Using medical images acquired during routine clinical care for research purposes: a comprehensive medical informatics approach applied to the study of brain development from birth through 2 years of age as indexed by diffusion MRI

Biologic variability and dramatic changes of brain development in children aged 0 to 2 years make it challenging to accurately detect subtle abnormalities in single Magnetic Resonance Imaging (MRI) scans. Diffusion MRI (dMRI) indices such as Apparent Diffusion Coefficient (ADC) are reliable measures of water content in the brain and thus an excellent surrogate marker for brain development. Developing robust age-specific diffusion biomarkers for quantitative measurement of normative brain evolution would enhance our ability to detect subtle alterations due to tissue injuries or neuropathological disorders. Obtaining significant numbers of normative MRI scans for this age group means redirecting clinical data from hospital databases for research purposes. Therefore, this pilot project demonstrates the feasibility of identifying, retrieving and analyzing pediatric clinical dMRI data to investigate normal brain development from birth to 2 years. Research Patient Data Registry (RPDR) at Massachusetts General Hospital (MGH) was used to collect patient medical information and identify healthy children according to radiology reports. Corresponding MRI data were retrieved from MGH Picture Archiving and Communication System (PACS) using the prototype of Medical Imaging Informatics Bench to Bedside (mi2b2) software. A specific pipeline was created to handle the volume of studies and extract technical scan information used to identify comparable diffusion series; 193 studies were used for analysis. Two markers, whole brain average of ADC and Fractional Anisotropy (FA) values (WBAADC and WBAFA), were computed for each patient, their age-evolution across patients was investigated with different models. WBAADC and WBAFA seem to exhibit biexponential decay and increase respectively and might be gender-specific. These results have clinical implications for potentially determining the health status of an unknown individual, and research utility for continued development of these tools.

Biophysical Basis of the Diffusion-Weighted Magnetic Resonance Signal in the Rat Brain

Nicolas Kunz

Nuclear magnetic resonance (NMR) can be used in-vivo in a vast array of applications, such as anatomical imaging (magnetic resonance imaging, MRI), localized chemical composition characterization (magnetic resonance spectroscopy, MRS), cellular structure assessment (diffusion tensor imaging, DTI) and cerebral activity mapping (functional imaging, fMRI) for the most important one. This thesis focused on the development of diffusion NMR spectroscopy and imaging methods at ultra-high magnetic field, with the aim of a better characterization of the diffusion mechanism in-vivo. DTI measures the water molecule displacement due to the thermal agitation in the sample. In cellular tissue, the molecules are restrained in compartments delimited by the cell membranes, which mean that DTI can provide information on the cerebral cellular microstructure. DTI is thus widely used to investigate cerebral disorders such as brain ischemia, trauma, and tumors, as well as the structural changes occurring during brain development and normal aging. However, the presence of water molecules in both the intra- and extra-cellular compartments may bias the assessed structural information. On the other hand, the metabolites, measured by 1H-MRS, are mostly localized in the intracellular compartment, which may provide a more precise insight of the cellular structure. With the availability of high magnetic field systems, 1H-MRS allows the measurement of 15 – 20 metabolites in-vivo in localized region in the order of &icro;l in the rodent brain. These metabolites are involved in cell energy management, neurotransmission, neuro-protection, neuronal development and membrane growth, and thus provide a rather complete status of the brain neurochemical profile. In this context, a diffusion tensor spectroscopy pulse sequence (DT-MRS) was implemented and used to investigate the metabolite diffusion properties in specific brain regions in-vivo at ultra-high magnetic field 14.1 T to benefit from the increased sensitivity and spectral resolution at high B0. The diffusion tensor of 5 metabolites were reconstructed in the corpus-callosum, CC, (NAA, Tau, Glu, Ins and Mac) and 8 in the cortex, Cx, (with in addition Cr, PCr, GABA and GSH). Metabolites apparent diffusion coefficients were significantly higher in the CC (0.115 – 0.153 µm2/ms) than in the Cx (0.087 – 0.107 µm2/ms) and so was the fractional anisotropy (0.51 – 0.62 and 0.34 – 0.51 respectively). In addition, the metabolite preferential diffusion directions showed an excellent agreement with the known rat brain structure. The aforementioned DW-MRS pulse sequence was then combined with an inversion recovery module, allowing direct assessment of the macromolecule (MM) signal, which overlaps with metabolites resonances. An inaccurate estimation of the MM baseline may result in a significant bias during the quantification process. At ultra-high magnetic field, the MM spectral shape becomes more complex, and therefore, the previously reported experimental techniques would require post-processing to remove residual metabolite signals. In this study, several metabolite contaminations remaining in the metabolite nulled spectrum were successfully identified and attributed to creatine, myo-inositol, taurine, choline, N-acetylaspartate, glutamine and glutamate. The macromolecule baseline assessed by DW-MRS resulted in a twofold reduction of the metabolite residuals with minimal effect on the macromolecule signal (i.e. < 8% signal reduction). Although at high magnetic field spectral resolution and sensitivity are improved, metabolite diffusion measurements remain very time consuming and challenging due to the low metabolite concentrations in-vivo (1 – 15 mM) compared to water (44 M). Therefore, a fast and robust EPI-based DTI pulse sequence was developed, implemented and validated to investigate the cerebral architecture in-vivo at 14.1 T through water diffusion. Combined with the use of a purpose-built quadrature surface coil and adiabatic RF-pulses, it resulted in images with a full coverage of the rat brain and a total acquisition time of one hour. The rapid signal loss arising from the use of a surface coil for the MRI acquisitions, which was previously in this thesis overcome experimentally, was also addressed in collaboration with the Signal Processing Core of CIBM, by a direct correction applied on reconstructed magnitude images. Based on a low-pass frequency mode; the developed method showed better accuracy in enhancing image intensity compared to standard approaches. In view of a future application of the presented diffusion NMR techniques, investigating structural changes during brain development, a study was performed to characterize different intra-uterine growth retardation (IUGR) models by 1H-MRS. Dexamethasone-treated animals (100 µg/kg/day induced from the day 15 of gestation) showed more severe alterations in cerebral energy metabolism and neurotransmitter action at postnatal day 7 and 21 than the reference IUGR model, caloric restriction (30% food intake of ad libitum fed dams from day 1 of gestation). In a continuation of the IUGR study, the exposure to the estrogen-like molecule (i.e. bisphenol-A, BPA), was investigated. Here, 1H-MRS showed a significant increase of the Glu/Asp ratio in the hippocampus of the BPA-treated group at 20 days old, indicating a possible deviation in the malate-aspartate shuttle. Finally, the diverse NMR techniques used and developed during this thesis have required the implementation of several data processing tools, which were centralized in a framework written in Matlab (Mathworks, Inc). Two principal modules were build: 1. to reconstruct DTI experiments data and 2. to facilitate and standardize the 1H-MRS spectra quantification by LCModel. The proposed NMR data processing framework was used to process most of the data presented in this thesis and has demonstrated good performance. In addition, this framework was tested and used by five external users to analyze for data-analysis in ongoing studies.

EPFL2010

Bio Simulation of Brain Ventricle Dilation in Normal Pressure Hydrocephalus

Kamal Shahim

Hydrocephalus is a brain disease wherein the ventricles dilate and compress the parenchyma towards the skull. It is primarily characterized by the disruption of the cerebrospinal fluid (CSF) flow within the ventricular system. Normal pressure hydrocephalus (NPH) is a form of hydrocephalus for which the enlargement of ventricles occurs although the intracranial pressure (ICP) remains close to normal. The pressure gradient between the source of CSF production in the ventricles and the absorption sites is reported to be very low (∼1 mm Hg), i.e. within the experimental errors. The mechanism of NPH evolution is still obscure and its distinction from the other causes of dementia such as Alzheimer and neurodegenerative diseases is difficult. The present work contributes to a better understanding of the NPH mechanism in terms of CSF disturbances and/or parenchyma defects. To this end, imaging techniques such as Magnetic resonance imaging (MRI), Diffusion tensor imaging (DTI) and Magnetic resonance elastography (MRE) are used together with a finite element (FE) model. As a final step, NPH onset and evolution are clarified via a theoretical model for healthy and NPH brains assuming a spherical geometry. The proposed mechanism is further analyzed in a realistic 3D model of the brain parenchyma. Geometries of ventricular system and skull are obtained from MRI images of a human brain. DTI data are used to establish the fiber tracts direction as well as the local frame of anisotropic elasticity and permeability. The brain parenchyma is considered as a poro-elastic material where the tissue displacement and CSF flow are modeled using the Biot's theory. A link between the CSF diffusion and CSF permeability in brain parenchyma is established and the importance of space dependent CSF content and transverse isotropic (TI) permeability is highlighted in case of low pressure gradient hydrocephalus. Calculations are carried out to simulate the ventricular dilation using FE softwares such as MATLAB® and COMSOL®. The numerical results show that consideration of space dependent CSF content and TI permeability leads to a much more realistic model for NPH in terms of CSF velocity and CSF content. Anisotropic MRE experiment is conducted over selected slices of a healthy human brain. The experimental results are statistically refined and further used to assess the healthy brain stiffness as well as the degree of anisotropy in elasticity. Moreover, the constitutive behavior of the white matter is modeled as a composite material containing fiber tracts surrounded by a matrix; with the assumption of a low fiber-matrix bonding and fiber tract undulation. A non-linear elastic model is proposed in order to take into account the load transfer from white matter matrix to fiber tracts when these are fully stretched. The unknown value of the elastic coefficients in a sick brain is determined by using inverse modeling, i.e. by adjusting these coefficients so that the right ventricle dilation is obtained. It is demonstrated that NPH development can be associated with a degradation of the brain parenchyma elastic stiffness in NPH patients. It is shown that during NPH development, a load transfer from the white matter matrix (cell bodies and interstitial fluid) to fiber tracts takes place, initiating elastic anisotropy in white matter tissues at rather large strains. An analytical approach is developed to seek the underlying NPH mechanism in a simplified model of brain. Without further refinement in the constitutive equation or adding complexity to the material behavior, the Biot's formulation is regarded as the basis. However, an absorption term is added to the Biot's model to consider the possible transparenchymal CSF resorption. The ventricle stability concept is introduced and is further utilized to investigate the equilibrium positions. The influence of different biomechanical parameters on the stable ventricle geometry is assessed and the healthy and NPH equilibrium positions are found to be dependent in particular on the CSF seepage through the ventricle wall and the absorption and permeability coefficients of the brain parenchyma. Although very simple, the proposed analytical model is able to predict the onset and development of NPH conditions as a deviation from healthy conditions. Incorporating the stability concept in a more realistic geometry of brain (3D), the respective equilibrium positions are recaptured using the parameter values provided by the analytical spherical model. The disruption of ventricle surface during the NPH development increases CSF seepage and consequently the medium permeability. A dilation dependent permeability is moreover incorporated in a 3D model of the brain. The results emphasize the importance of strain dependent permeability which favors the ventricle equilibrations in more realistic geometries of brain. Future works might consider the time dependent deformation (creep effects and stress induced remodeling) of ventricles and the incorporation of anisotropic permeability and elasticity in the 3D model. The geometry should be extended to the full ventricular system including the subarachnoid spaces (SAS).

EPFL2011

Fluid and white matter suppression with the MP2RAGE sequence

Giulio Gambarota, Tobias Kober, José Pedro Rebelo Ferreira Marques

Purpose: To develop a magnetic resonance imaging (MRI) sequence (fluid and white matter suppression, FLAWS) for generating two sets of images from a single acquisition: one with contrast similar to a T1-weighted magnetization-prepared rapid gradient-echo sequence (MPRAGE) for structural definition; the other with nulled white matter (WM) signal intensity, similar to the fast gray matter T1 inversion recovery (FGATIR) sequence, for improved delineation of subcortical brain structures. Materials and Methods: The recently proposed MP2RAGE, which is a modification of the MPRAGE and generates two image sets at different inversion times, was employed to generate the FGATIR-like contrast (FLAWS1) and MPRAGE-like contrast (FLAWS2). Five healthy volunteers were scanned at 3T and brain tissue contrast and contrast-to-noise were compared. Results: FLAWS1 and FLAWS2 exhibited similar tissue contrast and contrast-to-noise as the `` reference'' sequences, FGATIR and MPRAGE, respectively. Synthetic minimum value images generated from FLAWS1 and FLAWS2 provided a gray matter-dominant image. Conclusion: FLAWS provides two coregistered 3D volumes, one with nulled WM signal intensity and another with nulled cerebrospinal fluid. The coregistered nature of the two datasets allows for generating images that might be helpful in segmentation algorithms and clinical diagnosis. Key Words: brain; MRI; MPRAGE; brain tissue segmentation; globus pallidus; deep brain stimulation J. Magn. Reson. Imaging 2012; 35: 1063-1070. (C) 2011 Wiley Periodicals, Inc.