Magnetic Resonance Data Sets Research Articles

A fully automatic three-step liver segmentation method on LDA-based probability maps for multiple contrast MR images

Automatic 3D liver segmentation in magnetic resonance (MR) data sets has proven to be a very challenging task in the domain of medical image analysis. There exist numerous approaches for automatic 3D liver segmentation on computer tomography data sets that have influenced the segmentation of MR images. In contrast to previous approaches to liver segmentation in MR data sets, we use all available MR channel information of different weightings and formulate liver tissue and position probabilities in a probabilistic framework. We apply multiclass linear discriminant analysis as a fast and efficient dimensionality reduction technique and generate probability maps then used for segmentation. We develop a fully automatic three-step 3D segmentation approach based upon a modified region growing approach and a further threshold technique. Finally, we incorporate characteristic prior knowledge to improve the segmentation results. This novel 3D segmentation approach is modularized and can be applied for normal and fat accumulated liver tissue properties.

DYNAMIC CONTRAST-ENHANCED MAGNETIC RESONANCE IMAGING OF CANINE BRAIN TUMORS

We evaluated dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in canine brain tumors. Magnetic resonance data sets were collected on seven canine intracranial tumors with a 3 T magnet using a T1-weighted fast spin echo fluid attenuated inversion recovery sequence after an IV bolus injection (0.2 mmol/kg) of Gd-DTPA. The tumors were confirmed histopathologically as adenocarcinoma (n=1), ependymoma (n=1), meningioma (n=3), oligodendroglioma (n=1), and pituitary macroadenoma (n=1) The data were analyzed using a two-compartment pharmacokinetic model for estimation of three enhancement parameters, E(R) (rate of enhancement), Kel (rate of elimination), and Kep (rate constant), and a model-free phenomenologic parameter initial area under the Gd concentration curve (IAUGC) defined over the first 90s postenhancement. Pearson's correlations were calculated between parameters of the two methods. The IAUGC has a relatively strong association with the rate of enhancement E(R), with r ranges from 0.4 to 0.9, but it was weakly associated with Kep and Kel. To determine whether any two tumors differed significantly, the Kohnlmogorov-Smirnov test was used. The results showed that there were statistical differences (P < 0.05) between distributions of the enhancement pattern of each tumor. These kinetic parameters may characterize the perfusion and vascular permeability of the tumors and the IAUGC may reflect blood flow, vascular permeability, and the fraction of interstitial space. The kinetic parameters and the IAUGC derived from DCE-MRI present complementary information and they may be appropriate to noninvasively differentiate canine brain tumors although a larger prospective study is necessary.

Supervised Self Organizing Maps for Classification and Determination of Potentially Discriminatory Variables: Illustrated by Application to Nuclear Magnetic Resonance Metabolomic Profiling

The article describes the extension of the self organizing maps discrimination index (SOMDI) for cases where there are more than two classes and more than one factor that may influence the group of samples by using supervised SOMs to determine which variables and how many are responsible for the different types of separation. The methods are illustrated by an application in the area of metabolic profiling, consisting of a nuclear magnetic resonance (NMR) data set of 96 samples of human saliva, which is characterized by three factors, namely, whether the sample has been treated or not, 16 donors, and 3 sampling days, differing for each donor. The sampling days can be considered a null factor as they should have no significant influence on the metabolic profile. Methods for supervised SOMs involve including a classifier for organizing the map, and we report a method for optimizing this by using an additional weight that determines the relative importance of the classifier relative to the overall experimental data set in order to avoid overfitting. Supervised SOMs can be obtained for each of the three factors, and we develop a multiclass SOM discrimination index (SOMDI) to determine which variables (or regions of the NMR spectra) are considered significant for each of the three potential factors. By dividing the data iteratively into training and test sets 100 times, we define variables as significant for a given factor if they have a positive SOMDI in the training set for the factor and class of interest over all iterations.

Remote sites of structural atrophy predict later amyloid formation in a mouse model of Alzheimer's disease

Magnetic resonance (MR) imaging can provide a longitudinal view of neurological disease through repeated imaging of patients at successive stages of impairment. Until recently, the difficulty of manual delineation has limited volumetric analyses of MR data sets to a few select regions and a small number of subjects. Increased throughput offered by faster imaging methods, automated segmentation, and deformation-based morphometry have recently been applied to overcome this limitation with mouse models of neurological conditions. We use automated analyses to produce an unbiased view of volumetric changes in a transgenic mouse model for Alzheimer's disease (AD) at two points in the progression of disease: immediately before and shortly after the onset of amyloid formation. In addition to the cortex and hippocampus, where atrophy has been well documented in AD patients, we identify volumetric losses in the pons and substantia nigra where neurodegeneration has not been carefully examined. We find that deficits in cortical volume precede amyloid formation in this mouse model, similar to presymptomatic atrophy seen in patients with familial AD. Unexpectedly, volumetric losses identified by MR outside of the forebrain predict locations of future amyloid formation, such as the inferior colliculus and spinal nuclei, which develop pathology at very late stages of disease. Our work provides proof-of-principle that MR microscopy can expand our view of AD by offering a complete and unbiased examination of volumetric changes that guide us in revisiting the canonical neuropathology.

Cardinal series to sort out defective samples in magnetic resonance data sets

NMR signals are unavoidably impaired with noise stemming from the electronic circuits of the spectrometer. This noise is most often white and Gaussian and can be greatly reduced by applying low pass analogue and digital filters. Nevertheless, extra noise with other statistics than Gaussian may interfere with the signal, e.g. when auxiliary electrical devices are placed near the magnet of the NMR spectrometer. This paper reports on how one can make use of this difference in statistics to remove the noise caused by electrical devices before any further data processing. The algorithm is based on the use of a new linear low pass filter, which consists in fitting NMR data in the time domain with a cardinal series and whose spectral width can be controlled. Over other filtering methods such filter has the advantage of not distorting the signal neither at the beginning nor the end of the acquisition period. The performance of the method is demonstrated by applying it to a data set collected in a flow PGSE experiment and impaired with noise emanated from a brushed DC electric motor.

Assessment of the variability in the anatomical position and size of the subthalamic nucleus among patients with advanced Parkinson’s disease using magnetic resonance imaging

Targeting of the subthalamic nucleus (STN) during deep brain stimulation (DBS) surgery using standard atlas coordinates is used in some centers. Such coordinates are accurate for only a subgroup of patients, and subgroup size depends on the extent of inter-individual variation in STN position/size and degree to which atlas represents average anatomical relations. Few studies have addressed this issue. Sixty-two axial T(2)-weighted magnetic resonance (MR) images of the brain (1.5 T) were obtained before STN-DBS in 62 patients (37 males) with Parkinson's disease using a protocol optimized for STN visualization. Image distortion was within sub-millimeter range. Midcommissural point (MCP)-derived coordinates of STN borders, STN center, and other brain landmarks were obtained using stereotactic software. MR-derived measurements were compared to Schaltenbrand and Wahren Atlas. We evaluated 117 best-visualized STNs. STN dimensions and coordinates of its center were highly variable. STN lateral coordinate ranged 8.7 mm-14.5 mm from MCP, A-P coordinate 3.5 mm posterior to 0.5 mm anterior to MCP, and vertical coordinate 1.3 mm-6 mm below MCP. The antero-posterior nucleus dimension varied by 8 mm and lateral-medial dimension by 5.8 mm. Differences between mean values of MR-derived data sets and Atlas values were statistically significant but moderate, excluding AC-PC length, for which the Atlas value was below the 1st percentile of the MR data set. The STN lateral coordinate strongly correlated with the width of the third ventricle (r = 0.73, p < 0.001). It is now possible to directly evaluate STNs at 1.5 T with minimal image distortion, which reveals variation in STN position and dimensions in the range of nucleus size. This puts under question the rationale of use of standard STN coordinates during DBS surgery.

Unwrapping magnetic resonance phase maps with Chebyshev polynomials

A phase-unwrapping algorithm, based on the method of moments, is introduced in this work. The proposed algorithm expands the phase map in terms of a two-dimensional Chebyshev series. The expansion coefficients are calculated by exploiting the orthogonality of Chebyshev polynomials of the first kind. The performance of the proposed phase-unwrapping algorithm is tested on a synthetic phase map and experimental phase maps of a uniform phantom, a human brain and a mouse torso, all acquired from 3-T magnetic resonance (MR) scanners. To impose additional burdens on the algorithm, we introduced magnetic field inhomogeneities to the phantom and human brain data by an external gradient coil. The proposed phase-unwrapping algorithm is found to perform well on the phantom data set in a low signal-to-noise ratio (SNR) environment and on the synthetic data set. The proposed algorithm is also found to perform well in in vivo data sets of the human brain and mouse torso. Results obtained from the in vivo MR data sets show that the proposed algorithm produced unwrapped phase maps that are nearly identical to those produced by a widely used phase-unwrapping algorithm, PRELUDE 2D in the fMRI Software Library.

Gadofosveset-enhanced MR Angiography of Carotid Arteries: Does Steady-State Imaging Improve Accuracy of First-Pass Imaging? Comparison with Selective Digital Subtraction Angiography

To evaluate the diagnostic accuracy of gadofosveset-enhanced magnetic resonance (MR) angiography in the assessment of carotid artery stenosis, with digital subtraction angiography (DSA) as the reference standard, and to determine the value of reading first-pass, steady-state, and "combined" (first-pass plus steady-state) MR angiograms. This study was approved by the local ethics committee, and all subjects gave written informed consent. MR angiography and DSA were performed in 84 patients (56 men, 28 women; age range, 61-76 years) with carotid artery stenosis at Doppler ultrasonography. Three readers reviewed the first-pass, steady-state, and combined MR data sets, and one independent observer evaluated the DSA images to assess stenosis degree, plaque morphology and ulceration, stenosis length, and tandem lesions. Interobserver agreement regarding MR angiographic findings was analyzed by using intraclass correlation and Cohen kappa coefficients. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated by using the McNemar test to determine possible significant differences (P < .05). Interobserver agreement regarding all MR angiogram readings was substantial. For grading stenosis, sensitivity, specificity, PPV, and NPV were, respectively, 90%, 92%, 91%, and 91% for first-pass imaging; 95% each for steady-state imaging; and 96%, 99%, 99%, and 97% for combined imaging. For evaluation of plaque morphology, respective values were 84%, 86%, 88%, and 82% for first-pass imaging; 98%, 97%, 98%, and 97% for steady-state imaging; and 98%, 100%, 100%, and 97% for combined imaging. Differences between the first-pass, steady-state, and combined image readings for assessment of stenosis degree and plaque morphology were significant (P < .001). Gadofosveset-enhanced MR angiography is a promising technique for imaging carotid artery stenosis. Steady-state image reading is superior to first-pass image reading, but the combined reading protocol is more accurate.

Real-time multi-modal rigid registration based on a novel symmetric-SIFT descriptor

The purpose of image registration is to spatially align two or more single-modality images taken at different times, or several images acquired by multiple imaging modalities. Intensity-based registration usually requires optimization of the similarity metric between the images. However, global optimization techniques are too time-consuming, and local optimization techniques frequently fail to search the global transformation space because of the large initial misalignment of the two images. Moreover, for large non-overlapping area registration, the similarity metric cannot reach its optimum value when the two images are properly registered. In order to solve these problems, we propose a novel Symmetric Scale Invariant Feature Transform (symmetric-SIFT) descriptor and develop a fast multi-modal image registration technique. The proposed technique automatically generates a lot of highly distinctive symmetric-SIFT descriptors for two images, and the registration is performed by matching the corresponding descriptors over two images. These descriptors are invariant to image scale and rotation, and are partially invariant to affine transformation. Moreover, these descriptors are symmetric to contrast, which makes it suitable for multi-modal image registration. The proposed technique abandons the optimization and similarity metric strategy. It works with near real-time performance, and can deal with the large non-overlapping and large initial misalignment situations. Test cases involving scale change, large non-overlapping, and large initial misalignment on computed tomography (CT) and magnetic resonance (MR) datasets show that it needs much less runtime and achieves better accuracy when compared to other algorithms.

Congenital aortic disease: 4D magnetic resonance segmentation and quantitative analysis

Automated and accurate segmentation of the aorta in 4D (3D+time) cardiovascular magnetic resonance (MR) image data is important for early detection of congenital aortic disease leading to aortic aneurysms and dissections. A computer-aided diagnosis (CAD) method is reported that allows one to objectively identify subjects with connective tissue disorders from 16-phase 4D aortic MR images. Starting with a step of multi-view image registration, our automated segmentation method combines level-set and optimal surface segmentation algorithms in a single optimization process so that the final aortic surfaces in all 16 cardiac phases are determined. The resulting aortic lumen surface is registered with an aortic model followed by calculation of modal indices of aortic shape and motion. The modal indices reflect the differences of any individual aortic shape and motion from an average aortic behavior. A Support Vector Machine (SVM) classifier is used for the discrimination between normal and connective tissue disorder subjects. 4D MR image data sets acquired from 104 normal volunteers and connective tissue disorder patients MR datasets were used for development and performance evaluation of our method. The automated 4D segmentation resulted in accurate aortic surfaces in all 16 cardiac phases, covering the aorta from the aortic annulus to the diaphragm, yielding subvoxel accuracy with signed surface positioning errors of -0.07+/-1.16 voxel (-0.10+/-2.05mm). The computer-aided diagnosis method distinguished between normal and connective tissue disorder subjects with a classification correctness of 90.4%.

P.1.c.019 Regulation of pro-apoptotic Fas/FADD protein complex in brain regions of CB1-receptor deficient mice

In this paper we present a benchmarking framework for the validation of cardiac motion analysis algorithms. The reported methods are the response to an open challenge that was issued to the medical imaging community through a MICCAI workshop. The database included magnetic resonance (MR) and 3D ultrasound (3DUS) datasets from a dynamic phantom and 15 healthy volunteers. Participants processed 3D tagged MR datasets (3DTAG), cine steady state free precession MR datasets (SSFP) and 3DUS datasets, amounting to 1158 image volumes. Ground-truth for motion tracking was based on 12 landmarks (4 walls at 3 ventricular levels). They were manually tracked by two observers in the 3DTAG data over the whole cardiac cycle, using an in-house application with 4D visualization capabilities. The median of the inter-observer variability was computed for the phantom dataset (0.77 mm) and for the volunteer datasets (0.84 mm). The ground-truth was registered to 3DUS coordinates using a point based similarity transform. Four institutions responded to the challenge by providing motion estimates for the data: Fraunhofer MEVIS (MEVIS), Bremen, Germany; Imperial College London - University College London (IUCL), UK; Universitat Pompeu Fabra (UPF), Barcelona, Spain; Inria-Asclepios project (INRIA), France. Details on the implementation and evaluation of the four methodologies are presented in this manuscript. The manually tracked landmarks were used to evaluate tracking accuracy of all methodologies. For 3DTAG, median values were computed over all time frames for the phantom dataset (MEVIS = 1.20 mm, IUCL = 0.73 mm, UPF = 1.10 mm, INRIA = 1.09 mm) and for the volunteer datasets (MEVIS = 1.33 mm, IUCL = 1.52 mm, UPF = 1.09 mm, INRIA = 1.32 mm). For 3DUS, median values were computed at end diastole and end systole for the phantom dataset (MEVIS = 4.40 mm, UPF = 3.48 mm, INRIA = 4.78 mm) and for the volunteer datasets (MEVIS = 3.51 mm, UPF = 3.71 mm, INRIA = 4.07 mm). For SSFP, median values were computed at end diastole and end systole for the phantom dataset(UPF = 6.18 mm, INRIA = 3.93 mm) and for the volunteer datasets (UPF = 3.09 mm, INRIA = 4.78 mm). Finally, strain curves were generated and qualitatively compared. Good agreement was found between the different modalities and methodologies, except for radial strain that showed a high variability in cases of lower image quality.

P.1.c.018 β1-adrenergic receptor activation evokes a long-lasting increase in hippocampal theta (4−15Hz) rhythm in vitro

In this paper we present a benchmarking framework for the validation of cardiac motion analysis algorithms. The reported methods are the response to an open challenge that was issued to the medical imaging community through a MICCAI workshop. The database included magnetic resonance (MR) and 3D ultrasound (3DUS) datasets from a dynamic phantom and 15 healthy volunteers. Participants processed 3D tagged MR datasets (3DTAG), cine steady state free precession MR datasets (SSFP) and 3DUS datasets, amounting to 1158 image volumes. Ground-truth for motion tracking was based on 12 landmarks (4 walls at 3 ventricular levels). They were manually tracked by two observers in the 3DTAG data over the whole cardiac cycle, using an in-house application with 4D visualization capabilities. The median of the inter-observer variability was computed for the phantom dataset (0.77 mm) and for the volunteer datasets (0.84 mm). The ground-truth was registered to 3DUS coordinates using a point based similarity transform. Four institutions responded to the challenge by providing motion estimates for the data: Fraunhofer MEVIS (MEVIS), Bremen, Germany; Imperial College London - University College London (IUCL), UK; Universitat Pompeu Fabra (UPF), Barcelona, Spain; Inria-Asclepios project (INRIA), France. Details on the implementation and evaluation of the four methodologies are presented in this manuscript. The manually tracked landmarks were used to evaluate tracking accuracy of all methodologies. For 3DTAG, median values were computed over all time frames for the phantom dataset (MEVIS = 1.20 mm, IUCL = 0.73 mm, UPF = 1.10 mm, INRIA = 1.09 mm) and for the volunteer datasets (MEVIS = 1.33 mm, IUCL = 1.52 mm, UPF = 1.09 mm, INRIA = 1.32 mm). For 3DUS, median values were computed at end diastole and end systole for the phantom dataset (MEVIS = 4.40 mm, UPF = 3.48 mm, INRIA = 4.78 mm) and for the volunteer datasets (MEVIS = 3.51 mm, UPF = 3.71 mm, INRIA = 4.07 mm). For SSFP, median values were computed at end diastole and end systole for the phantom dataset(UPF = 6.18 mm, INRIA = 3.93 mm) and for the volunteer datasets (UPF = 3.09 mm, INRIA = 4.78 mm). Finally, strain curves were generated and qualitatively compared. Good agreement was found between the different modalities and methodologies, except for radial strain that showed a high variability in cases of lower image quality.

THREE-DIMENSIONAL VISUALIZATION OF NEUROVASCULAR COMPRESSION

Abstract Objective: To assess the usefulness of presurgical simulation of microvascular decompression (MVD) by virtual endoscopy (VE), a new tool to analyze three-dimensionally reconstructed magnetic resonance data sets in patients with trigeminal neuralgia or hemifacial spasm (HFS). Methods: In 17 patients (10 with trigeminal neuralgia and seven with HFS) determined to be candidates for MVD, we performed presurgical simulation of MVD using VE. We used constructive interference in steady-state imaging and magnetic resonance angiography to obtain the original images. VE findings were compared with surgical findings. Results: The three-dimensional relations between visible structures seen on VE were consistent with intraoperative findings in all patients. In total, 20 (91%) of 22 neurovascular compression sites in all 17 patients were correctly delineated on VE, with the exception of two small branches identified as offending vessel in two patients with HFS. Perforators that were not apparent on VE limited our ability to accomplish transpositioning of the offending vessels as simulated. The positions of structures that can affect individual surgical approaches, such as the petrosal vein, cerebellar flocculus, and vertebral artery, were also adequately predicted on VE. All patients had excellent surgical outcomes. Conclusion: Presurgical VE in patients with trigeminal neuralgia or HFS is a novel technique that provides excellent visualization of the three-dimensional relations between neurovascular structures and allows simulation of MVD.

SPECIAL ASPECTS OF DIAGNOSTIC IMAGING FOR RADIOSURGERY OF ARTERIOVENOUS MALFORMATIONS

Radiosurgery can be considered a well-established option for the treatment of arteriovenous malformations (AVMs). The exact application of the therapeutic dose is based on the availability of imaging data sets with superior image quality that can be superimposed using an image fusion algorithm. For follow-up studies, the quantitative comparison of the respective image data sets also plays an important role. Until now, digital subtraction angiography (DSA) has been a mandatory tool for treatment planning and follow-up procedures. The aim of this study was to investigate whether a suitable computed tomographic (CT) and/or magnetic resonance (MR) angiography procedure can replace DSA and, if so, in which cases. For 34 AVM patients, various MR data sets were used together with the stereotactically localized CT and DSA data sets for treatment planning. To define the AVM nidus precisely, all available MR data sets were fused onto the CT data set by the use of an automatic image fusion algorithm. The nidus was outlined in both localized DSA projections, resulting in the DSA target volume. Subsequently, the DSA target volume was adapted by inclusion of the available CT/MR data sets (localized and/or fused, slice by slice), resulting in the final target volume. Finally, both volumes were compared and analyzed. For precise comparison purposes, all available digital follow-up studies were fused. In all cases, the thin-slice MR data sets (1-mm slice width) that included T1-weighted series and time of flight angiographies have been precisely fused onto the stereotactically localized treatment planning CT. The final target volume was compared with the DSA target volume as follows. In 19 cases, the final target volume was larger than the DSA target volume; in six cases, it was smaller; and in five cases, it was approximately equal. The difference was significant (Wilcoxon test, difference <0.0001; t test, t = 3.01; P > 0.005). In four cases, outlining the AVM was not possible without DSA. In five patients, a two- or three-vessel DSA was needed because there were different AVM compartments. In cases in which a previous partial embolization had been undergone by the patient, the use of superimposed CT sets with and without contrast medium was important to define the completely embolized partial volumes that were not subject to treatment. The inclusion of the DSA images enabled a better identification of those arterialized veins that did not belong to the nidus. In six cases, the follow-up MR studies showed contrast enhancements overlapping the AVM nidus as a result of brain-blood barrier disturbances (T1-weighted series with contrast). In seven cases, perifocal reactions were primarily observed (T2-weighted series) 12 months after treatment with rather low clinical relevance. By integrating all available imaging modalities, the exact three-dimensional definition of the AVM nidus was safely realized for all patients. Stereotactic DSA data acquisition remains a crucial tool for safe nidus definition in radiosurgery treatment planning and cannot, therefore, be discarded at present. It is recommended that a quantitative comparison of all MR follow-up studies be established.

Mapping of Hepatic Vascular Anatomy: Dynamic Contrast-enhanced Parallel MR Imaging Compared with 64–Detector Row CT

The study was approved by the institutional review board, and informed consent was obtained from all patients. The purpose of this study was to retrospectively evaluate the feasibility, reliability, and accuracy of breath-hold dynamic contrast material-enhanced parallel gradient-echo (GRE) magnetic resonance (MR) imaging for mapping the hepatic vascular anatomy, with contrast-enhanced 64-detector row computed tomography (CT) as the reference standard. The parallel GRE MR data sets of 100 patients acquired at 1.5 T were evaluated independently by two blinded readers with respect to (a) image quality for depiction of the hepatic arteries and the portal and hepatic veins and (b) presence of arterial stenosis and variant hepatic vasculature. The readers rated image quality to be good or excellent for 91.1%-100% of the vessels. At parallel GRE MR imaging, the readers diagnosed variant hepatic vessels and arterial stenosis with 94%-100% accuracy. They concluded that parallel GRE MR imaging, as compared with 64-detector row CT, is feasible for hepatic vascular mapping and enables reliable and accurate detection of variant hepatic vasculature and diagnosis of arterial stenosis. http://radiology.rsnajnls.org/cgi/content/full/2453062103/DC1.

A New Validation Method for X-ray Mammogram Registration Algorithms Using a Projection Model of Breast X-ray Compression

Establishing spatial correspondence between features visible in X-ray mammograms obtained at different times has great potential to aid assessment and quantitation of change in the breast indicative of malignancy. The literature contains numerous nonrigid registration algorithms developed for this purpose, but existing approaches are flawed by the assumption of inappropriate 2-D transformation models and quantitative estimation of registration accuracy is limited. In this paper, we describe a novel validation method which simulates plausible mammographic compressions of the breast using a magnetic resonance imaging (MRI) derived finite element model. By projecting the resulting known 3-D displacements into 2-D and generating pseudo-mammograms from these same compressed magnetic resonance (MR) volumes, we can generate convincing images with known 2-D displacements with which to validate a registration algorithm. We illustrate this approach by computing the accuracy for two conventional nonrigid 2-D registration algorithms applied to mammographic test images generated from three patient MR datasets. We show that the accuracy of these algorithms is close to the best achievable using a 2-D one-to-one correspondence model but that new algorithms incorporating more representative transformation models are required to achieve sufficiently accurate registrations for this application.

Magnetic Resonance‐Based Female Pelvic Anatomy as Relevant for Maternal Childbirth Injury Simulations

The objectives of the study are to review the female pelvic floor anatomy relevant to childbirth simulations, to discuss available methods for clinical evaluation of female pelvic floor function, and to review the variation in pelvic floor changes after vaginal childbirth. A high-resolution magnetic resonance (MR) data set from an asymptomatic nullipara was used to illustrate the MR anatomy of the female pelvic floor. Manual segmentation was performed and three-dimensional reconstructions of the pelvic floor structures were generated, which were used to illustrate the 3D anatomy of the pelvic floor. Variation in the post partum appearance of the levator ani muscles is illustrated using other 2D MR data sets, which depict unilateral and bilateral disruptions in the puborectalis portion of levator ani, as well as shape variations, which may be seen in the post partum levator. The clinical evaluation of the pelvic floor is then reviewed. The female pelvis is composed of a bony scaffold, from which the pelvic floor muscles (obturator internus, levator ani) are suspended. The rectum fits in a midline groove in the levator ani. The vagina is suspended across the midline, attaching bilaterally to the obturator and levator ani. The vagina supports the bladder and urethra. MR studies have demonstrated disruptions in levator ani attachments after vaginal childbirth. Such disruptions are rare in women who have not given birth vaginally. Changes to the neuromuscular apparatus of the pelvic floor can also be demonstrated after vaginal delivery. The combination of childbirth-related anatomic and neurological injury to the pelvic floor may be associated with pelvic floor dysfunction (PFD). These changes are difficult to study in vivo but may be studied through simulations. Appropriate consideration of clinical anatomy is important in these simulations.

Automatic Segmentation of the Caudate Nucleus From Human Brain MR Images

We describe a knowledge-driven algorithm to automatically delineate the caudate nucleus (CN) region of the human brain from a magnetic resonance (MR) image. Since the lateral ventricles (LVs) are good landmarks for positioning the CN, the algorithm first extracts the LVs, and automatically localizes the CN from this information guided by anatomic knowledge of the structure. The face validity of the algorithm was tested with 55 high-resolution T1-weighted magnetic resonance imaging (MRI) datasets, and segmentation results were overlaid onto the original image data for visual inspection. We further evaluated the algorithm by comparing automated segmentation results to a "gold standard" established by human experts for these 55 MR datasets. Quantitative comparison showed a high intraclass correlation between the algorithm and expert as well as high spatial overlap between the regions-of-interest (ROIs) generated from the two methods. The mean spatial overlap +/- standard deviation (defined by the intersection of the 2 ROIs divided by the union of the 2 ROIs) was equal to 0.873 +/- 0.0234. The algorithm has been incorporated into a public domain software program written in Java and, thus, has the potential to be of broad benefit to neuroimaging investigators interested in basal ganglia anatomy and function.

Automatic estimation of the noise variance from the histogram of a magnetic resonance image

Estimation of the noise variance of a magnetic resonance (MR) image is important for various post-processing tasks. In the literature, various methods for noise variance estimation from MR images are available, most of which however require user interaction and/or multiple (perfectly aligned) images. In this paper, we focus on automatic histogram-based noise variance estimation techniques. Previously described methods are reviewed and a new method based on the maximum likelihood (ML) principle is presented. Using Monte Carlo simulation experiments as well as experimental MR data sets, the noise variance estimation methods are compared in terms of the root mean squared error (RMSE). The results show that the newly proposed method is superior in terms of the RMSE.

Aortoiliac hemodynamic and morphologic adaptation to chronic spinal cord injury

Reduced lower limb blood flow and resistive hemodynamic conditions potentially promote aortic inflammation and aneurysmal degeneration. We used abdominal ultrasonography, magnetic resonance imaging, and computational flow modeling to determine the relationship between reduced infrarenal aortic blood flow in chronic spinal cord injury (SCI) subjects and risk for abdominal aortic aneurysm (AAA) disease. Aortic diameter in consecutive SCI subjects (n = 123) was determined via transabdominal ultrasonography. Aortic anatomic and physiologic data were acquired via magnetic resonance angiography (MRA; n = 5) and cine phase-contrast magnetic resonance flow imaging (n = 4) from SCI subjects whose aortic diameter was less than 3.0 cm by ultrasonography. Computational flow models were constructed from magnetic resonance data sets. Results were compared with those obtained from ambulatory control subjects (ultrasonography, n = 129; MRA/phase-contrast magnetic resonance flow imaging, n = 6) who were recruited at random from a larger pool of risk factor-matched individuals without known AAA disease. Age, sex distribution, and smoking histories were comparable between the SCI and control groups. In the SCI group, time since injury averaged 26 +/- 13 years (mean +/- SD). Aortic diameter was larger (P < .01), and the prevalence of large (> or = 2.5 cm; P < .01) or aneurysmal (> or = 3.0 cm; P < .05) aortas was greater in SCI subjects. Paradoxically, common iliac artery diameters were reduced in SCI subjects (< 1.0 cm; 48% SCI vs 26% control; P < .0001). Focal preaneurysmal enlargement was noted in four of five SCI subjects by MRA. Flow modeling revealed normal flow volume, biphasic and reduced oscillatory flow, slower pressure decay, and reduced wall shear stress in the SCI infrarenal aorta. Characteristic aortoiliac hemodynamic and morphologic adaptations occur in response to chronic SCI. Slower aortic pressure decay and reduced wall shear stress after SCI may contribute to mural degeneration, enlargement, and an increased prevalence of AAA disease.