Книги по МРТ КТ на английском языке / Functional Neuroimaging in Child Psychiatry Ernst 1 ed 2000

54G. F. Eden and T. A. ZeYro

high variance near tissue boundaries, e.g., the brain surface/cerebrospinal Xuid boundary, may identify interscan motion. Translational motion that has a peak-to-peak amplitude of less than 10% of the image voxel width will not usually contribute substantially to variance in the resulting statistical images. Using these methods, it is possible to obtain good estimates of interscan motion to determine the success of subject restraint procedures. They also may be employed before and after corrective procedures are applied to determine their eYcacy.

Having determined that suYcient head motion is present to warrant correction, it is possible to employ an automated realignment procedure to determine the coordinate transformation that will bring the members of the time series back into register (Woods et al., 1998a, b). This is usually the most time consuming and computationally intensive part of the entire analysis procedure. For fMRI datasets, a linear realignment algorithm, allowing variation in the translation and rotational degrees of freedom, is usually employed. Having determined the appropriate coordinate transformation to reregister the volumes, a resampling algorithm is employed to generate the realigned image volume. For this procedure, there is a tradeoV between time and accuracy, with the most accurate resampling procedures (sinc interpolation) being signiWcantly slower than the less accurate procedures (nearest neighbor or trilinear interpolation). Most laboratories utilize trilinear interpolation for this step as it provides a reasonable compromise between execution speed and resampling accuracy. An example of the eYcacy of head motion correction in children and adults is shown in Fig. 3.2 (p. 82).

Global and local signal variation

There are multiple sources of spatially invariant, global signal intensity changes with time. These variations may be slow, as occurs with instrumental drift, or of mixed frequency, as may occur with physiologically based Xuctuations. For example, respiratory activity may have a mechanical eVect on the observed MR signal, with intrathoracic pressure modulations causing venous sinus pulsations; it may also have an indirect physiologic eVect on the signal by causing changes in the blood partial pressure of carbon dioxide that result in global cerebral blood Xow changes. These changes in global cerebral blood Xow will be accompanied by changes in the BOLD-contrast signal and will, therefore, confound attempts to measure taskrelated changes in the BOLD-contrast signal. Whatever their source, global changes in signal intensity may be corrected using the technique of ratio normalization (Friston

et al., 1994), in which the mean value of each volume of the time series is adjusted to a speciWed mean value. This approach assumes that, as in PET cerebral blood Xow images, there is a linear relationship between local and global signal intensity in EPI.

Even after global signal variations are corrected, it is not uncommon to observe regionally speciWc modulations of the MR signal with time, particularly in areas close to the brain surface. Many of these artifacts result from uncompensated head motion; that is, motion with eVects that are not removed by rigid-body transformations. These eVects usually occur at frequencies below those of the task of interest and may be removed with linear detrending or high-pass digital Wltering. In this instance it is assumed that signal Xuctuations occurring at low frequencies are not related to task-related signal changes and may be safely removed.

Statistical map generation

After processing the time series to remove instrumental and physiologic sources of error variance, parametric statistical techniques are employed to detect regionally speciWc task-related signal changes. These techniques include simple categorical contrasts (t-test), correlation, linear regression, analysis of variance, and many others. A thorough review of these methods is beyond the scope of this chapter and the reader is referred to a number of more detailed reviews of this topic (Bandettini et al., 1992; Bandettini, 1993; Friston et al., 1994, 1998).

Structure_function correlation

One approach for determining the neuroanatomic localization of task-related signal changes in individual subjects involves the coregistration of statistical maps with corresponding high-resolution structural images using intermodality linear spatial registration techniques (Woods et al., 1998a, b). This approach allows direct visualization of the loci of signal change in a neuroanatomic framework and permits labeling of responses according to sulcal or gyral landmarks.

An alternative approach involves transforming the statistical maps into a generalized coordinate system, usually an anatomic atlas (Talairach and Tourneaux, 1988), allowing standardized reporting across experiments and laboratories. An advantage of mapping all scans to a common coordinate system is that it allows intersubject averaging of responses for pixelwise statistical testing (ZeYro et al., 1997). However, this coordinate system is based on an adult brain, and pediatric counterparts have not been

implemented to date. Given the changes in size of cortical and subcortical structures throughout development (Giedd et al., 1996), analogous pediatric templates for the developmental stages will be useful for making accurate standardized maps.

Data interpretation

Statistical maps can be characterized in diVerent ways and currently there is no concensus on which analysis oVers the best measure of task-related signal change. Various groups have used the mean Z-score in a cluster of activated voxels, the spatial extent of a cluster, the peak Z-score within a cluster, or the corresponding percentage signal change at this local maximum.

In the pediatric studies described above, it has been reported that frontal activation during a working memory task is very similar between children and adults (Casey et al., 1997a). However, the results show that the percentage increase in MR signal increases with increasing age and yet the activation in children appears to be more diVuse than that in adults (also reported by Hertz-Pannier et al. (1997)). While these response measures may be highly correlated in some circumstances, in others they are not (Eden et al., 1999). Further investigations into the relationships of these measures are needed.

Example of a pediatric language study

Children with developmental dyslexia have diYculties with phonologic processing. The term ªphonologic awarenessº refers to skills of manipulation and segmentation of the constituent sound of words. For example, rhyme judgement (such as ªHat, cat, dog, mat ± which is the odd one out?º) can measure phonologic awareness and predict reading outcome (Bradley and Bryant, 1983).

The neuronal mechanisms that subserve phonologic segmentation which are impaired in dyslexia are currently under active investigation (Rumsey et al., 1992; Rumsey and Eden, 1997; Shaywitz et al., 1998). With advances in neuroimaging technology, it is now possible to study an individual subject's performance during a wider variety of tasks than was previously possible because of radiation dosimetry limits. As a result, an eVort has begun in our laboratory to study individuals with normal cognitive development to understand the functional organization of the brain for language. These eVorts will allow a deeper understanding of the eVects of developmental dyslexia on a range of sensory and phonologically related skills. An example of such a study is as follows. We utilized the BIG technique (see above) to investigate phonologic awareness

skills in adults and children with and without dyslexia. The task utilized phoneme elision, which requires an awareness of the sound structure of language in order to be able to delete the Wrst sound of a word. In this study, subjects viewed a series of words and were instructed to either read the word aloud (baseline) or to say the word after omitting the Wrst sound (phonologic manipulation). The sound elision condition was then compared with the word reading condition as well as with a visual Wxation condition. As illustrated in Fig. 3.2 (p. 82), fMRI data in individuals typically revealed involvement of temporoparietal areas, the inferior frontal gyri, and cerebellum. This Wgure also demonstrates that the amount of head movement produced by a child is greater than that of the adult but can be corrected for in the postprocessing procedures.

Conclusions and future directions

Clearly the opportunity to study pediatric populations using noninvasive fMRI has tremendous potential for furthering our understanding of human development in health and disease. Not only does fMRI allow acquisition of physiologic information in children that was previously unobtainable by PET because of restrictions on radiation exposure, but it also allows the information to be sampled at multiple time points. This has implications for longitudinal studies and allows for the observation of physiologic changes at times when behavioral changes may be occurring.

Reorganization of brain function following early brain injury in humans has been studied by making behavioral observations of language and visuospatial development. These behavioral observations have led to theories of functional reorganization across the two hemispheres involving homologous sites. While studies of altered behavioral performance have led to explanations such as crowding (Teuber, 1974), the true reorganization of the brain can only be assessed physiologically. The use of fMRI provides an opportunity to follow individuals in whom large portions of the brain have for some reason become modiWed. One such case report has been published by Levin et al. (1996), in which an adolescent showed interhemispheric reorganization of visuospatial skills from the right to the left hemisphere that were measured with fMRI.

Utilization of functional neuroimaging techniques is providing new information concerning the neuroanatomic localization of the systems aVected in developmental dyslexia (Eden and ZeYro, 1998). However, the details of how these ªalteredº brain areas are changed as a result of remediation has not been addressed to date. Rehabilitative

56G. F. Eden and T. A. ZeYro

recovery of reading functions after stroke has been assessed with fMRI (Small et al., 1996), demonstrating altered brain physiology in the inferior parietal cortex after phonological training in acquired dyslexia. Despite evidence from animal models establishing anatomic and physiologic correlates of recovery or learning, these phenomena have not been investigated in children with developmental disorders. With the ability to scan children repeatedly and noninvasively, in vivo human brain reorganization can be measured. Examination of these physiologic changes will improve our understanding of the plasticity of this disorder; it may also allow the most suitable remediation techniques to be identiWed and may possibly allow determination of the physiologic predispositions most suited to intervention. The investigation of these questions will further our understanding of the mechanisms of developmental disorders and provide information on the eVectiveness of various remediation techniques.

Acknowledgements

We would like to thank Kimberley Noble and Jane Joseph for editing the manuscript. This work was supported by the Charles A. Dana Foundation and the Department of Defense (DAMD17±93±V±3018).

Appendix: suggested sources of equipment for fMRI experiments

LCD projector

Toshiba LCD products http://www.toshiba.com

Sharp Electronics Corporation 1-800-BE-SHARP

http://www.sharp-usa.com, http://www.fei.com nView Computer and Video Projection

http://www.nview.com Projection lens

BUHL Optical, USA

Contact: (412) 321-0076 Rear projection screen

Da-Lite Screen Company, USA (800) 622-3737 http://www.da-lite.com

Response device (optical) Current Designs, Inc.

(215) 387-5456 bdugan@netaxs.com

Systems

Psychology Software Tools (412) 271-5040 info@pstnet.com

Resonance Technology http://mri-video.com/rtc

Neuroscan Inc. (703) 444-7100 sales@neuron.com

Sensor Systems (703) 437-7651 www.sensor.com

iReferencesi

Bandettini, P. A. (1993). MRI studies of brain activation: temporal characteristics. In Proceedings of the First Annual Meeting of the International Society of Magnetic Resonance in Medicine, pp. 143±51, Dallas, TX.

Bandettini, P. A., Wong, E. C., Hinks, R. S., Tikofsky, R. S. and Hyde, J. S. (1992). Time course EPI of human brain function during task activation. Magn. Reson. Med., 25, 390±7.

Beischer, D. E. (1962). Human tolerance to magnetic Welds.

Astronautics, 42, 24±5.

Belliveau, J. W., Kennedy, D. N., McKinstry, R. C. et al. (1991). Functional mapping of the human visual cortex by magnetic resonance imaging. Science, 254, 716±19.

Benson, R. R., Logan, W. J., Cosgrove, G. R. et al. (1996). Functional MRI localization of language in a 9-year-old child. Can. J. Neurolog. Sci., 23, 213±19.

Binder, J. R., Rao, S. M., Hammeke, T. A., Frost, J. A., Bandettini, P. A. and Hyde, J. S. (1994). EVects of stimulus rate on signal response during functional magnetic resonance imaging of auditory cortex. Cogn. Brain Res., 2, 31±8.

Binder, J. R., Frost, J. A., Hammeke, T. A., Cox, R. W., Rao, S. M. and Prieto, T. (1997). Human brain language areas identiWed by functional magnetic resonance imaging. J. Neurosci., 17, 353±62.

Biswal, B., de Yoe, A. E. and Hyde, J. S. (1996). Reduction of physiological Xuctuations in fMRI using digital Wlters. Magn. Reson. Med., 35, 107±13.

Bookheimer, S. Y., ZeYro, T. A., Blaxton, T., Gaillard, W. and Theodore, W. (1995). Regional cerebral blood Xow during object naming and word reading. Hum. Brain Map., 3, 93±106.

Bradley, L. and Bryant, P. (1983). Categorizing sounds and learning to read ± a causal connection. Nature, 301, 419±21.

Buckner, R. L., Bandettini, P. A., O'Craven, K. M. et al. (1996). Detection of cortical activation during averaged single trials of a cognitive task using functional magnetic resonance imaging.

Proc. Natl. Acad. Sci. USA, 93, 14878±83.

Casey, B. J., Cohen, J. D., Jezzard, P. et al. (1995). Activation of prefrontal cortex in children during a nonspatial working memory task with functional MRI. Neuroimage, 2, 221±9.

Casey, B. J., Cohen, J. D., King, S. W. et al. (1997a). A developmental functional MRI study of cortical activation during a spatial working memory task. Neuroimage, 5, S69.

Casey, B. J., Trainor, R. J., Orendi, J. L. et al. (1997b). A developmental functional MRI study of prefrontal activation during performance of a go±no±go task. J. Cogni. Neurosci., 9, 835±47.

Cho, Z. H., Chung, S. C., Lim, D. W. and Wong, E. K. (1998). EVects of the acoustic noise of the gradient systems on fMRI: a study on auditory, motor, and visual cortices. Magn. Reson. Med., 39, 331±6.

Cohen, M. S. (1996). Rapid MRI and functional applications. In

Brain Mapping: The Methods, ed., A. W. Toga and J. C. Mazziotta, pp. 223±52. San Diego, CA: Academic Press.

Dhankhar, A., Wexler, B., Fulbright, R. F., Halwes, T., Blamire, A. and Shulman, R. G. (1997). Functional magnetic resonance imaging assessment of the human brain auditory cortex response in increasing word presentation rates. Am. Physiol. Soc., 77, 476±83.

Eden, G. F. and ZeYro, T. A. (1998). Neural systems aVected in developmental dyslexia revealed by functional neuroimaging.

Neuron, 21, 279±82.

Eden, G. F., Joseph, J. E., Brown, H. E., Brown, C. P. and ZeYro, T. A. (1999). Utilizing hemodynamic delay and dispersion to detect fMRI signal change without auditory interference: the behavior interleaved gradients technique. J. Magn. Reson. Med.,

41, 13±20.

Edmister, W. B., Talavage, T. M., Ledden, P. J. and WeisskoV, R. M. (1999). Improved auditory cortex imaging using clustered volume acquisitions. Hum. Brain Map., 7, 89±97.

Fiez, J. A., Raichle, M. E., Miezin, F. M., Petersen, S. E., Tallal, P. and Katz, W. F. (1995). PET studies of auditory and phonological processing: eVects of stimulus characteristics and task demands. J. Cogn. Neurosci., 7, 357±75.

Fox, P. T. and Raichle, M. E. (1984). Stimulus rate dependence of regional cerebral blood Xow in human striate cortex, demonstrated by positron emission tomography. J. Neurophysiol., 51, 1109±20.

Friston, K. J., Jezzard, P. and Turner, R. (1994). Analysis of functional MRI time-series. Hum. Brain Map., 1, 153±71.

Friston, K. J., Josephs, O., Rees, G. and Turner, R. (1998). Nonlinear event-related responses in fMRI. Magn. Reson. Med., 39, 41±52.

Frost, J. A., Binder, J. R., Newby, R. F. et al. (1997). Phonological processing in developmental dyslexia: an fMRI study. Neuroimage,

7, S568.

Giedd, J. N., Vaituzis, A. C., Hamburger, S. D. et al. (1996). Quantitative MRI of the temporal lobe, amygdala, and hippocampus in normal human development: ages 4±18 years. J. Comp. Neurol., 366, 223±30.

Guimaraes, A. R., Melcher, J. R., Talavage, T. M. et al. (1998). Imaging subcortical auditory activity in humans. Hum. Brain Map., 6, 33±41.

Haxby, J. V., Horwitz, B., Ungerleider, L. G., Maisog, J. M., Pietrini, P. and Grady, C. L. (1994). The functional organization of human extrastriate cortex: a PET±rCBF study of selective attention to faces and locations. J. Neurosci., 14, 6336±53.

Hertz-Pannier, L., Gaillard, W. D., Mott, S. et al. (1994). Pre-opera-

tive assessment of language by fMRI in children with complex partial seizures: preliminary study. In Proceedings of the 2nd Annual Meeting of the International Society of Magnetic Resonance in Medicine, vol. 1, p. 326, Dallas, TX.

Hertz-Pannier, L., Gaillard, W. D., Mott, S. H. et al. (1997). Noninvasive assessment of language dominance in children and adolescents with functional MRI: a preliminary study.

Neurology, 48, 1003±12.

International Non-Ionizing Radiation Committee (of the International Radiation Protection Association (IRPA/INIRC) (1991). IRPA/INIRC guidelines: protection of the patient undergoing a magnetic resonance examination. Health Physics, 61, 923±8.

Jackson, G. D (1994). New techniques in magnetic resonance and epilepsy. Epilepsia, Suppl. 6, S2±13.

Jackson, G. G., Connelly, A., Cross, J. H., Gordon, I. and Gadian, D. G. (1994). Functional magnetic resonance imaging in focal seizures. Neurology, 44, 850±6.

Kwong, K. K., Belliveau, J. W., Chesler, D. A. et al. (1992). Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc. Natl. Acad. Sci. USA, 80, 5675±9.

Levin, H. S., Scheller, J., Rickard, T. et al. (1996). Dyscalculia and dyslexia after right hemisphere injury in infancy. Arch. Neurol.,

53, 88±96.

Ogawa, S., Tank, D. W., Menon, R. et al. (1992). Intrinsic signal changes accompanying sensory stimulation: functional brain mapping using MRI. Proc. Natl. Acad. Sci. USA, 89, 5951±5.

Poncelet, B. P., Wedeen, V. J., WeisskoV, R. M. and Cohen, M. S. (1992). Brain parenchyma motion: measurement with cine echo-planar MR imaging. Radiology, 185, 645±51.

Price, C., Wise, R., Ramsay, S. et al. (1992). Regional response diVerences within the human auditory cortex when listening to words. Neurosci. Lett., 146, 179±82.

Price, C. J., Wise, R. J. S., Watson, J. D. G., Patterson, K., Howard, D. and Frackowiak, R. S. J. (1994). Brain activity during reading: the eVects of exposure duration and task. Brain, 117, 1255±69.

Price, C. J., Wise, R. J. S., Warburton, E. A. et al. (1996). Hearing and saying: the functional neuro-anatomy of auditory word processing. Brain, 119, 919±31.

Rees, G., Howseman, A., Josephs, O. et al. (1997). Characterizing the relationship between BOLD contrast and regional cerebral blood Xow measurements by varying the stimulus presentation rate. Neuroimage, 6, 270±8.

Rumsey, J. M. and Eden, G. F. (1997). Functional neuroimaging of developmental dyslexia: regional cerebral blood Xow in dyslexic men. In SpeciWc Reading Disability: AView of the Spectrum, ed. B. Shapiro, P. J. Accardo and A. J. Capute, pp. 35±62. Timonium:York Press.

Rumsey, J. M., Andreason, P., Zametkin, A. J. et al. (1992). Failure to activate the left temporoparietal cortex in dyslexia. Arch. Neurol.,

49, 527±34.

Rumsey, J. M., Horwitz, B., Donohue, B. C., Nace, K., Maisog, J. M. and Andreason, P. (1997a). Phonologic and orthographic components of word recognition: a PET±rCBF study. Brain, 120, 739±59.

Rumsey, J. M., Nace, K., Donohue, B. C., Wise, D., Maisog, J. M. and Andreason, P. (1997b). A positron emission tomography study of impaired word recognition and phonological processing in dyslexic men. Arch. Neurol., 54, 562±73.

Shaywitz, S. E., Shaywitz, B. A., Pugh, K. R. et al. (1998). Functional disruption in the organization of the brain for reading in dyslexia. Proc. Natl. Acad. Sci. USA, 95, 2636±41.

Shulman, G. L., Corbetta, M., Buckner, R. L. et al. (1997). Top-down modulation of early sensory cortex. Cereb. Cortex, 7, 193±206.

Slifer, K., Cataldo, M. F., Cataldo, M. D., Llorente, A. M. and Gerson, A. C. (1993). Behavior analysis of motion control for pediatric neuroimaging. J. Appl. Behav. Anal., 26, 469±70.

Small, S. L., Noll, D. C., Perfetti, C. A., Hlustik, P., Wellington, R. and Schneider, W. (1996). Localizing the lexicon for reading aloud: replication of a PET study using fMRI. Neuroreport, 7, 961±5.

Talairach, J. and Tourneoux, P. (1988). Co-planar Stereotactic Atlas of the Human Brain: 3-Dimensional Proportional System: An Approach to Cerebral Imaging. Stuttgart: Thieme.

Tenforde, T. S. and Budinger, T. F. (1985). Biological eVects and physical safety aspects of NMR imaging and in vivo spectroscopy. In NMR in Medicine: Instrumentation and Clinical Applications, ed. S. R. Thomas and R. L. Dixon, p. 493. New York: American Association of Physicists in Medicine.

Teuber, H. L. (1974). Why two brains? Cambridge, MA: MIT Press. Turner, R. and Jezzard, P. (1993). Magnetic resonance studies of brain function activation using echo-planar imaging. In

Functional Neuroimaging, ed. R. W. Thatcher, M. Hallett, E. R. John and M. Huerta, pp. 69±78. San Diego, CA: Academic Press.

Turner, R., Jezzard, P., Wen, H. et al. (1993). Functional mapping of the human visual cortex at 4 and 1.5 Tesla using deoxygenated contrast EPI. Magn. Reson. Med., 29, 281±3.

van Meter, J.W., Maisog, J. M., ZeYro, T. A., Hallett, M., Herscovitch, P. and Rapoport, S. I. (1995). Parametric analysis of functional neuroimages: application to a variable-rate motor task.

Neuroimage, 2, 273±83.

Woods, R. P., Grafton, S. T., Holmes, C. J., Cherry, S. R. and Mazziotta, J. C. (1998a). Automated image registration: I. J. Comp. Assist. Tomogr., 22, 139±52.

Woods, R. P., Grafton, S. T., Watson, J. D., Sicotte, N. L. and Mazziotta, J. C. (1998b). Automated image registration: II. J. Comp. Assist. Tomogr., 22, 153±65.

ZeYro, T. A., Eden, G. F., Woods, R. P. and van Meter, J. W. (1997). Intersubject analysis of fMRI data using spatial normalization.

Adv. Exp. Med. Biol., 413, 235±40.

Introduction

The application of magnetic resonance spectroscopy (MRS) techniques to the study of neuropsychiatric disorders in childhood provides an extraordinary opportunity to advance our understanding of the neurobiological mechanisms underlying these disorders. Recent developments in imaging technology have aVorded researchers the capability to examine in vivo not only brain structure but also neurochemistry and functional architecture. This review will focus on the use of MRS. All MR methods (MRS, magnetic resonance imaging (MRI) and functional MRI, (fMRI)) rely on the same basic principles. In the study of human brain, all three of these procedures use the same hardware. Of these techniques, MRS was the Wrst technology to be developed, exploiting the magnetic properties of nuclei with unpaired protons and neutrons. The earliest observations of NMR signals from bulk matter were made independently by Purcell and colleagues (1946) and by Bloch et al. (1946). However, a number of important technical developments were necessary before the Wrst spectra could be obtained from human brain in the 1980s (Krishnan and Doraiswamy, 1997). A glossary of technical terms used in functional neuroimaging is provided at the end of this book (p. 480).

Methods in MRS

The importance of MRS for understanding neuropathologic processes is derived from the information this technique provides regarding the chemical content of the tissue being studied. The application of this technology allows investigators to acquire data that describe both the chemistry and the physical environment of the tissue. It is, therefore, possible to examine and quantify changes in

metabolite levels of chemical substances such as N- acetylaspartate (NAA), phosphocreatine (PCr), creatine plus phosphocreatine (Cr), and cytosolic choline compounds (Cho) and relate them to changes in structural pathology. As with MRI technology, spectroscopic techniques are rapidly evolving.

MR visible compounds give rise to distinct peaks, or resonances. In general, the area of the resonance intensity is proportional to the concentration of molecules that contribute to the resonance. Thus, quantiWcation of resonance intensities may be used to derive tissue concentration estimates for brain chemicals. In practice, these calculations also require knowledge of the relaxation times, T1 and T2, of the molecule of interest, the data acquisition parameters, time to repetition (TR) and time to echo (TE), the tissue volume of interest, and the eYciency of signal detection. The collection of these parameters is very time consuming; consequently, it is a common practice to express MRS data as metabolite ratios or in terms of institutional (standardized) units. In addition, the chemical composition of gray matter and white matter are quite diVerent. For many study hypotheses it is important to assess the tissue content of speciWc brain regions. QuantiWcation of resonance intensities is possible for a number of nuclei, including proton (1H), carbon (13C), Xuorine (19F), and phosphorus (31P). So far, 31P MRS and 1H MRS have been applied to the study of neuropsychiatric disorders. The aim of these studies has been to characterize the concentrations of metabolites such as NAA, Cr, and Cho in speciWc brain regions, which may be related to neuropathologic processes.

In general, MR methods are based on the fact that speciWed nuclei may align with or against a static magnetic Weld, at slightly diVerent energy levels. Two events are necessary for the generation and observation of MR signals. Two magnetic Welds are needed for a signal to be recorded:

60D. A. Yurgelun-Todd and P. F. Renshaw

Wrst, a stable Weld (usually in the range 1.5±4.0T), which is characteristic of the magnetic strength of the MR scanner, and, second, a transient magnetic Weld that is introduced at a speciWc frequency, causing a transition of some lower energy spins (aligned with the static magnetic Weld) to the higher energy level (aligned against the static magnetic Weld). This second Weld is typically introduced using a radiofrequency (rf) coil, or antenna. For each magnetic nucleus in the brain, a given static magnetic Weld is associated with a particular resonance frequency, often called the Larmor frequency. For example, at 1.5T, the proton resonance frequency is 63.88MHz. When the second magnetic Weld is removed, usually by turning oV the second rf magnetic Weld, energy at the resonance frequency is released from the higher energy spins as they align with the Wrst magnetic Weld and this energy may be detected. This release of energy comprises the MR signal. The process by which the nuclei realign themselves with the static con-

stant Weld is called relaxation, or recovery (T1 relaxation is within the longitudinal axis and T2 is within the transverse axis). More detailed descriptions of the MR method have

been published elsewhere (e.g., Krishnan and Doraiswamy, 1997; Mukherji, 1998; see also Chapter 3.)

The design of an MRS experiment requires that study parameters are set to optimize signal-to-noise from a clearly delineated brain region in as rapid a time course as possible. In addition to hardware considerations, and the selection of nuclei to be studied, both localization strategy and relaxation times must be speciWed. Therefore, the selection of the nucleus to be studied is only the Wrst in a series of important decisions to be made in deWning an MRS protocol. Given that MRS techniques have low sensitivity, most study strategies are designed to optimize the available signal.

Nuclei that have been assessed through the application of MRS to the human brain include 1H, 31P, 13C, 19F, lithium-7 (7Li), and sodium-23 (23Na) (Table 4.1). Although psychiatric investigations have generally been limited to the study of lithium, Xuorine, hydrogen, and phosphorus, recent studies

of neurologic patients have reported interesting Wndings with 23Na and 13C. Sodium-23 gives rise to a single resonance line, and changes in the sodium MRS resonance have been associated with cerebral ischemia (e.g., Tyson et al., 1996). Carbon-13 is a stable isotope with a low natural abundance, which makes it possible to administer and detect labeled compounds (Mason et al., 1996). At present, 13C-labeled compounds are very expensive, which limits experimentation. Over time, it is likely that 13C MRS methods will be developed that will permit the direct observation of neurotransmitter cycling by introducing a neurotransmitter or its precursor labeled with 13C. Current investigations in children with psychiatric disorders, however, have focused on

Table 4.1. Relative NMR sensitivities

Note: Spin quantum number is a term used to describe the angular momentum of nuclei.

a limited number of metabolites such as NAA, Cr, and Cho. Studies of each nucleus are associated with speciWc advantages and limitations. From a practical perspective, since each nucleus will have a unique resonance frequency at a given Weld strength, additional rf coils and ampliWers are usually necessary for each MRS nucleus. Therefore, most clinical studies of patients report Wndings based on a single nucleus.

Brain 31P, 1H, and 13C spectra give rise to multiple resonance lines, all of which are generated by compounds containing these nuclei. The diVerence in resonance frequencies for diVerent compounds arises from interactions within molecules (Bovey et al., 1988), making it possible to assess the concentrations of a number of diVerent compounds using these methods. Each nucleus has a diVerent MR sensitivity (Table 4.1), which in turn limits the spatial resolution of the MR experiment. Sensitivity is an important consideration in MRS studies in that the metabolites being observed are typically present at concentrations in the millimolar range. In contrast, the concentration of brain water is of the order of 40 mol/l, the primary reason why MR images of brain water have such striking contrast. At the other end of the sensitivity spectrum, radionuclide imaging (e.g., positron emission tomography (PET) and single photon emission computed tomography (SPECT)) allows the detection of molecules that are present at nanomolar concentrations, and is used to measure, among other applications, brain receptor distributions (Chapters 1 and 2). Parameters for MRS experiments can be adjusted to maximize the detection of metabolites by increasing their ªMRS visibilityº. The most common Weld strength for human MRS studies at the present time is 1.5T. However, a number of research centers are currently installing scanners at a Weld strength of 3 or 4T. These higher Weld scanners are particularly valuable for MRS studies, as MR signals increase linearly with Weld strength, which in turn results in increased sensitivity.

As we have described above, MRS allows for the noninvasive, in vivo visualization and examination of brain metabolites in a manner that was previously not possible. Moreover, MRS oVers several advantages over radionuclide neuroimaging techniques (PET and SPECT) for the study of children and adolescents, in particular its absence of ionizing radiation, which allows repeated examinations of a single study subject. MRS data are also acquired with the same basic hardware as structural imaging data, thereby facilitating the collection of both data types within the same scanning session. Several methods are currently used for the acquisition of spectroscopic data and include both single-voxel and chemical shift imaging (CSI). Spectroscopic data may be obtained from either single, predeWned tissue volumes (voxels) or from twoor three-dimensional arrays of tissue (spectroscopic images). It is generally easier to perform single-voxel MRS as opposed to spectroscopic imaging as it requires less imaging time for the subject. However, when spectroscopic imaging parameters have been optimized, data sets may be obtained without penalty in terms of data acquisition time or spatial resolution. Therefore, it is usually preferable to obtain spectroscopic imaging data when it is possible. Single-voxel spectroscopy provides information about a speciWc cube of tissue localized in a particular brain region, deWned with help from the magnetic Weld gradients. CSI is actually a multivoxel method in which a signal is collected from a wide region of tissue that may include up to an entire brain slice. This array of data is later decoded into individual spectra from each of the voxels (Fig. 4.1). In practice, most single-voxel studies of human subjects include anywhere from one to three voxels per imaging session. The resolution of single-voxel spectroscopy is considered superior, as the magnetic Weld can be optimally homogenized for the volume selected, while CSI has the advantage of being able to acquire data from multiple regions of the brain simultaneously. For clinical investigations aimed at the identiWcation of focal pathology, the single-voxel method may be most advantageous.

Although studies utilizing MRS yield an abundance of information about both the structure and chemical composition of tissue, MR technology is limited by a number of factors. As with any neuroimaging technique, MRS requires that study participants remain completely still for the duration of the examination, which may be diYcult for children. Additionally, the signal from one nucleus or from a group of nuclei may be dispersed into two or more through spin coupling. While often used for the identiWcation of speciWc resonances, this splitting of the signal results in a reduced signal-to-noise ratio and complicates the spectra, making the interpretation of the data more diYcult. Perhaps the most signiWcant limitation of

Fig. 4.1. Coronal image through the medial temporal lobe depicting the placement of a multivoxel grid used in chemical shift imaging. Twelve proton spectra are generated from the 12 voxels within the larger volume of interest.

MRS is its lack of sensitivity. The signal strength of a particular nucleus is dependent upon its inherent magnetogyric ratio and the external applied magnetic Weld strength. The magnetic Weld needs to be precisely homogenized in order to acquire narrow, clear resonance peaks. The sensitivity of MRS can be increased by altering a number of imaging and study design parameters, for example increasing the applied magnetic Weld strength.

Measurement of brain metabolites for speciWc brain regions is accomplished by measurement of the spectrum or spectra after the data collection is complete. An example of a proton spectrum is presented in Fig. 4.2. The MR signal is displayed as a spectrum with characteristic peaks associated with diVerent elements. The area under individual peaks is measured relative to speciWed reference compounds, although estimation of the area under each peak

is often confounded by overlapping peaks. Some investigators report their Wndings as absolute values while others report their data as a ratio of one metabolite to another. Interpretation of spectral data requires that saturation and relaxation eVects be considered and that the underlying tissue content of the regions of interest are known and examined. Although the MRS procedure requires a thoughtful, detailed approach, this exciting method will

yield in vivo biochemical data that were previously not available for the study of human brain.

Limitations of MRS

In this review, we will consider evidence for abnormal concentrations of various metabolites detected with MRS in

the brains of children with psychiatric disorders. Additionally, this chapter will address how these MRS Wndings Wt with data from studies of children with other neurologic pathologies, and how the MRS abnormalities may relate to primary pathology as well as being associated with risk factors and clinical features of behavioral disorders. For example, brain abnormalities found in adult studies of schizophrenic patients have generally been interpreted by investigators as reXective of neuropathology associated with the primary schizophrenic process. However, the meaning ascribed to such cortical changes depends crucially on the theoretical models within which the data are viewed. Caveats in interpreting previous research and suggested directions for new studies will be discussed, along with potential clinical applications for future research.

A critical evaluation of the published research suggests that two major factors may contribute to inconsistency in the results from neuroimaging studies in children with psychiatric disorders. The Wrst involves the nature and extent of potential biases in the selection of both patient and control subjects. Subject selection, sample size, and variables on which patients and controls are matched have all been demonstrated to be important factors for the outcome of neuroimaging studies (Hendren et al., 1995, Jacobsen et al., 1996). In studies of children and adolescents, the eVects of age and sex are potent determinants of hemispheric laterality, regional morphometry, metabolite concentration, and cortical activation (Kreis et al., 1985; Witelson, 1985) and may critically aVect the ability to identify cortical changes between children from diVerent diagnostic groups. A second factor that restricts the interpretation of neuroimaging studies is the inconsistency in study methods. As with any new technique, there remains debate as to the optimal parameters and procedures for the application of MRS to the study of childhood psychiatric disorders.

Initial studies applying MRS methods to psychiatric populations generally examined a single voxel in the brain. In contrast, recent studies using CSI have provided more extensive, quantitative measurements of metabolite concentrations in multiple brain regions. Findings often vary considerably. Even recent studies using sophisticated MR acquisition techniques are constrained by the absence of completely objective and reliable anatomic landmarks to demarcate precise and speciWc brain regions of interest. This diYculty in localization results, in part, from an absence of formal agreement as to the cortical landmarks to be used in human MRS studies and the inability to establish isotropic cortical volumes in study subjects. For patient populations that display only small quantitative diVerences from control subjects in brain structure and

chemistry, the inability of imaging techniques to characterize speciWc cortical regions precisely contributes to the diYculty in reliably detecting group diVerences. Methodologic limitations also exist both for MR data acquisition and image-analytic techniques. Few investigators, for example, have utilized the same MRS acquisition parameters, postimaging data processing techniques, or statistical approaches to data analyses and hypothesis testing. Finally, conceptual approaches regarding how best to deWne and demonstrate what is abnormal, and what falls within the spectrum of normal brain variance, has remained a point of debate.

Despite the limitations outlined above, MRS studies of children with psychiatric disorders may have some advantages. It has been hypothesized that children with psychiatric disorders may represent more homogeneous diagnostic subtypes and, therefore, be more likely to yield unique biological correlates for illnesses such as depression and schizophrenia. This would be of considerable clinical importance, as neurobiologic subtypes have been proposed to be related to clinical variables, such as chronicity of illness, poor treatment response, presence of negative symptoms, and neurocognitive function. Studies of children are also more likely to provide data on the etiologic processes as neurobiologic abnormalities are less prone to changes through treatment or to the disease process. To date, brain MRS research has targeted adult populations, in part because of the relative diYculty in obtaining good-quality scans from children. Additionally, children and adolescents are more likely to have imaging data confounded by motion artifact or dental braces and some psychiatric illnesses, such as schizophrenia, are often not diagnosed until the late teens and early twenties.

Metabolites measured by MRS

MRS employs standard MRI devices to make measurements of chemical levels within the brain. MRS-visible compounds that can be measured noninvasively in human brain include psychotropic medications, such as lithium (Sachs et al., 1995; Jensen et al., 1996; Soares et al., 1996; Riedl et al., 1997; Renshaw and Wicklund, 1998) and some Xuorinated polycyclic drugs (Komoroski et al., 1991; Renshaw et al., 1992; Miner et al., 1995; Strauss et al., 1997, 1998). Spectra arising from either 7Li or 19F, in drugs that contain Xuorine, are typically single lines; however, these lines may also include signal from active and inactive metabolites. These studies have generally reported whole brain drug concentrations because of the relatively low brain concentrations of the therapeutic agents (0.1±1.0