Brain Imaging Research (BIR) - Center for Imaging of Neurodegenerative Diseases

4/4/2007
Dr. David Saloner
Novel assessment of neuro-interventional procedures in an XMR suite

Interventional MRI procedures are typically performed under x-ray fluoroscopy guidance. X-ray fluoroscopy provides high contrast, large field-of-view images that have high temporal resolution and are well-suited to procedures that require rapid dynamic updating. However they contain little or no information on the surrounding soft tissue, in particular that of the end-organ, are generally only obtained as projectional 2D images and do not provide information on the three-dimensional geometry of the vasculature, and cannot determine functional measures such as velocity fields or perfusion maps. MR imaging provides excellent visualization of soft tissue anatomy but is difficult to use as an interventional methodology because of incompatibilities with the devices that are used, the difficulty in providing high-quality roadmap images of the kind that are used in x-ray fluoroscopy, and the relatively low frame rates. We have investigated the use of a combined X-ray/MRI suite – the XMR suite – for image-guided interventions. Patients can be rapidly moved from one scanning device to another while remaining completely immobile on the imaging table. Using this technology we have been able to evaluate patients immediately pre and immediately post neuro-interventional procedures such as carotid stenting, or embolization of meningiomas. These results show that new insights into physiological responses can be obtained that were not previously available.

2/21/2007
Andreas Ebel
Accelerated 3D Echo-Planar Spectroscopic Imaging at 4 Tesla Using Modified Blipped Phase-Encoding

Whole-brain echo-planar spectroscopic imaging often substantially lengthens MRI/MRSI protocols. To halve acquisition time, application of a blipped phase-encoding gradient during the EPSI readout was previously suggested to accomplish phase-encoding (PE) of the even readout echoes in k-space at an interstitial location along k_PE, separated from the odd readout echoes. Thus, the number of phase-encoding steps, and hence acquisition time, could be reduced by a factor of 2. However, the approach is very susceptible to phase inconsistencies between even and odd readout echoes in the presence of B0 inhomogeneities and gradient imbalance, leading to ghosting in the phase-encoding direction. In this work, the blipped phase-encoding gradient is placed in between pairs of even/odd readout gradient lobes to prevent phase inconsistencies between even and odd echoes from interfering with the phase-encoding. This approach is demonstrated in a phantom and in normal human brain in vivo at 4T.

01/12/1007
Kio Kim
Ph.D. candidate in the Department of Physics at Brown University
Mosaicing and super-resolution of forward looking sonar images

Multibeam forward looking sonar (FLS) is a novel underwater imaging modality that is popularly employed for surveillance, navigation and exploration. Its specialized acquisition process yields good geometrical properties that enables image enhancement via image fusion methods. However, FLS images have a unique transformation geometry and image degradation model that differs from conventional optical images. In this talk, I will present algorithms for mosaicing and super-resolution for FLS image sequences. I will discuss this geometry and introduce robust methods for registering low quality sonar images and fusing images with varying reliability.

12/13/2006
Ashish Raj
Spatial Priors for Better MR imaging

I will introduce the concept of using spatial image priors to solve a number of estimation tasks in MR. Most MR images tend to possess spatial coherence, that is, the expectation that the physical attributes of voxels dont change abruptly from one to the next. I will show how to encode these intuitive expectations in a way that allows us to obtain better estimates of relevant MR quantities. I will describe a few concrete examples, including MR image reconstruction, estimation of T1 from multiple flip angles, etc.

11/29/2006
Jerry Matson
Arterial Spin Labeling: Comparison of Continuous and Pseudo-Continuous Methods

Blood perfusion measurements by continuous arterial spin labeling make use of velocity-driven adiabatic fast passage for inversion of flowing blood. For multiple slice imaging, the control scan typically makes use of a double inversion, which returns the blood magnetization to its original orientation. The power of the RF used for the double inversion is adjusted to provide the same magnetization transfer effects as produced by the labeling RF (producing the single inversion). This method is compared with the pseudo-continuous method developed by the Alsop group (1), in which different series of slice-selective pulses are used for label (inversion) and control (null) scans. Although the original null pulse for the sequence did not perform well, we were able to improve the null pulse performance (2). The mechanisms by which the Alsop pulses perform are elucidated, and the advantages and disadvantages of both methods are compared.

6/28/2006
Linda Chao
fMRI

Gulf War Illness (GWI) is the name given to a variety of psychological and physical symptoms suffered by veterans of the first Gulf War. The symptoms have been remarkably wide-ranging, and sometimes somewhat ill-defined, but typically include chronic fatigue, loss of muscle control, diarrhea, migraines, dizziness, and memory problems. The goal of this study was to use functional MRI (fMRI) to determine whether there are any differences between the brains of veterans with GWI and those without GWI that are not accounted for by confounds such as post traumatic stress disorder (PTSD) during memory retrieval. Behaviorally, there were no significant differences between veterans with GWI (N= 13) and those without GWI (N = 11) on behavioral tests of episodic memory. Functionally, veterans without GWI activated a region in the left inferior frontal cortex more than veterans with GWI during memory retrieval. This inferior frontal brain region has previously implicated in the successful memory retrieval. Three of the veterans with GWI also had PTSD. When the data was analyzed as a function of PTSD, we found more hippocampal activation in veterans without PTSD than in veterans with GWI and PTSD during memory retrieval. These preliminary suggests that there may be functional differences between the brains of veterans with GWI and those with PTSD that are not detectable with behavioral episodic memory tests.

5/17/2006
Karl Young
NiPy

Considerable chaos and confusion exists re. attempting to choose appropriate algorithms and software for analyzing neuroimaging data. Some speculation as to the cause of the state of affairs and how to alleviate it will be offered. In particular it will be argued that a recent development, birth of the NiPy package, has the makings of a potential solution to this problem.

Power Point Presentation

4/5/2006
Valerie Cardenas-Nicholson
Deformation Morphometry: Application in Prediction of Cognitive Decline and Frontotemporal Dementia

Previous studies have shown that brain atrophy predicts conversion to dementia, but these studies have generally investigated limited brain regions using region of interest analysis. Few studies have investigated the anatomical substrate of changes in cognition over time. This talk will present results of determining the spatial pattern of brain atrophy at baseline associated with future cognitive decline. This talk will also present preliminary results determining the spatial pattern of brain atrophy that distinguishes FTLD (fronto-temporal lobar degeneration, one of the most common causes of dementia in the elderly) from AD (Alzheimer's disease), and underlies cognitive deficits in dementia.

1/25/2006
Karl Young
Complexity derived cortical thinning

Recent developments in algorithm development and application of complexity measures to study of neurodegeneration will be described. In particular application of the methods to a data set generated by Lerch and Evans to study various techniques for measurement of cortical thinning will be discussed.

Power Point Presentation

12/14/2005
Stefan Gazdzinski
Body fat and the brain.

Increased amount of body fat over lifetime was recently identified as a risk factor for late-life dementia. The biological link between these conditions is not known. It may involve low-grade tissue inflammation, cerebrovascular disease, and/or diabetes, which may have detrimental effects on brain tissue before onset of dementia. In this talk, I will present relationships between body mass index (measure of body fat) and brain morphology and brain metabolites (NAA, choline, creatine, and myoinositol) in a group of healthy controls from studies performed in this lab.

11/16/2005
Stefan Gazdzinski / Tim Durazzo
Quantitative brain MRI in alcohol dependence: Preliminary evidence for effects of concurrent chronic cigarette smoking on regional brain volumes.

Although, approximately 80% of alcohol-dependent individuals are regular smokers, previous alcohol research did not consider the potential effects of smoking. In this seminar of this series, we will present evidence of detrimental effects of cigarette smoking on regional brain volumes in recovering alcoholics.

8/10/2005
Xiaoping
Data Collection and Processing in Parallel imaging

The main reason we are interested in parallel images is that shorter echo train lengths of EPI meant less geometrical distortion and less motion artefacts. Sensitivity encoding (SENSE) reconstruction was employed for diffusion tensor imaging (DTI) and three dimension echo planar imaging (3D-EPSI) of the brain. 1) Sensitivity Encoded Diffusion Tensor Imaging 8 health volunteers were scanned at 4T. DTI data were acquired. SENSE reconstruction was performed. Fraction anisotropy (FA) maps were obtained. We draw attention that with high acceleration factor SENSE significantly amplified noise, induced bias, and altered biologically distinction features, such as age correlation of FA in genu (slide 1 - 2). 2) Sensitivity Encoded 3D Echo Planar Spectroscopic Imaging (3D-EPSI) A method of incorporating SENSE to 3D echo-EPSI was based on simultaneously acquired metabolite and interleaved water reference 3D-EPSI. Typical proton spectra from gray matter areas and images of brain metabolites, including NAA, Cr and Cho, are presented, demonstrating the feasibility (slide 3 and 4). However severe SNR degradation of SENSE reconstructed 1H spectra was found. We consider that the reconstruction algorithm suffers the problem of phase cancellation. Other parallel imaging method, such generalized auto-calibrating partially parallel acquisitionm, which is more flexible for implementation of phase alignment between coil images, is more appropriate for reconstruction of spectroscopic images, such as 3D-EPSI.

Power Point Presentation

1/26/2005
Karl Young
The big picture, Data-mining, Data infrastructure

Data mining methods and infrastructure are required for extraction of useful information from the large amounts of multi-modal imaging data currently available from studies of brain structure and function. Recent advances will be discussed and speculation on the appropriate steps for building and utilizing a robust medical imaging data mining infrastructure will be provided.

Power Point Presentation

9/22/2004
Karl Young
Complexity: Theory and Preliminary Results

Progress in Application of Complexity Measures in Imaging Studies A recently developed package which allows for the simulation of MR images given various hypothesized conditions will be described in this talk. The simulator was used to simulate MR spectroscopic images which were in turn used to test the efficacy of information theoretic complexity measures at detecting subtle changes in the images. These results and a description of the theoretical background of the use of complexity measures in imaging will be provided.

Power Point Presentation

6/30/2004
Gerald Matson
RF Pulse Design

At high field (4.0 Tesla and higher) the RF (B1) field is necessarily inhomogeneous, resulting in spatially dependent excitation, with resulting loss in tissue contrast, and/or loss in S/N. In addition, suppression of the larger susceptibility effects at high field requires that the slice-selective pulses be applied in the presence of large slice-selective gradients. However, the limited RF power typically available at high field limits the RF pulse bandwidth, particularly for spin echo pulses, thus limiting the gradient strength that can be used. RF pulse designs to remedy both of these problems are presented. To remedy the non-uniform excitation, the design of B1 insensitive pulses is presented in which uniform excitation is achieved even in the presence of non-uniform B1 fields. To remedy the RF pulse power limitation, the design of lowered voltage spin echo pulses is presented. Both designs are based on concatenation of selective RF pulses applied in the presence of gradients of alternating sign.

5/5/2004
Lara Stables
High field MRI review

Structural Imaging at 4T This talk will outline some of the challenges and opportunities which arise as we optimize our structural imaging sequences for the 4 T scanner. I will review the literature on high field (3-8T) structural imaging and discuss some new sequences which we might consider. I will also show the results of computer simulations of gray/white/csf contrast for various permutations of IR, FLASH, and MPRAGE sequences.

4/21/2004
Karl Young
Complexity

Data Mining and Information Theory for Multi-modal Imaging Studies. Given the large amounts of multi-modal imaging data currently obtained in various studies of brain structure and function, methods for efficiently extracting maximal information from this data would be of obvious benefit. Toward this end an overall data mining infrastructure and specific information theoretic techniques for data summarization will be described in this talk.

Power Point Presentation

4/7/2004
XiaoPing Zhu
Dynamic pefusion at 4T

To simultaneously measure perfusion and permeability is important. Dynamic ASL-MRI is entirely non-invasive. Dynamic ASL-MRI can be repeated after very short interval. The purpose of this talk is to introduce a new kinetic model, which includes effects of flow, transit times and vessel leakage, and apply the model to the labeled-blood time course of ASL-MRI data. The new four-phase single-capillary step-wise model divides the time-course of the ASL signal into 4 phases regarding to: 1) the arrival time of labeled water at the voxel of interest; 2) the transit time through the arteries in the voxel; 3) the duration of the bolus of labeled spins. Preliminary esults demonstrated the feasibility of simultaneously perfusion and endothelial permeability from human subjects (slide 1). A comparison of fitting uncertainties between data acquired from 1.5T and 4T scanner shows it might be possible to measure white matter perfusion at 4T (slide 2) with better signal to noise ratio.

Power Point Presentation

2/4/2004
Valerie Cardenas-Nicholson
Statistical Models for TBM

Switching the Dependent and Independent Variables in Deformation Morphometry (a bit of linear algebra and some surprising results). In voxel-by-voxel linear modeling of brain images (e.g., analysis of PET images, voxel-based morphometry, deformation morphometry), the images have always been the dependent variable in the linear model. For some models, it may be desirable for the image to be the independent variable, but this is computationally inefficient. This talk will present linear algebra short-cuts to increase computation speed when the image is the independent variable, and demonstrate the effects of switching the dependent and independent variables in deformation morphometry.

11/19/2003
Linda Chao
NAA inCIND and outlook for fMRI

Chronic alcohol-associated brain injury has been described in neuropathological studies and brain atrophy in alcohol-dependent individuals has been studied in alcoholics in treatment. Here we report our recent findings of cortical atrophy and frontal brain metabolite abnormalities that can be detected by MR methods in non-treatment seeking heavy drinkers. This sample constitutes a much larger fraction of alcohol-dependent individuals than the alcohol abuse treatment samples previously studied.