Hardware for Performing Hyperpolarized Helium Imaging on a Clinical MR Imager

Hardware for Performing Hyperpolarized Helium Imaging on a Clinical MR Imager by Angela C. Tooker Submitted to the Department of Electrical Engineering and Computer Science in Partial Fulfillment of the Requirements for the Degrees of Bachelor of Science in Electrical Engineering and Computer Science and Master of Engineering in Electrical Engineering and Computer Science at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY January 24, 2002 MAsSACHUSETTS INSTITUTE OF TECHNOLOGY 2002 Angela C. Tooker. All rights reserved. The author hereby grants to M.I.T. permission to reproduce and distribute publicly paper and electronic copies of this thesis and to grant others the right to do so. JUL 3 1 2002 LIBRARIES Author Department of Electrical Engineering and Computer Science January 24, 2002 Certified b3 Mitchell Albert Thesis Supervisor Certified by_ Frederick Bowman Thesis Supervisor Certified by Chairman, Department Committee on Gn,. Smith Ae Theses

Hardware for Performing Hyperpolarized Helium Imaging on a Clinical MR Imager by Angela Tooker Submitted to the Department of Electrical Engineering and Computer Science January 24, 2002 In Partial Fulfillment of the Requirements for the Degrees of Bachelor of Science in Electrical Engineering and Computer Science and Master of Engineering in Electrical Engineering and Computer Science Abstract For the past forty years, magnetic resonance imaging (MRI) has been one of the most common methods for obtaining volume images of human organs. However, some organs in the human body, such as the lungs, are resistant to these techniques, due to their lack of protons. In 1994, hyperpolarized noble gas MRI was introduced, and today is viewed as the best solution to the MR imaging dilemma. The different resonance frequencies of proton and helium, however, present an obstacle for hyperpolarized helium imaging. Conventional MR scanners are proton-based and, thus, are incapable of transmitting and receiving at frequencies other than that of proton. Broadband MR scanners, which are capable of transmitting and receiving at a variety of frequencies, are expensive and are not readily available for all nuclei. A heterodyne system is an inexpensive system that can be added to conventional MR systems to enable them to image a variety of nuclei. This system transforms the output of the MR system, at the frequency of proton, to the frequency of the nuclei of interest. The resulting signal obtained from the subject is transformed back to the frequency of proton by the heterodyne system, thereby enabling the MR system to create an image of the subject. With this system, hyperpolarized helium images of human lungs were obtained using a variety of different protocols. This heterodyne system opens up the possibility of new types of imaging, based on nuclei other than proton, on conventional MR scanners. Thesis Supervisor: Mitchell Albert Title: Project Supervisor Thesis Supervisor: Frederick Bowman Title: MIT Thesis Supervisor 2

Acknowledgements I would like to thank my thesis supervisor at Brigham and Women's Hospital, Dr. Mitch Albert, who was always there to advise me and provide support. Special thanks to Professor Bob Lenkinski for providing the basic design for the heterodyne system, and for helping in the inital setup and testing of the system. Thanks to my lab-mates, Arvind Venkatesh and Adelaide Zhang, who taught me so much and helped during the initial setup and testing of the heterodyne system. Many thanks to Ken Hickson, the GE Field Engineer, for helping me debug the gating problem and for his continued help with the intricacies of making our equipment talk to the Signa equipment. Many thanks to Ralph Hashoian who taught me so much about RF coils and mixers, and who always had wonderful suggestions for solving the problems with our coils, our T/R switch and preamplifier, and the Signa. I would like to thank my MIT thesis supervisor, Dr. Frederick Bowman, for his support and help with the seemingly endless paperwork. Thanks to Dr. Kwan Soo Hong, who always performed the experiments with me. Special thanks to Erin McKinstry for her help with the experiments and for her suggestions regarding my thesis. Finally, many thanks to my parents for all their support during my years at MIT and for helping to proofread my thesis. 3

Contents 1 Introduction...8 2 Magnetic Resonance Theory... 10 2.1 Atoms and Magnetic Dipole Moments... 10 2.2 Precession and the Resonance (Larmor) Frequency... 13 2.3 Bulk Magnetization...14 2.4 Radio Frequency Coils and Pulses... 16 2.5 Magnetic Resonance Imaging and Gradients...18 3 Hyperpolarized Noble Gas MRI...20 3.1 Introduction.... 20 3.2 Method for Hyperpolarizing 3 He... 21 3.3 Advantages of Hyperpolarized Noble Gas MRI... 21 3.4 Disadvantages of Hyperpolarized Noble Gas MRI...22 4 Heterodyne System Design...24 4.1 B ackground... 24 4.2 Overview of the Heterodyne System... 25 4.3 Transmit Channel... 26 4.4 Radio Frequency Amplifier... 29 4.5 Transmit/Receive Switch, Pre-Amplifier, and Coil...29 4.6 R eceive C hannel... 31 5 Heterodyne System Results and Discussion...33 5.1 Hyperpolarized 3 He Human Lung MR Images...33 5.2 Voltage Lost Considerations... 35 5.3 G radient R eversal... 38 5.4 Gating and the Radio Frequency Amplifier... 40 4

6 C onclusions and Future W ork... 42 6.1 Conclusions... 42 6.2 Recom m endations for Future W ork... 43 R eferences...-......-44 5

List of Figures Figure 2.1 A spinning charged nucleus creates a magnetic field, similar to that created by a bar magnet. Figure 2.2 The direction of the magnetic field depends directly on the direction in which the nucleus is spinning. Figure 2.3 (A) The magnetic fields for paired protons cancel, resulting in no net magnetic field. (B) Unpaired protons, on the other hand, create a net magnetic field. Figure 2.4 With no external magnetic field, the magnetic dipole moments from individual nuclei cancel, resulting in a zero net magnetic field. Figure 2.5 In the presence of an external magnetic field, B 0, the axes of the magnetic dipoles align with the field, creating net magnetization. Figure 2.6 (A) With no external magnetic field, the nucleus spins about its dipole axis. (B) In the presence of an external magnetic field, the nucleus spins about its dipole axis and precesses about the external magnetic field axis. Figure 2.7 Motion of the magnetization vector, after a RF pulse has been applied, in both (A) stationary and (B) rotating frames of reference. Figure 2.8 Decaying sine wave of the received signal (FID). Figure 2.9 When a linear gradient is applied to a body, the effective applied magnetic field at different points in the body change. This, in turn, changes the resonance frequency of those points. Figure 4.1 Overview of the heterodyne system (shaded in gray) and the connections to the MR system, RF Amplifier, T/R switch, pre-amplifier, and coil. Figure 4.2 The transmit channel of the heterodyne system. Figure 4.3 The receive channel of the heterodyne system. Figure 5.1 Proton MR image of the lungs in the coronal plane. 6

Figure 5.2 3 He image of the lungs in the coronal plane, obtained using the heterodyne system. Figure 5.3 3He image of the lungs in the axial plane, obtained using the heterodyne system. Figure 5.4 3He image of the lung airways obtained using the heterodyne system. Figure 5.5 Voltage output by the MR system and the transmit channel of the heterodyne system. Figure 5.6 Proton frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. Figure 5.7 Helium frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. Notice the reversal of the applied linear gradient. Figure 5.8 Proton frequencies, and their corresponding helium frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. Notice, each proton slice corresponds to the opposite slice for helium. 7

Chapter 1 Introduction Conventional magnetic resonance imaging (MRI) techniques are proton based. By exciting the protons in water and measuring the resulting output signals, images of a variety of structures within the body can be obtained. However, these techniques provide little useful information about the lungs because the inhomogeneous air-tissue interfaces in the lungs make it difficult to acquire the signal fast enough. Consequently, other techniques are needed to make MR images of the lungs. One of the most promising techniques involves the use of hyperpolarized noble gas [1,2]. After the hyperpolarization process, the 3He or 1 29 Xe has a signal 105 times greater than its equilibrium signal, making it easily detectable by MRI. Using a specialized MR scanner, hyperpolarized helium MR images of the lungs that are considerably more detailed than conventional proton images can be obtained. Hyperpolarized helium images can be used to discern information about the lungs for possible diagnoses, tracking the time-course of pulmonary diseases, such as cystic fibrosis, asthma, and emphysema [2-6], and staging the effectiveness of various treatments. 8

Proton and helium have different resonance frequencies. Conventional MR systems are only able to transmit and receive at the resonance frequency necessary to excite protons and, hence, cannot be used to excite other nuclei, such as helium. There are specialized MR systems that can be used to image nuclei other than proton. Their high cost, however, prevents most institutions from obtaining these systems, and for some nuclei, these specialized MR systems are not available. The goal for this project is to develop an inexpensive heterodyne system that can be added to an MR system, enabling it to transmit and receive at a variety of frequencies. This system can be used to image a variety of nuclei on any conventional MR system. The particular application of interest for this project is hyperpolarized helium MR of the lungs. With the heterodyne system, conventional MR scanners can be used to obtain hyperpolaried helium MR images of the lungs, showing considerable detail from the lung periphery and airway structures. 9

Chapter 2 Magnetic Resonance Theory 2.1 Atoms and Magnetic Dipole Moments The nucleus of an atom, comprised of protons and neutrons, has a positive charge. Each proton within the nucleus spins, creating an electromagnetic field, like a typical bar magnet (Figure 2.1) [7-11]. -- Greater magnetic field <- Spinning nucleus with charge Bar Magnet Figure 2.1 A spinning charged nucleus creates a magnetic field, similar to that created by a bar magnet. The hydrogen nucleus, containing a single proton, has a spin quantum number of '/2. Hence, there are two possible energy states, denoted -2 and + 2, and two possible directions in which the protons may spin. Due to the different spinning directions 10

possible for the protons, magnetic fields are created in opposite directions (Figure 2.2), which, by analogy to the bar magnet, are often referred to as north and south. Direction of magnetic field Direction of spin Direction of magnetic field Figure 2.2 The direction of the magnetic field depends directly on the direction in which the nucleus is spinning. If there are even number of protons in the nucleus, then each proton with a magnetic field pointing north has a paired proton with a magnetic field pointing south. These anti-parallel magnetic fields cancel out, creating a zero net magnetic field (Figure 2.3A). On the other hand, if there are an odd number of protons, there will be one unpaired proton. This unpaired proton, pointing either north or south, creates a non-zero net magnetic field, or a magnetic dipole moment (MDM) represented by the vector p (Figure 2.3B). (No magnetic field) (Net magnetic field) Paired protons Unpaired protons A Figure 2.3 (A) The magnetic fields for paired protons cancel, resulting in no net magnetic field. (B) Unpaired protons, on the other hand, create a net magnetic field. B 11

Each individual proton, in a collection of protons, spins about its own axis and has its own magnetic field, or magnetic dipole moment (MDM) [7-11]. Each axis points in a random direction and the resulting sum of the individual MDMs is zero (i.e. there is no net magnetic field) (Figure 2.4). Bo off: :Net MF= 0 Figure 2.4 With no external magnetic field, the magnetic dipole moments from individual nuclei cancel, resulting in a zero net magnetic field. If this collection of protons is placed in an external magnetic field, B 0, the axes of the individual MDMs will align along the direction of B 0. Some will point in the same (parallel) direction as h 0 and some will point in the opposite (anti-parallel) direction (Figure 2.5). Bo on: 4+4+44 Figure 2.5 In the presence of an external magnetic field, B 0, the axes of the magnetic dipoles align with the field, creating net magnetization. Initially, exactly half of the MDMs are in the same direction as B 0 and half are in the opposite direction. Over time, however, the number of MDMs pointing in the same direction as B. increases, creating a net magnetization, M, in the direction of B 0. The net magnetization grows exponentially with time: 12

M = N(H)(1 - e~"i,) (Eqn. 2.1) where T is the relaxation time and N(H) is the density of mobile protons in the sample. Eventually, all of the dipole axes of the protons will be parallel to B 0. 2.2 Precession and the Resonance (Larmor) Frequency If a proton is placed in an external magnetic field, its own small MDM causes it to both spin about its own MDM axis and precess about the axis of the external magnetic field (Figure 2.6B) [7-11]. Bo OFF Bo ON A B Figure 2.6 (A) With no external magnetic field, the nucleus spins about its dipole axis. (B) In the presence of an external magnetic field, the nucleus spins about its dipole axis and precesses about the external magnetic field axis. The angular frequency of this precession is called the Larmor, or resonance, frequency (co); it is proportional to both the magnitude of the external magnetic field (B 0 ) and the gyromagnetic ratio (y ) of the nucleus co = B 0 y. (Eqn. 2.2) 13

Since the gyromagnetic ratio is element specific, different atoms will have different resonance frequencies even when the magnetic field strengths are the same. For example, the gyromagnetic ratio of hydrogen is y, = 42.6MHz / Tesla, whereas the gyromagnetic ratio of helium is Y He = 32.4MHz / Tesla. Therefore, at a magnetic field strength of BO = 1.5Tesla, the resonance frequency of hydrogen is O H= 63.87MHz and the resonance frequency of helium is coh,4 = 48.65MHz. 2.3 Bulk Magnetization Due to its natural abundance in the human body, hydrogen is the basis of the signal in conventional magnetic resonance imaging (MRI). When a human body is placed in an external magnetic field (B 0 ), each individual proton within the sample aligns its MDM (i ) axis either parallel or anti-parallel to the external magnetic field (see Figure 2.5) [7-11]. Hence, a magnetization vector M can be defined which is the sum of the individual MDMs within the sample: NS AI = I I, (Eqn. 2.3) n2=1 where NS is the total number of spinning protons in the body and i, is the MDM for the nth proton. From quantum theory, there are two possible spin states for proton. The Boltzmann relationship relates the population difference to the energy difference in these two spin states: N N = exp,e (Eqn. 2.4) KT, 14

where N, is the number of spins pointing upward, NI is the number of spins pointing downward, AE is the energy difference between the two spin states, K is the Boltzmann constant (1.38 x 10-23 J/K), and T, is the absolute temperature of the body. Since in practice, AE << KT (Eqn. 2.5) the ratio can be rewritten N I ~1+ A KT (Eqn. 2.6) where y is the gyromagnetic ratio, h is Planck's constant (6.6 x 10 34 J-s), and B 0 is the magnitude of the external magnetic field. Using the relations NS= N +NU NS,~ 2N-2Nt ( qn.. ) where NS is the total number of spins, the above ratio in Eqn. 2.6 can be rewritten as NT- N NhB 2kTS (Eqn. 2.8) This indicates there is a small excess of spins in the upward direction. This difference occurs since a spin will more likely be in the lower-energy state (with an upward pointing spin) than in the higher-energy state (with a downward pointing spin). This small population difference results in a non-zero magnetization vector, M. The magnitude of the magnetization vector is related to this population difference: (NT - NJy 2 (Eqn. 2.9) Hence, the magnitude of the bulk magnetization is 15

2 h 2 B 0 N~ N ( E q n. 2.1 0 ) 4KI:, _ = 2 2 so Although the individual protons precess around the external magnetic field,, B, the bulk magnetization vector, M, does not precess around BO. The individual protons are out of phase with each other and, hence, when their individual MDMs are added together, there is a large component in the direction of BO, but the phase differences cancel each other. Therefore, the bulk magnetization vector does not precess. 2.4 Radio-Frequency Pulses and Coils A radio-frequency (RF) pulse is an electromagnetic wave that is applied to the sample via a RF coil. Coils are electrical devices composed of multiple loops of wires that can generate and/or detect magnetic fields [7-12]. Typically RF pulses of high intensity and short duration are applied to the body at the resonance frequency of proton, in a direction perpendicular to the external magnetic field. The RF pulse causes the bulk magnetization vector to tip away from the external magnetic field and into the plane containing the rotating RF pulse. The magnetization vector tips because the protons with a resonance frequency equal to that of the RF pulse absorb the energy in the RF pulse and are, thus, excited to a higher energy state. (Nuclei with different resonance frequencies do not absorb any energy and, therefore, are not excited.) While it is tipping away from equilibrium, and as it is relaxing back (once the RF pulse is turned off), the magnetization vector precesses around the external magnetic field, B 0, at the resonance frequency. When viewed from a stationary frame of reference, the protons precess around the 16

external magnetic field axis at the resonance frequency. If, instead, the spinning protons were viewed in a frame of reference rotating at the resonance frequency, they appear to be stationary. Hence, when viewing the precessing magnetization vector in the stationary frame of reference, the magnetization vector traces out a spiral (Figure 2.7A). On the other hand, if the precessing magnetization vector is viewed in the rotating frame of reference, the magnetization vector traces an arc into the plane containing the RF pulse (Figure 2.7B). Z Spiral observed from a point outside the coordinate system Z Simple arc visualized from a rotating reference point x X RF A B Figure 2.7 Motion of the magnetization vector, after a RF pulse has been applied, in both (A) stationary and (B) rotating frames of reference. This rotating magnetization is an oscillating magnetic field that can be detected by the coil. The signal detected is a damped oscillating sine wave called the free induction decay (FID) signal (Figure 2.8). Received signal Time Figure 2.8 Decaying sine wave of the received signal (FID). 17

2.5 Magnetic Resonance Imaging and Gradients To create a magnetic resonance image (MRI), gradients are used to obtain information about the volumes of interest [7,8,10]. If no gradients are applied, the entire body would see the same magnetic field, B 0. Hence, the FID received by the coil as a result of the RF pulse would be from the entire body. Gradients are used to vary the external magnetic field applied to the subject, thereby changing the resonance frequencies of different slices within the body. Consider a body in an external magnetic field, B 0, shown in Figure 2.9. 1.55T- - 1.5T=Bo Figure 2.9 When a linear gradient is applied to a body, the effective applied magnetic field at different points in the body change. This, in turn, changes the resonance frequency of those points. To define various slices within the sample, a linear gradient is applied. As shown in Figure 2.9, this applied gradient changes the effective magnetic field seen at each point in the body. Since the resonance frequency varies with the applied magnetic field, this causes a change in the resonance frequency at each point. Therefore, if a RF pulse is applied at a particular resonance frequency, only one corresponding line of the body is excited. By applying a RF pulse at a range of frequencies (a bandwidth), many lines within the body, i.e. a "slice," can be excited. As the bandwidth increases, a larger slice 18

within the body will be excited by the RF pulse. To excite other slices, RF pulses at different frequencies need to be applied. Once the gradient has been applied, any FID resulting from the application of an RF pulse with a specified bandwidth will contain information from only one slice of the body. By applying gradients in varying configurations, using all three spatial directions, specific volumes within the sample can be specified. Any signal obtained after the RF pulse will be due only to the nuclei in the specific region being excited. The strength of the received signal is directly proportional to the number of protons excited in the specific volume. An MR image of the body can be obtained using various transforms once the positions and number of the protons are known. 19

Chapter 3 Hyperpolarized Noble Gas MRI 3.1 Introduction The field of hyperpolarized noble gas MRI is rapidly growing. One limitation of conventional proton-based MRI is that regions lacking water protons, such as the lungs or the lipid bilayer membranes of the brain, cannot be imaged. Other techniques are needed; hyperpolarized noble gas MRI is one technique that can be used to image these inhomogeneous regions of the lungs [2,13]. Hyperpolarized 3He images of the lungs, in both humans, rats, dogs, and pigs, have already been obtained. In addition, hyperpolarized 129Xe images have been obtained of the lung and brain structures in rats and humans. Current hyperpolarized noble gas MRI requires either the inhalation of the gas by the subject or the injection of the gas in the form of microspheres [2,13,14]. The MR image is then created from the signal obtained after applying a pulse at the proper resonance frequency. Thermal equilibrium polarization of noble gases is only about 0.04%, which is insufficient for imaging. However, using a laser optical pumping 20

process, the polarization can be increased by a factor of 105, making MR imaging of noble gases possible in the human body. 3.2 Method for Hyperpolarizing 3He A laser optical pumping process is used to hyperpolarize 3 He. In this method, the helium nuclei undergo a spin exchange with optically pumped rubidium (Rb) atoms [2,13,15-17]. Rb is an alkali metal that vaporizes at 85 'C. When vaporized and placed in an external magnetic field, Rb splits into different electron spin states under the excitation of a diode laser. Ground state electrons in Rb atoms can be excited by absorbing photons of light with a wavelength of 794.7 nm. If these polarized Rb atoms are then brought together with 3 He atoms, the collisions between the atoms will result in the transfer of angular momentum from Rb valence electrons to the 3 He nuclei. This spin exchange process increases the noble gas spin population by about 25% over the equilibrium state and, therefore, enhances the resulting MR signal by up to 105 times its thermal equilibrium value, making MR imaging possible. A similar process can be used to hyperpolarize other noble gases, such as 129 Xe. To achieve a polarization of about 10-20% requires approximately seven hours for 'He. If the hyperpolarized noble gas is cryogenically cooled and kept in a magnetic field, the polarization can last for hours, even days. 3.3 Advantages of Hyperpolarized Noble Gas MRI There are several advantages to using hyperpolarized noble gases for MRI. The MR signal from hyperpolarized noble gases can be up to ten times larger than the signal 21

from a corresponding volume of protons. Hence, a new range of exploitable responses and contrast, as well as increases in space and time resolutions, can be obtained [1,2,13,15,16]. In addition, since the polarization process in hyperpolarized MRI is independent of magnetic field strength, less expensive, low-field (< 0.1 Tesla) magnets can be used. Currently, people with metal implants, such as aneurysm clips or pacemakers, cannot undergo MRI. With low-field magnets, it might be possible for these people to avail themselves of this technology [18]. Another advantage of hyperpolarized noble gas MRI is that it is possible to visualize biological systems invisible to proton-based MRI techniques. For example, conventional MRI detects the signal from the water proton molecules in the body. However, the lung-space is a nearly water-free environment. As such, details about the lungs are not visible in MR images. Hyperpolarized helium and xenon, however, can be used to discern information about the lungs in MR images [2,13], for possible diagnoses, tracking the time-course of pulmonary diseases, such as cystic fibrosis, asthma, and emphysema, and staging the effectiveness of various treatments [2-6]. Likewise, the lipid-filled regions of the brain cannot be imaged with conventional MRI. Hyperpolarized xenon, which rapidly dissolves in lipids, can be used to image these regions of the brain [1]. Such information could be used in the diagnosis, treatment, and staging of various white-matter diseases, such as multiple sclerosis. 3.4 Disadvantages of Hyperpolarized Noble Gas MRI Despite these promising advantages, there are two disadvantages for hyperpolarized noble gas MRI. First, hyperpolarizing helium or xenon is a time- 22

consuming, costly procedure. The hyperpolarization process can take upwards of seven hours and there is a limit to the amount of gas that can be hyperpolarized at one time. The second disadvantage for hyperpolarized noble gas MRI is that only specialized MR systems can be used, due to the differences in the resonance frequencies of proton and 3 He and 129Xe. The conventional MR systems that are routinely used are incapable of transmitting and/or receiving at a frequency other than that of proton. Only some MR systems, equipped with specialized hardware can transmit and receive at a variety of different frequencies. The large cost or lack of availability prevents most institutions from adding to their conventional MR systems the specialized hardware necessary for performing MR imaging of other nuclei. Thus, before hyperpolarized noble gas MRI can be used routinely, an inexpensive method for imaging other nuclei on conventional MR scanners must be found. The heterodyne system is an inexpensive means that enables conventional MR scanners to image other nuclei, such as helium. 23

Chapter 4 Heterodyne System Design 4.1 Background Conventional MR uses the hydrogen proton as the nucleus for imaging. However, it is possible to use nuclei other than proton, such as atomic elements with intrinsic spins, like the noble gases, helium and xenon, for MR imaging. In order to image using these nuclei, specialized MR systems with broadband hardware are required. MR systems, with broadband hardware, are capable of transmitting and receiving a variety of frequencies and are used to image different nuclei. In contrast, conventional MR systems have narrowband hardware and are only capable of transmitting and receiving at the frequency of proton; thus, these systems cannot be used for imaging other nuclei. Specialized MR systems with broadband hardware are very expensive. Thus, most institutions have narrowband MR systems. Given that most MR scanners in both clinical and research environments are narrowband and that imaging of other nuclei, such as helium and xenon, is rapidly growing, an important question is how can conventional MR scanners be inexpensively converted to allow imaging of these other nuclei? 24

An inexpensive solution to this problem is the heterodyne system [19]. A heterodyne system can be added to conventional MR systems allowing imaging to be performed on a variety of different nuclei, including hyperpolarized helium. 4.2 Overview of the Heterodyne System There are two main components to the heterodyne system: the transmit channel and the receive channel. (The heterodyne system was designed and tested for hyperpolarized helium MRI at 1.5 T. With only minor modifications, however, this system can be used for a variety of different nuclei at different magnetic field strengths.) The primary task of the transmit channel is to convert the signal transmitted by the MR system from 63.87 MHz (the resonance frequency for proton at 1.5 T) to 48.65 MHz (the resonance frequency for helium at 1.5 T). For the receive channel, the primary task is to convert the 48.65 MHz helium signal received from the coil to the 63.87 MHz proton signal that is detectable by the MR system. Figure 4.1 contains a diagram of the heterodyne system (shaded in gray) and its connection to the MR system. The Radio- Frequency (RF) amplifier, Transmit/Receive (T/R) switch, pre-amplifier, and coil are not part of the heterodyne system; they are required, however, for imaging of all nuclei on all MR systems. 25

Broadband RF Amplifier MR SYSTEM + T/R Switch COIL FPre-Amplifier Figure 4.1 Overview of the heterodyne system (shaded in gray) and the connections to the MR system, RF Amplifier, T/R switch, pre-amplifier, and coil. 4.3 Transmit Channel The main goal of the transmit channel is to convert the 63.87 MHz proton signal transmitted by the MR system to the 48.65 MHz signal needed to excite the helium nuclei. A frequency mixer and a filter can be combined together to convert the original proton signal. (See Figure 4.2 for a diagram of the transmit channel.) The frequency mixer has two inputs and outputs their convolution. The two inputs to the mixer are: 1) the 63.87 MHz signal output by the MR system and 2) the local oscillator, which is a pure sine wave at 112.52 MHz. The local oscillator is a Hewlett-Packard (Palo Alto, California) 8648B Signal Generator. The voltage output by the local oscillator is 300 mvrms. The frequency mixer is a Level 7 ZP-3 double-balanced mixer from Mini- Circuits (Brooklyn, New York). The resulting output of the frequency mixer is the convolution of the two input signals, i.e. the sum of two sine waves one at 112.52MHz - 63.87MHz = 48.65MHz (Eqn. 4.1) 26

and one at 112.52MHz + 63.87MHz = 176.39MHz. (Eqn. 4.2) Thus, by changing the frequency of the local oscillator, the sine waves output by the frequency mixer changes accordingly. The heterodyne system, therefore, can be used to image other nuclei, by changing the frequency of the local oscillator. Mini-Circuits ZP-3 Level 7 Frequency Mixer Transmitted Proton Signal (63.87 MHz) from MR System Local Oscillator (112.52 MHz) Hewlett-Packard 8648B Signal Generator Mini-Circuits BLP-90 Low-Pass Filter Transmitted Helium Signal (48.65 MHz) to RF Amplifier Figure 4.2 The transmit channel of the heterodyne system. The signal output by the MR system is not continuous; rather, it is a pulsed signal. The frequency mixer does not require a continuous signal input, and outputs a signal only when receiving input. Hence, the output of the mixer has the same pulsed shape as the original, only the frequency of oscillation changes. The same local oscillator is used in both the transmit and receive channels. It is important that the local oscillator used in both channels has the same frequency. To eliminate the need for another signal generator, a power splitter is used to create the 27

signals needed for both channels. A power splitter is used to split the output of the local oscillator into two identical signals, each with the same frequency as the original and approximately half the amplitude of the original. (The power splitter is a ZFSC-2-2 2- Way-0 0 Power Splitter/Combiner from Mini-Circuits.) If, instead, the local oscillator outputs a sine wave with a frequency of 15.23 MHz, the frequency mixer would output the sum of two sine waves: one at 48.65 MHz and the other at 79.09 MHz. Either local oscillator frequency, 112.52 MHz or 15.23 MHz, could be used in the heterodyne system, since both produce the desired resonance frequency of helium, 48.65 MHz. The local oscillator frequency is set to 112.52 MHz to enable better filtering. To avoid damaging the coil and other electrical components, as well as for the safety of the patient, it is necessary that only the desired frequency, 48.65 MHz, leave the heterodyne system. The other frequency output by the frequency mixer must be eliminated. The frequency difference between the two sine waves is 127.74 MHz for the local oscillator frequency of 112.52 MHz and only 30.44 MHz for the local oscillator frequency of 15.23 MHz. It is easier to filter a frequency that is 127 MHz away from the desired frequency than a frequency that is only 30 MHz away from the desired frequency. Hence, the higher local oscillator frequency, 112.52 MHz, was chosen. A low-pass filter is used to filter the sine wave at 176.39 MHz, leaving only the desired 48.65 MHz signal. A BLP-90 Low-Pass Filter (Mini-Circuits) is used. This filter has a passband from DC to 81 MHz (i.e. only frequencies in the 0-81 MHz range pass through un-attenuated). Hence, upon leaving the filter, the sine wave at 176.39 MHz has negligible amplitude, while the sine wave at 48.65 MHz has no decrease in amplitude. (If other nuclei are used, the output of the frequency mixer will contain different frequencies 28

and, thus, a different filter will be required.) Therefore, with a frequency mixer and a filter, the transmit channel of the heterodyne system converts the proton signal output by the MR system to the desired signal at the frequency of helium. 4.4 Radio-Frequency Amplifier Once a signal of the desired frequency has been obtained by the transmit channel, it needs to be amplified before it is sent to the coil so that the pulse will be strong enough to excite the nuclei. Conventional MR systems use narrowband Radio-Frequency (RF) amplifiers. They are only capable of amplifying a signal at the frequency of proton and actually attenuate signals at other frequencies. Thus, a separate RF Amplifier is needed to amplify the signal at the frequency of helium. While a narrowband RF amplifier that amplifies the frequency of interest, in this case helium, can be used, a broadband RF amplifier is a better choice as it amplifies a wide range of frequencies and, thus, can be used for a variety of different nuclei. (Broadband RF amplifiers, while often more expensive than narrowband RF amplifiers, are much less expensive than specialized broadband MR systems.) Hence, with the broadband MRI-2000 Linear Pulse Amplifier (ENI) used in this system, the 48.65 MHz signal transmitted to the coil will be large enough to excite the helium nuclei. 4.5 Transmit/Receive Switch, Pre-Amplifier, and Coil A Transmit/Receive (T/R) switch is used to transmit the signal from the RF amplifier to the coil, as well as to receive the signal from the coil. The T/R switch must be tuned to operate at the frequency of interest, in this case, the frequency of helium, 29

48.65 MHz. The MR system adds a 15 V bias to the transmitted signal traveling from the RF amplifier to the T/R switch. This bias tells the T/R switch that the signal must be transmitted to the coil. The T/R switch removes the DC bias from the signal and then transmits it to the coil The coil must be tuned to operate at the frequency of interest, 48.65 MHz. The coil employed is a flexible, quadrature coil built by Midwest RF, LLC. The coil wraps around the torso of the subject so that hyperpolarized helium images of the lung can be obtained. This coil was specially built for lung imaging; however, any type of coil, for any application, could be used with the heterodyne system. The signal detected by the coil from the excited helium nuclei travels back to the T/R switch. Since this signal has no DC bias, the T/R switch recognizes it as a "received" signal and transmits it to the pre-amplifier. The received signals are generally very small, so the pre-amplifier is used to increase the gain of the signal. Like the T/R switch and the coil, the pre-amplifier must be tuned to the proper frequency, 48.65 MHz. The output of the pre-amplifier, at 48.65 MHz, is then sent to the receive channel of the heterodyne system. The T/R switch, pre-amplifier, and coil are not part of the heterodyne system, although they are required for hyperpolarized helium imaging, or imaging of any other nuclei, on all MR systems. 30

4.6 Receive Channel Mini-Circuits ZP-3 Level 7 Frequency Mixer Local Oscillator Received (112.52 MHz) Hewlett- Helium Signal Packard 8648B Signal (48.65 MHz) Generator from Coil Received Proton Signal (63.87 MHz) to MR System Figure 4.3 The receive channel of the heterodyne system. The receive channel is similar to the transmit channel (see Figure 4.3). The signal received from the coil is at the frequency of helium, 48.65 MHz. However, conventional MR systems are incapable of detecting this frequency. Thus, the receive channel must convert the 48.65 MHz helium signal received from the coil (via the T/R switch and preamplifier) back to the frequency of proton, 63.87 MHz. This is done in a manner similar to that for the transmit channel. Another ZP-3 Level 7 double-balanced frequency mixer from Mini-Circuits is used (along with the same local oscillator as in the transmit channel) to convert the received signal at 48.65 MHz to the frequency of proton, 63.87 MHz. The output of the mixer will be the sum of two sine waves: one at 112.52MHz - 48.65MHz = 63.87MHz (Eqn. 4.3) and one at 112.52MHz+ 48.65MHz =161.17MHz. (Eqn. 4.4) 31

Since the MR system is only capable of receiving signals at the frequency of proton, no external filter is necessary to filter the 161.17 MHz sine wave from the mixer output. Since only the frequency of oscillation of the received signal is changed, the MR system can render an image in the same manner as for proton. The receive channel has extra, protective circuitry, as shown in Figure 4.3. The T/R switch requires a 15 V bias in order to transmit the signal to the coil. The inductor and capacitors are used to prevent this DC bias from damaging the frequency mixer and the MR system, and to allow the oscillating received signal to reach the frequency mixer. With this circuitry, the receive channel of the heterodyne system is able to convert the received helium signal, at 48.65 MHz, to a signal at the frequency of proton, 63.87 MHz, that the MR system then transforms into an image. 32

Chapter 5 Heterodyne System Results and Discussion 5.1 Hyperpolarized "He Human Lung MR Images Conventional proton MR images of the lungs provide limited information. As shown in Figure 5.1, the lungs appear as a black void in proton MR images; no information regarding the lung periphery or airway structure is visible. (The white areas within the black voids of the lungs are the blood vessels in the lungs.) Figure 5.1 Proton MR image of the lungs in the coronal plane. In contrast, hyperpolarized 3 He MR images of the lungs show considerably more detail. Figure 5.2 shows a 3 He MR image of one slice of a subject's lungs obtained on the GE 33

Signa LX 1.5 T MRI system (GE Medical Systems, Milwaukee, WI), running Software Revision 8.4, equipped with the heterodyne system described in Chapter 4. With the attached heterodyne system, the pulse sequences and options available on the Signa MRI system can be used. Figure 5.2 3 He image of the lungs in the coronal plane, obtained using the heterodyne system. Figure 5.3 3 He image of the lungs in the axial plane, obtained using the heterodyne system. Figure 5.3 shows one slice of a 3He image of a subject's lungs in the axial plane, as opposed to the image in Figure 5.2, which is in the coronal plane. The heterodyne system does not limit the planes in which the images can be obtained. In the images shown in Figures 5.2 and 5.3, only the lung periphery is visible; in contrast, lung airways are also visible in the image shown in Figure 5.4. A pulse sequence different from the one used for the images in Figures 5.2 and 5.3 was used to obtain the image in Figure 5.4. Thus, the various pulse sequences and imaging options available on the MR scanner for conventional proton imaging can also be used with the heterodyne system. 34

Figure 5.4 3 He image of the lung airways obtained using the heterodyne system. As demonstrated with these images, the heterodyne system enables conventional MR scanners to image other nuclei. The pulse sequences and imaging options available on the MR scanner are available for use with the heterodyne system to perform a variety of different types of imaging for various clinical and experiment protocols. With hyperpolarized helium, these options can be used to visualize lung periphery and airway structures. 5.2 Voltage Loss Considerations If the power reaching the coil is not sufficient to excite the nuclei, no image will be obtained. Hence, it is important to minimize the power lost by the heterodyne system. Due to the signal conversions performed by the heterodyne system, there is the possibility for signal loss. The voltage output by both the MR system and the transmit channel was measured. The voltage output by the MR system can be altered in a variety of different ways; for these measurements, the voltage output is varied by changing the transmit gain 35

(TG), i.e. the power in the RF pulses transmitted to the coil. Larger TG values correspond to larger power outputs. As can be seen in Figure 5.5, the voltage output by the low-pass filter in the heterodyne system is significantly smaller than the voltage output by the MR system. For large inputs (i.e. large TG values), more than 50% of the input signal is lost. For smaller inputs, the signal loss is smaller; however, it is never zero. 300 Voltage Output E 200 20-+- MR System E 100 - + Transnit Channel 0 0 100 200 Figure 5.5 Voltage output by the MR system and the transmit channel of the heterodyne system. The voltage loss is due to the frequency mixer. The particular frequency mixers used in the heterodyne system are double-balanced mixers [12]. One of the characteristics of double-balanced mixers is a conversion loss, i.e. as the convolution of the input signals is performed, voltage is lost between the input and output signals. 50% of the voltage can be lost, depending on the input signal and the local oscillator. To ensure minimal conversion loss by the frequency mixers, the voltage from the input signal (the signal from the MR system) must be less than the voltage from the local oscillator. Thus, as the input voltage from the MR system increases (i.e. as the TG increases), the voltage lost by the frequency mixer increases (as seen in Figure 5.5). This implies that for a given input voltage from the MR system, less voltage will be lost 36

during conversion for larger local oscillator voltages. These measurements were repeated for other local oscillator voltages. For larger voltage outputs from the local oscillator, the voltage lost by the frequency mixer for a given input signal is less than for smaller local oscillator voltages. However, even for small input signals from the MR system and large local oscillator voltages, there is always some voltage loss. Although there is a voltage loss in the transmit channel of the heterodyne system, it is not detrimental to obtaining hyperpolarized 3 He images with the GE Signa MR system and the quadrature coil. One possible reason is, despite the loss in the transmit channel, the RF amplifier provides enough amplification of the signal so that when it is transmitted to the coil it is still sufficient to excite the nuclei. There are several possible methods for eliminating this voltage loss. Instead of using double-balanced frequency mixers in the heterodyne system, active frequency mixers can be used [12]. Active frequency mixers require a DC voltage input; with this extra input, the resulting conversion loss by the frequency mixer is negligible. Hence, there would be no voltage lost between the input and output signals of the active frequency mixer. Another possible solution is the addition of an amplification stage. This would consist of either 1) an amplifier, distinct from the RF amplifier, on the output of the low-pass filter that would amplify the output signal to offset any signal loss by the frequency mixer or 2) a different RF amplifier that would provide more amplification. While all three solutions would solve the voltage loss problem in the heterodyne system, the active frequency mixer is the best solution. The addition of an amplifier, either at the output of the low-pass filter or via a different RF amplifier, would result in an amplified signal, containing amplified noise. The active frequency mixer, in addition to being less 37

expensive than the amplifier, would provide amplification of the signal, without the simultaneous amplification of the noise. (Similar losses are seen in the receive channel, and the solutions outlined above can also be used to minimize these losses.) 5.3 Gradient Reversal As explained in Section 2.5, MR imaging requires the use of gradients. With these gradients, spatial information about the object being imaged can be obtained [7,8,10]. The gradients change the applied magnetic field for different slices within the object. These changes in magnetic field cause changes in the resonance frequency of the nuclei in those slices. Figure 5.6 shows an applied gradient in a 1.5 T external magnetic field and the resulting changes in the resonance frequencies of proton for the different slices. RF pulses are transmitted by the MR system at the different frequencies for each slice. Proton Frequencies for Different Slices Applied Linear Gradient 66.87 MHz 65.87 MHz 64.87 MHz 63.87 MHz External 62.87 MHz Magnetic Field BO = 1.5 T 61.87 MHz 60.87 MHz Figure 5.6 Proton frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. 38

The transmit channel in the heterodyne system converts the frequency for proton to the frequency for helium. Figure 5.7 shows the resulting helium frequencies output by the transmit channel for different slices described in Figure 5.6. As can be seen, this corresponds to a reversal of the applied gradient field. Helium Frequencies for Different Slices Applied Linear Gradient 112.52 MHz - 66.87 MHz = 45.65 MHz 112.52 MHz - 65.87 MHz = 46.65 MHz 112.52 MHz - 64.87 MHz = 47.65 MHz 112.52 MHz - 63.87 MHz = 48.65 MHz 112.52 MHz - 62.87 MHz = 49.65 MHz 112.52 MHz - 61.87 MHz = 50.65 MHz 112.52 MHz - 60.87 MHz = 51.65 MHz External Magnetic Field Bo = 1.5 T Figure 5.7 Helium frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. Notice the reversal of the applied linear gradient. Proton Frequencies for Different Slices Helium Frequencies for Corresponding Slices Applied Linear Gradient 66.87 MHz 65.87 MHz 64.87 MHz 63.87 MHz 62.87 MHz 61.87 MHz 60.87 MHz X 45.65 MHz t 46.65 MHz 47.65 MHz 48.65 MHz External 49.65 MHz Magnetic Field Bo = 1.5 T 50.65 MHz 51.65 MHz Figure 5.8 Proton frequencies, and their corresponding helium frequencies for different slices of a body in a 1.5 T external magnetic field, with an applied linear gradient. Notice, each proton slice corresponds to the opposite slice for helium. 39

However, since the gradients are applied as shown in Figure 5.6, the RF pulse from the MR system selecting a particular slice with the proton frequency would be transformed by the transmit channel of the heterodyne system to an RF pulse selecting the opposite slice with the helium frequency (as shown in Figure 5.8). Hence, the transmit channel of the heterodyne system creates an effective gradient reversal. It is important that the slices in the body be selected and labeled correctly. With the current implementation of the heterodyne system, the slice selected by the MR system will not be the slice that is actually imaged. Two possible means for correcting this problem are: first, the gradients on the MR system can be reversed so that the slices selected with the proton frequency correspond to the same slices with the helium frequency. Second, if the local oscillator is changed to the other frequency, as described in Section 4.3, (15.23 MHz instead of 112.52 MHz), when the transmit channel converts the frequencies, there will be no apparent gradient reversal. With this second method, each slice in proton corresponds to the exact slice in helium. (The reason for choosing the local oscillator frequency of 112.52 MHz is explained in Section 4.3.) 5.4 Gating and the Radio-Frequency Amplifier Both the RF amplifier used with the heterodyne system and the narrowband RF amplifier used by the Signa MRI system are pulsed amplifiers. They do not continually transmit a signal; only when triggered by an external input (called the gating input) will they transmit the RF pulse. This gating signal turns on before the RF pulse is transmitted by the MR system and turns off only once the pulse has been transmitted. The gating signal is a square wave output by the MR system. When an RF pulse is being transmitted 40

by the MR system, the gating signal outputs 5 V, and when no RF pulse is transmitted, the gating signal outputs 0 V. Hence, the narrowband RF amplifier that is part of the MR system transmits when it receives a 5 V gating input and does not transmit when it receives a 0 V gating input. The broadband RF amplifier used with the heterodyne system, on the other hand, transmits when it receives a 0 V gating input and does not transmit when it receives a 5 V gating input. Therefore, unless the gating signal output by the MR system is altered, the broadband RF amplifier used with the heterodyne system will always be off when the RF pulse is transmitted. This means no signal will be transmitted to the coil and, thus, no image can be obtained. With the use of an inverter, the gating signal was transformed. The gating signal was input to an inverter. Thus, when the gating signal output 0 V (signaling that no RF pulse is being transmitted), the inverter output 5 V to the RF amplifier, turning it off. Likewise, when the gating signal output 5 V (signaling that an RF pulse is being transmitted) the inverter outputs 0 V, thereby turning the RF amplifier on. With this inverter, the broadband RF amplifier is turned on when then RF pulse is transmitted and, thus, images can be obtained. 41