This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Non-invasive imaging biomarkers hold tremendous potential for the evaluation of treatment response and accelerating drug development for diseases such as Alzheimer s, fatty liver (figure 1), hemochromatosis and multiple sclerosis. Biomarkers that are able detect early and subtle changes in patients can provide opportunity for pre-emptive intervention.
Magnetic resonance imaging (MRI) parameters such as proton density, longitudinal relaxation time (T1), transverse relaxation times (T2, T2*) and resonance frequency are sensitive and valid indicators for the changes that occur in patients with neurodegenerative diseases and diseases related to iron overload and insulin resistance.
Importantly, MRI uses no ionizing radiation and is more suitable than other imaging modalities such as computer tomography (CT) and x-ray for longitudinal studies and for imaging children. There is an increasing interest to assess therapies using MRI parameters as surrogate markers of disease (e.g.: decreased myelin water-fraction in neurodegenerative diseases or increased hepatic fat-fraction in fatty liver disease).
There is an urgent need for methods that can accurately assess the myelin water-fraction, longitudinal relaxation time (T1) and transverse relaxation times (T2, T2*) for early detection and staging of neurodegenerative diseases and diseases related to iron overload and insulin resistance. I aim to create a research program that focuses on the development and validation of non-invasive quantitative magnetic resonance imaging (qMRI) biomarkers for accurate diagnosis and grading of diseases such as Alzheimer s, hemochromatosis and multiple sclerosis. In addition to reducing bias, minimizing the variance of the estimates is important in quantitative measurements. I aim to improve the accuracy and precision of the quantitative MRI measurements by computing the optimal acquisition parameters that maximize the noise performance, using Cram r-Rao bound analysis.
Quantitative measurement using magnetic resonance imaging is challenging due to confounding factors such as B0 and B1 inhomogeneities, eddy currents, gradient non-linearities, spectral complexity and noise. Correction of the errors in the quantitative MRI measurements arising from the confounding factors often requires acquisition of more data and the resulting longer scan durations reduce patient comfort. Improving the data acquisition speed is of critical importance for translating quantitative MRI methods to clinical practice. Scan duration can be reduced by using low resolution acquisitions. But, low resolution acquisitions may introduce partial volume effects that might corrupt the parameter estimates. For a given scan duration, improving resolution often reduces signal-to-noise-ratio (SNR). Achieving the optimum tradeoff between resolution, scan duration and SNR is important.
MRI is non-invasive, multi-parametric and provides excellent qualitative soft-tissue contrast. However, the above mentioned and other similar confounding factors need to be addressed to bring a paradigm shift from qualitative to quantitative MR imaging.
Non-alcoholic fatty liver disease (NAFLD) is recognized as the most prevalent chronic liver disease in USA, affecting up to 80 million Americans. Chemical-shift2-4 based fat quantification methods that aim to detect and quantitatively grade NAFLD acquire MR images at different echo times, during which T2* decay occurs. Generally, methods that correct for this exponential decay to improve the accuracy of fat quantification assumed a common T2* value for water and fat signals (single T2* method5,6).
Independent Estimation of T2* for water and Fat
For my doctoral dissertation, I developed and validated a novel chemical-shift based fat-water separation method7,8 that independently estimates the T2* for water and fat for accurately quantifying the hepatic fat-fraction in the presence of field inhomogeneities and spectral complexity of fat. The dual T2* method estimates the parameters using a modified Gauss-Newton method for multiple variables. The estimates from the single T2* method, used as the initial guess, were iteratively updated using Taylor s first order approximation, and were further optimized to reduce the least squares error by performing a linear search. The dual T2* method improved the accuracy of fat-fraction estimates by 25%, in a fat-water-superparamagnetic iron oxide (SPIO) phantom as compared to the fat-fraction estimates from the single T2* method (figure 2).
Cram r-Rao Bound Analysis
In general, adding additional degrees of freedom to provide more accurate estimates of fat-fraction through T2* correction leads to reduced bias, but at the cost of worse SNR performance. Cram r-Rao bound analysis1,9-12 showed that noise performance of the estimation methods depends on the sampling times (echo times) at which the MR signal is measured. Previous works used Cram r-Rao bound analysis for unbiased estimators, but for single T2* correction and without T2* correction the estimates will be biased, in general. I analyzed the tradeoff between bias and variance for different fat quantification methods using the Cram r-Rao bound analysis for biased estimators (CRBBE). I proposed a noise performance metric for fat-fraction estimation and the theoretically evaluated the metric using CRBBE at different echo combinations to derive the optimal echo combination for best noise performance in fat quantification. From figure 3, it can be seen that ?/2 and 2?/3 are the optimal echo shifts for best noise performance for single and dual T2* correction, respectively. 2?/3 is also the optimal echo spacing for methods without T2* correction. All the three methods demonstrate tremendous improvement in noise performance with very short first echo times. To my knowledge this is the first work that presented CRBBE for any multi-point chemical-shift based fat-water separation method.
In-vivo Assessment of Hepatic Fat-Fractions
I also analyzed the role of independent correction for T2* of water and fat in the quantification of hepatic fat in-vivo in an institutional review board approved cohort of 100 subjects. A comparison of hepatic fat-fraction estimates, segmented to avoid large vessels and biliary structures, from the single and dual T2* methods in a patient with severe steatosis are shown in figure 4.
In summary, I developed a novel chemical-shift based qMRI method that corrects for the independent T2* decay of water and fat to improve the accuracy of fat measurement. I experimentally and theoretically analyzed the method s optimality of for diagnosis and grading of fatty liver disease.
In this section, I will describe my future research interests in developing quantitative MR imaging biomarkers. Development of non-invasive imaging biomarkers will provide an unprecedented opportunity for pre-emptive intervention and prevention of diseases such as Alzheimer s, hemochromatosis and multiple sclerosis. I also plan to use the principles of estimation theory and post-processing methodologies to improve the accuracy and precision in the quantitative MRI measurements.
Quantification of Amyloidosis and Iron Content in the Brain of Patients with Alzheimer s Disease
Alzheimer s disease is estimated to be affecting approximately 5 million Americans and accounts for more than 60% of all dementia cases. Abnormal deposition of b-amyloid (Ab) protein and neurofibrillary tangles are the pathologic hallmarks of Alzheimer s disease. There is an increasing interest in understanding the role of iron in neurodegenerative diseases. The presence of Ab together with its associated iron deposition is expected to alter the biophysical environment in the brain. I aim to detect and quantify amyloidosis and brain iron content through measurement of surrogate biomarkers such as T2, T2* and T1. Noise analysis for simultaneous T1 and T2 mapping has been performed using numerical simulations for typical acquisition parameters. But, Cram r-Rao bound analysis to determine the optimal acquisition parameters for simultaneous T1 and T2 quantification has not been performed. Improving the noise performance in quantitative measurement of iron content in the brain using results from Cram r-Rao bound analysis would not only help to detect and quantitatively grade diseases such as Alzheimer s and Parkinson s but would also be useful in better assessment of cerebral microbleeds in stroke patients. I will validate the theoretical results using phantom experiments and will collaborate with Radiologists to translate the methods to clinical practice.
Improving Spatial Resolution and SNR for Diffusion Weighted Imaging and MR Angiography
Partial volume effects arising from low resolution acquisitions are one of the confounding factors for accurate quantitative MRI measurements. In certain applications such as diffusion-weighted imaging (DWI), functional BOLD MRI (fMRI), and proton-density or T2-weighted imaging, true three-dimensional (3-D) image acquisitions are not always possible and therefore acquiring a set of two-dimensional (2-D) slices is a general practice. A limiting factor of the minimum slice thickness for 2-D scans is the SNR. Thicker slices have higher SNR; however, they introduce partial volume effects (PVE) when two tissues interface within a single voxel. In time-of-flight (TOF) MRA, thin 2-D slices may be desirable to minimize saturation of slow or in-volume flow; however, that must be weighted against the higher SNR of 3-D TOF methods. I plan to develop post-acquisition image reconstruction methods that use separate interleaved slice acquisitions to reduce partial volume effects in DWI, MRA and in the quantitative MRI measurements of myelin water fraction, T2, T2* and T1.
Measurement of Ferritin and Hemosiderin Iron in Patients with Hemochromatosis and Steatosis
Hemochromatosis afflicts more than 1.5 million Americans and also occurs frequently in the course of thalassemia major. Hemochromatosis and/or multiple transfusions of red cells cause iron over load in organs such as liver and heart. Excess iron could cause oxidative stress leading to cell death and tissue damage. Ferritin and hemosiderin are the short-term (distributed) and long-term (aggregate) storage forms of iron. Distributed and aggregate forms of iron were shown to have different effects on MR signal decay. The differential effects of these different forms of iron were quantified in liver without steatosis. Fat and iron often exist concomitantly in hepatic diseases. I recently analyzed1 the effect of iron overload from transfusional hemosiderosis on the R2* of liver with steatosis (figure 5). I plan to extent this work to analyze the differential effects of ferritin and hemosiderin on the T2* of liver in the presence of fat to provide greater pathological specificity and to quantitatively assess the efficacy of the iron-chelating regimens that are used in the treatment of hemochromatosis and thalassemia major.
Assessment of Myelin and Ferritin Distributions in the Brain of Patients with Multiple Sclerosis
Multiple sclerosis is a chronic demyelinating disease of the central nervous system that afflicts approximately 400,000 Americans and 2.5 million individuals worldwide13. Recent high-resolution, high field MRI studies have shown myelin and iron (ferritin) to be important contributors to the significant contrast variations observed in the brain. Laminar variation of myelin and ferritin content in the cortical gray matter was also observed14. I plan to use my doctoral research experience in multicomponent T2* analysis to quantitative measure the individual contributions of iron and myelin to the MR contrast seen in the brain. This analysis would be useful in analyzing the anatomical relationship between myelin and ferritin distributions across the brain. Understanding the role of ferritin in myelination would be very valuable for developing effective therapies for demyelinating disorders such as multiple sclerosis.
I developed and validated a novel non-invasive quantitative MR imaging biomarker for diagnosis and staging of non-alcoholic fatty liver disease (NAFLD). Using my unique research experience1-12, I aim to develop novel accurate and precise MR imaging biomarkers for neurodegenerative diseases and diseases related to iron overload and insulin resistance. This research has tremendous potential for receiving funding support from Alzheimer s Disease Neuroimaging Initiative (ADNI), National Institute on Aging, National Institute of Biomedical Imaging and Bioengineering (NIBIB), National Multiple Sclerosis Society and the Foundation for the National Institutes of Health. Development of quantitative magnetic resonance imaging biomarkers will greatly benefit the public by accelerating the detection and treatment of a range of diseases.
1. Chebrolu VV, Bice E, Yu H, et al. On the Performance of T2* Correction Methods for Quantification of Hepatic Fat Content over Clinically Relevant Fat-Fractions. Magn Reson Med 2010;in preparation.
2. Chebrolu V, Shahani P; A method and system for separating water and fat components in a magnetic resonance Image patent IPCOM000147728D. 2007 Mar 23, 2007.
3. Brodsky E, Chebrolu V, Block W, Reeder S. Frequency Response of Multipoint Chemical Shift Based Spectral Decomposition. Journal of Magnetic Resonance Imaging 2010;in press.
4. Johnson K, Chebrolu V, Reeder S. Absolute Temperature imaging with Non-Linear Fat/Water Signal Fitting. Proceedings 16th Scientific Meeting, International Society for Magnetic Resonance in Medicine. Toronto, Canada; 2008. p 1236.
5. Chebrolu V, Yu H, Brodsky E, McKenzie C, Reeder S. Impact of T2* Decay on the Quantification of Hepatic Steatosis with MRI. Proceedings 16th Scientific Meeting, International Society for Magnetic Resonance in Medicine. Toronto, Canada; 2008. p 1585.
6. Hines C, Yu H, Shimakawa A, et al. Validation of Fat Quantification with T2* Correction and Accurate Spectral Modeling in a Novel Fat-Water-Iron Phantom. Proceedings 17th Scientific Meeting, International Society for Magnetic Resonance in Medicine. Honolulu, Hawaii; 2009. p 2707.
7. Chebrolu V, Hines C, Yu H, et al. Independent Estimation of T2* for Water and Fat for Improved Accuracy of Fat Quantification. Proceedings 17th Scientific Meeting, International Society for Magnetic Resonance in Medicine. Honolulu, Hawaii; 2009. p 2847.
8. Chebrolu VV, Hines CD, Yu H, et al. Independent estimation of T2* for water and fat for improved accuracy of fat quantification. Magn Reson Med 2010;63(4):849-857.
9. Chebrolu VV, Yu H, Pineda AR, McKenzie CA, Brittain JH, Reeder SB. Noise analysis for 3-point chemical shift-based water-fat separation with spectral modeling of fat. J Magn Reson Imaging 2010;32(2):493-500.
10. Chebrolu V, Yu H, Pineda A, McKenzie C, Brittain J, Reeder S. Noise Analysis for 3-pt Chemical Shift Based Water-Fat Separation with Accurate Spectral Modeling. Proceedings 17th Scientific Meeting, International Society for Magnetic Resonance in Medicine. Honolulu, Hawaii; 2009. p 2681.
11. Chebrolu V, Yu H, Pineda A, McKenzie C, Brittain J, Reeder S. Noise Analysis for Chemical Shift Based Water-Fat Separation with Independent T2* Correction for Water and Fat. ISMRM-ESMRMB Joint Annual Meeting Stockholm, Sweden.; 2010. p 2681.
12. Bice E, Chebrolu V, Yu H, Brittain J, Reeder S, Pineda A. Cramer-Rao Bound for Estimating Non-linear Parameters in a Model for Chemical Species Separation using Magnetic Resonance Imaging. 116th Annual Meeting of the American Mathematical Society. San Francisco, California; 2010. p 1056.
13. Oke SL, Tracey KJ. The Inflammatory Reflex and the Role of Complementary and Alternative Medical Therapies. Annals of the New York Academy of Sciences 2009;1172(1):172-180.
14. Fukunaga M, Li T-Q, van Gelderen P, et al. Layer-specific variation of iron content in cerebral cortex as a source of MRI contrast. Proceedings of the National Academy of Sciences 2010;107(8):3834-3839.