     

MR signal change in venous thrombus relates organizing process and thrombolytic response in rabbit

Venous thrombus is subsequently organized and replaced by fibrous connective tissue. However, the sequential changes in venous thrombi are not reliably detected by current noninvasive diagnostic techniques. The purpose of this study is to reveal whether magnetic resonance (MR) can detect venous thrombus, define thrombus age and predict thrombolytic responses. Thrombus in the rabbit jugular vein was imaged with a 1.5-T MR system at 4 h and at 1, 2 and 4 weeks using three-dimensional (3D) fast asymmetric spin echo T2-weighted (T2W) and 3D-gradient echo T1-weighted (T1W) sequences. The jugular veins were histologically assessed at each time point. Magnetic resonance imaging (MRI) was also performed in vivo before and 30 min after tissue plasminogen activator (t-PA) administration. The thrombi in MRI were comparable in size to histological sections. The signal intensity (SI) of thrombi at 4 h was heterogeneously high or low on T2W or T1W images, respectively. The SI of thrombi on T2W images decreased time-dependently, but increased on T1W images at 1 and 2 weeks. Morphological analysis showed time-dependent decreases in erythrocyte, platelet and fibrin areas and time-dependent increases in smooth muscle cell, macrophage, collagen and iron areas. The t-PA administration significantly decreased thrombus volume at 4 h but not at 1, 2 and 4 weeks. Venous thrombosis can be reliably and noninvasively detected by MRI. Measurement of SI might support assessments of thrombus age and thrombolytic response.     

Aligning 3D time-of-flight MRA datasets for quantitative longitudinal studies: evaluation of rigid registration techniques

Objective

3D Time-of-flight (TOF) magnetic resonance angiography is commonly used for vascular analyses. A quantification of longitudinal morphological changes usually requires the registration of TOF image sequences acquired at different time points. The aim of this study was to evaluate the precision of different 3D rigid registration setups such that an optimal quantification of morphological changes can be achieved.

Methods

Eight different rigid registration techniques were implemented and evaluated in this study using the target registration error (TRE) calculated based on 554 landmarks defined in twenty TOF datasets. The registration techniques differed in integration of brain and vessel segmentation masks and usage of a multi-resolution framework. Furthermore, the benefit of a prior volume-of-interest definition for registration accuracy was evaluated.

Results

The results revealed that the highest registration accuracies can be achieved using a multi-resolution framework and a cerebrovascular segmentation as mask. Numerically, a mean TRE of 1.1 mm was calculated. If applicable, a prior definition of a volume-of-interest allows a reduction of the TRE to only 0.6 mm.

Conclusion

TOF datasets should be registered using vessel segmentations as mask, multi-resolution framework and previous volume-of-interest definition if possible to obtain the highest registration precision. This is especially the case for longitudinal datasets that are separated by several months while the registration technique seems less important for datasets that are only separated by a few days.     

A modified fuzzy clustering algorithm for operator independent brain tissue classification of dual echo MR images.

Methods for brain tissue classification or segmentation of structural magnetic resonance imaging (MRI) data should ideally be independent of human operators for reasons of reliability and tractability. An algorithm is described for fully automated segmentation of dual echo, fast spin-echo MRI data. The method is used to assign fuzzy-membership values for each of four tissue classes (gray matter, white matter, cerebrospinal fluid and dura) to each voxel based on partition of a two dimensional feature space. Fuzzy clustering is modified for this application in two ways. First, a two component normal mixture model is initially fitted to the thresholded feature space to identify exemplary gray and white matter voxels. These exemplary data protect subsequently estimated cluster means against the tendency of unmodified fuzzy clustering to equalize the number of voxels in each class. Second, fuzzy clustering is implemented in a moving window scheme that accommodates reduced image contrast at the axial extremes of the transmitting/receiving coil. MRI data acquired from 5 normal volunteers were used to identify stable values for three arbitrary parameters of the algorithm: feature space threshold, relative weight of exemplary gray and white matter voxels, and moving window size. The modified algorithm incorporating these parameter values was then used to classify data from simulated images of the brain, validating the use of fuzzy-membership values as estimates of partial volume. Gray:white matter ratios were estimated from 20 twenty normal volunteers (mean age 32.8 years). Processing time for each three-dimensional image was approximately 30 min on a 170 MHz workstation. Mean cerebral gray and white matter volumes estimated from these automatically segmented images were very similar to comparable results previously obtained by operator dependent methods, but without their inherent unreliability.     

Application of fuzzy c-means segmentation technique for tissue differentiation in MR images of a hemorrhagic glioblastoma multiforme

The application of a raw data-based, operator-independent MR segmentation technique to differentiate boundaries of tumor from edema or hemorrhage is demonstrated. A case of a glioblastoma multiforme with gross and histopathologic correlation is presented. The MR image data set was segmented into tissue classes based on three different MR weighted image parameters (T₁-, proton density-, and T₂-weighted) using unsupervised fuzzy c-means (FCM) clustering algorithm technique for pattern recognition. A radiological examination of the MR images and correlation with fuzzy clustering segmentations was performed. Results were confirmed by gross and histopathology which, to the best of our knowledge, reports the first application of this demanding approach. Based on the results of neuropathologic correlation, the application of FCM MR image segmentation to several MR images of a glioblastoma multiforme represents a viable technique for displaying diagnostically relevant tissue contrast information used in 3D volume reconstruction. With this technique, it is possible to generate segmentation images that display clinically important neuroanatomic and neuropathologic tissue contrast information from raw MR image data.