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

394M. Luciana and C. A. Nelson

10% improvement in her use of strategy was noted. Tower of London performance remained unaltered. Set-shifting and recognition memory skills remained identical to those measured immediately postsurgery.

To summarize, the most dramatic gain observed from before surgery to 1 year postsurgery was in the length of her memory span, which increased from Wve to eight items. While test±retest reliability studies of CANTAB performance in children have not been conducted, this magnitude of gain would not be expected within a single year of normal development based on cross-sectional normative data. Her spatial recognition memory performance had also improved from 65 to 80% correct. Spatial working memory errors demonstrated a transient increase immediately postsurgery, suggesting a temporary disorganization in the brain network mediating this behavior, but stabilized 1 year later, showing a 30% improvement over initial testing.

Other clinical groups

CANTAB has been used to assess the patterns of impairment observed in several distinct populations of children with neurologic and/or neurodevelopmental impairments. The results are summarized in Table 22.3.

In addition to the work described above conducted in our laboratories, several other sites aYlated with the MacArthur Research Network on Psychopathology and Development are using the CANTAB to study children, including those with attention-deWcit hyperactivity disorder (Dr Ellen Lipman at McMaster University), autism (Dr Geraldine Dawson at the University of Washington), anxiety disorders (Dr Kathleen Merikangas at Yale University), children of depressed mothers (Dr Geraldine Dawson), and adolescents with histories of violent oVenses (Drs Laurence Steinberg and Elizabeth CauVman at Temple University and Stanford University, respectively). Through these endeavors, the discriminative validity of speciWc CANTAB tasks for identifying individuals with speciWc disorders aVecting cognitive function will be further assessed.

In summary, the CANTAB is a useful tool for the study of cognitive development in children. While the CANTAB appears to be a sensitive instrument for the assessment of neurobehavioral injury in children, its speciWcity with respect to localized brain impairment has yet to be demonstrated (linkage to discrete brain regions cannot be determined at this time). Nonetheless, because its neural correlates have been established in adult populations, informed hypotheses about localized brain dysfunction in children, such as those involved in our studies of PKU and

neonatal neurologic injury, are available. Although the clinical samples that we have tested involve cases with heterogeneous presentations (e.g., the NICU children), we believe that they are representative of the types of case that present for assessment and treatment at pediatric neuropsychology clinics.

Strategies for further validation of CANTAB in childhood assessment

Although additional behavioral studies of children with focal brain abnormalities would be a useful parallel to the adult data that has been gathered from neurosurgical patients, continued study of normative populations is also needed. Such studies have the goal of establishing that the brain±behavior correlates observed in adult populations are present in children. For instance, although it is apparent from several studies that CANTAB's spatial working memory task is sensitive to frontal lobe impairment in adults (cf., Owen et al., 1996c), a deWnitive link to frontal lobe function in children or adolescents has yet to be established. However, work has been initiated to develop neuroimaging protocols for use with children, utilizing CANTAB subtasks to address this issue (C. A. Nelson et al., unpublished data). This endeavor has been initiated through the use of a DMTS recognition memory task. This CANTAB subtest has been omitted from our normative studies because of its time demands. BrieXy, a geometric pattern (the ªsampleº) is displayed in the center of the computer screen for a brief interval. Following delays of 0, 4, 8 or 12 s, four patterns are displayed on-screen, one of which is the sample pattern. The child must correctly recall, and respond by touching, the sample pattern.

For use in fMRI, this task has been redesigned as a delayed nonmatch-to-sample (DNMTS). DNMTS task was selected because it has been extensively studied in nonhuman primates to determine its neural correlates and is widely considered to be the prototypical task for measuring visual recognition memory (e.g., Mishkin and Delacour, 1975). In DNMTS, the subject (generally a monkey) is Wrst presented with several training trials in which a stimulus object is baited with a reward. Following training, the subject is presented with the original object and a novel object. If the subject reaches for the novel object, s/he is again rewarded. This task requires formulating a memory from a visual perception of the Wrst object and creating a motoric response based on the memory. Lesion studies with monkeys have shown that this task is highly dependent on the hippocampus and underlying perirhinal cortex (Meunier et al., 1993; Alvarez et al., 1994),

as well as on the ventromedial prefrontal cortex (Kowalska et al., 1991).

For use with fMRI, the stimuli consist of the same abstract colored visual patterns that are used in CANTAB's DMTS task. On each trial, subjects are Wrst presented with a single target stimulus for a 3 s duration, after which the screen is blank for a 12 s delay interval. After this delay, the target stimulus and a novel stimulus appear together on screen for 3 s. Subjects are instructed to press a button that corresponds to the screen side of the novel stimulus. As in the monkey studies, this task requires that the subjects create a visual memory and form a motoric action based on the memory. In a perceptual control condition, the task is exactly the same except that there are no memory demands. SpeciWcally, the stimulus is presented on the screen for the 3 s familiarization period and remains on the screen throughout the 12 s delay and the response period. As a result, rather than relying on memory to determine which button to press, the subject can determine the correct answer perceptually. Then, following standard fMRI methodology (e.g., Cohen et al., 1993), the functional brain activation from the perceptual control task is subtacted from the memory task to identify the brain structures that are involved in the visual memory component of the task. In our initial work with four adult subjects (ages 22±30 years), a network of neural structures was activated. SpeciWcally, similar to the monkey studies, the medial temporal lobe (including the hippocampus and parahippocampal gyrus, along with the underlying perirhinal cortex) and the dorsolateral prefrontal cortex were activated.

Once past the pilot stage, studies will be conducted to determine if these Wndings can be extended to children. Behavioral studies with CANTAB's DMTS task, using stimuli identical to the ones used in the fMRI DNMTS task protocol, have found that adult-like performance on this task is not reached until 7 years of age (Fig. 20.5; Luciana and Nelson, 1998). Therefore, this study will examine the covariance between behavioral performance and patterns of neural activity in children between the ages of 6 years and adolescence (we currently lack Institutional Review Board permission to test healthy children below 6 years of age in fMRI studies). Of interest are the brain structures that subserve task performance in children. One might speculate that relatively mature structures, such as those within the medial temporal lobe, would be relatively more activated in children than in adults and would permit successful performance for children. Moreover, it may be possible to understand better how the lack of maturity of other structures (e.g., regions of the prefrontal cortex) alters the characteristics of the task-relevant neural network in children.

Summary

The CANTAB is useful for the evaluation of frontal and temporal lobe dysfunction in adults with acquired lesions. Whether these same neural correlates hold in the course of normal development has not been investigated. Neuroanatomic data suggest a protracted course of frontal lobe development through childhood, a Wnding that is consistent with our observation that children have not reached adult levels of performance on frontal lobe tasks before the age of 12 years (Luciana and Nelson, 1998). Although several studies of neurologically impaired pediatric samples suggest that CANTAB is able to discriminate between groups of children with and without neurologic impairment, its speciWcity as a diagnostic tool has yet to be determined. However, the use of a well-validated battery of tests, like the CANTAB, within a multimethod approach that includes neuroimaging, may lead to more reWned assessments of brain±behavior relations in children.

iReferencesi

Alvarez, P., Zola-Morgan, S. and Squire, L. R. (1994). The animal model of human amnesia: long term memory impaired and short term memory intact. Proc. Natl. Acad. Sci. USA, 91, 5637±41.

Baker, S. C., Rogers, R. D., Owen, A. M. et al. (1996). Neural systems engaged by planning: a PET study of the Tower of London task.

Neuropsychologia, 34, 515±26.

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. Neuroimaging, 2, 221±9.

Chugani, H. T. (1994). Development of regional brain glucose metabolism in relation to behavior and plasticity. In Human Behavior and the Developing Brain, eds. G. Dawson and K. Fischer, pp. 153±75. New York: Guilford Press.

Chugani, H. T. and Phelps, M. E. (1986). Maturational changes in cerebral function in infants determined by [18O]FDG positron emission tomography. Science, 231, 840±3.

Cohen, J. D., Noll, D. C. and Schneider, W. (1993). Functional magnetic resonance imaging: overview and methods for psychological research. Behav. Res. Meth. Instruments Computers, 25, 101±13.

Damasio, A. (1994). Descartes Error. New York: Plenum Press. Damasio, A. and Anderson, S. W. (1994). The frontal lobes. In

Clinical Neuropsychology, 3rd edn, eds. K. M. Heilman and E. Valenstein, pp. 409±60. New York: Oxford University Press.

Diamond, A. (1990). The development and neural bases of memory formation as indexed by the AB and delayed response tasks in human infants and infant monkeys. Ann. N. Y. Acad. Sci., 761, 267±317.

Diamond, A. and Goldman-Rakic, P. S. (1989). Comparison of human infants and rhesus monkeys on Piaget's AB task: evidence for dependence on dorsolateral prefrontal cortex. Exp. Brain Res., 74, 24±40.

Diamond, A., Prevor, M. B., Callendar, G. and Druin, D. P. (1997). Prefrontal cortex cognitive deWcits in children treated early and continuously for PKU. Monogr. Soc. Res. Child Dev., 62, 1±207.

Dias, R., Robbins, T. W. and Roberts, A. C. (1996). Dissociation in prefrontal cortex of attentional and aVective shifts. Nature, 380, 69±72.

Downes, J. J., Roberts, A. C., Sahakian, B. J., Evenden, J. L., Morris, R. G. and Robbins, T. W. (1989). Impaired extra-dimensional shift performance in medicated and unmedicated Parkinson's disease: evidence for a speciWc attentional dysfunction.

Neuropsychologia, 27, 1329±43.

Flavell, J. H., Beach, D. R. and Chinsky, J. M. (1966). Spontaneous verbal rehearsal in a memory task as a function of age. Child Dev., 37, 283±99.

Fray, P. J., Robbins, T. Wl. and Sahakian, B. J. (1996). Neuropsychiatric applications of CANTAB. Int. J. Geriat. Psychiatry, 11, 329±36.

Fuster, J. M. (1995). Temporal processing. Ann. N.Y. Acad. Sci., 769, 173±83.

Goldman-Rakic, P. S. (1987). Circuitry of primate prefrontal cortex and the regulation of behavior by representational memory. In

Handbook of Physiology, the Nervous System, Higher Functions of the Brain, Vol. 1, ed. F. Plum, pp. 373±417. Bethesda, MD: American Physiological Society.

Hughes, C., Russell, J. and Robbins, T. W. (1994). Evidence for executive dysfunction in autism. Neuropsychologia, 32, 477±92.

Huttenlocher, P. R. (1994). Synaptogenesis, synapse elimination, and neural plasticity in human cerebral cortex. In Minnesota Symposium on Child Psychology, Vol. 27, Threats to Optimal Development: Integrating Biological, Psychological, and Social Risk Factors, ed. C. A. Nelson, pp. 35±54. Hillsdale, NJ: Lawrence Erlbaum.

Huttenlocher, P. R. and Dabhholkar, A. S. (1997). Regional diVerences in synaptogenesis in human cerebral cortex. J. Comp. Neurol., 387, 167±78.

Jacobsen, C. F. (1936). Studies of cerebral function in primates.

Comp. Psychol. Monogr., 13, 1±68.

Kowalska, D. M., Bachevalier, J. and Mishkin, M. (1991). The role of the inferior prefrontal convexity in performance of delayed nonmatching to sample. Neuropsychologia, 29, 583±600.

Luciana, M., Lindeke, L., GeorgieV, M., Mills, M. and Nelson, C. A. (1999). Neurobehavioral evidence for working memory deWcits in school-age children with histories of prematurity. Devel. Med. Child Neurol., 41, 521±33.

Luciana, M. and Nelson, C. A. (1998). The functional emergence of prefrontally-guided working memory systems in four-to-eight year-old children. Neuropsychologia, 36, 273±93.

Meunier, M., Bachevalier, J., Mishkin, M. and Murray, E. A. (1993). EVects on visual recognition of combined and separate ablations of the entorhinal and perirhinal cortices in rhesus monkeys. J. Neurosci., 13, 5418±32.

Milner, B. (1964). Some eVects of frontal lobectomy in man. In The Frontal Granular Cortex and Behavior, eds. J. M. Warren and K. Akert, pp. 313±34. McGraw-Hill: New York.

Milner, B. (1971). Interhemispheric diVerences and psychological processes. Br. Med. Bull., 27, 272±7.

Mishkin, M. and Delacour, J. (1975). An analysis of short-term visual memory in the monkey. J. Exp. Psychol. Anim. Behav. Process., 1, 326±34.

Morris, R. G., Ahmed, S., Syed, G. M. and Toone, B. K. (1993). Neural correlates of planning ability: frontal lobe activation during the Tower of London test. Neuropsychologia, 31, 1367±78.

Nelson, C. A. (1995). The ontogeny of human memory: a cognitive neuroscience perspective. Dev. Psychol., 31, 723±38.

Owen, A. M., Downes, J. J., Sahakian, B. J., Polkey, C. E. and Robbins, T. W. (1990). Planning and spatial working memory deWcits following frontal lobe lesions in man. Neuropsychologia,

28, 1021±34.

Owen, A. M., Roberts, A. C., Pokley, C. E., Sahakian, B. J. and Robbins, T. W. (1991). Extra-dimensional versus intra-dimen- sional set-shifting performance following frontal lobe excisions, temporal lobe excisions, or amygdalo-hippocampectomy in man. Neuropsychologia, 29, 993±1006.

Owen, A. M., Sahakian, B. J., Semple, J., Polkey, C. E. and Robbins, T. W. (1995). Visuospatial short-term recognition memory and learning after temporal lobe excision, frontal lobe excision, or amygdalo-hippocampectomy in man. Neuropsychologia, 33, 1±24.

Owen, A. M., Doyon, J., Petrides, M. and Evans, A. C. (1996a). Planning and spatial working memory: a positron emission tomography study in humans. Eur. J. Neurosci., 8, 353±64.

Owen, A. M., Evans, A. C. and Petrides, M. (1996b). Evidence for a two-stage model of spatial working memory processing within the lateral frontal cortex: a positron emission tomography study.

Cereb. Cortex, 6, 31±8.

Owen, A. M., Morris, R. G., Sahakian, B. J., Polkey, C. E. and Robbins, T. W. (1996c). Double-dissociation of memory and executive functions in working memory tasks following frontallobe excisions, temporal lobe excisions or amygdalo-hippocam- pectomy in man. Brain, 119, 1597±615.

Passler, M. A., Isaac, W. and Hynd, G. (1985). Neuropsychological development of behavior attributed to frontal lobe functioning in children. Dev. Neuropsychol., 1, 349±70.

Petrides, M. and Milner, B. (1982). DeWcits on subject-ordered tasks after frontal and temporal lobe lesions in man.

Neuropsychologia, 20, 249±62.

Robbins, T. W. (1996). Dissociating executive functions of the prefrontal cortex. Proc. R. Soc. Lond. Ser. B, in press.

Robbins, T. W. and Sahakian, B. J. (1994). Computer methods of assessment of cognitive function. In Principles and Practice of Geriatric Psychiatry, eds. J. R. M. Copeland, M. T. Abou-Saleh and D. G. Blazer, pp. 205±9. Chichester, UK: Wiley.

Robbins, T. W., Roberts, A. C., Owen, A. M. et al. (1994). Monoaminergic-dependent cognitive functions of the prefrontal cortex in monkey and man. In Motor and Cognitive Functions of the Prefrontal Cortex, ed. A. M. Thierry, pp. 93±111. New York: Springer-Verlag.

Ross, G., Boatright, S., Auld, P. A. and Nass, R. (1996). SpeciWc cognitive abilities in two year-old children with subependymal and mild intraventricular hemorrhage. Brain Cognit., 32, 1±13.

Sattler, J. M. (1992). Assessment of Children, 3rd edn. San Diego: Jerome M. Sattler.

Shallice, T. (1982). SpeciWc impairments in planning. Philos. Trans. R. Soc. Lond. Ser. B, 298, 199±209.

Smith, M. L. and Milner, B. (1988). Estimation of frequency of occurrence of abstract designs after frontal or temporal lobectomy. Neuropsychologia, 26, 297±306.

Veale, D. M., Sahakian, B. J., Owen, A. M. and Marks, I. M. (1996).

SpeciWc cognitive deWcits in tests sensitive to frontal lobe dysfunction in obsessive compulsive disorder. Psychol. Med., 26, 1261±9.

Volpe, J. (1995). Neurology of the Newborn, 3rd edn. Philadelphia, PA: Saunders.

Weinberger, D. L. (1988). Evidence for dorsolateral prefrontal involvement in the pathogenesis of schizophrenia. Arch. Gen. Psychiatry, 45, 609±15.

Welsh, M. C., Pennington, B. F. and Grossier, D. B. (1991). A normative development study of executive function: a window on prefrontal function in children. Dev. Neuropsychol., 7, 131±49.

Introduction

The future of functional neuroimaging in child psychiatry is bound not only to technologic progress but also to the reWnement of our skills in exploiting these techniques to answer the ultimate questions of why and how symptoms occur. These skills require a clear understanding of the physiologic meaning of the signals recorded by the various neuroimaging techniques, the formulation of rational models of normal and deviant processes that can be evaluated systematically in a logical and stepwise fashion, and the intimately interactive and synergistic collaboration with other disciplines. Indeed, clinical taxonomy, psychopharmacology, molecular genetics, and basic neuroscience all provide critical clues for understanding the mechanisms underlying psychiatric disorders.

Ultimately, hypothesis generation and the interpretation of functional neuroimaging Wndings in child psychiatry will rely heavily upon the synthesis of scientiWc evidence originating from neuroscience and clinical research. The impact of brain maturation and brain plasticity lies at the forefront of the understanding of brain imaging Wndings in children and adolescents. How one approaches questions depends on one's theoretical assumptions concerning the extent of the brain's capacity for plasticity, the regional convergence and divergence of inputs, and the degree of equipotency of brain regions for adaptation. Functional neuroimaging can serve to bridge clinical science and basic neuroscience and, in doing so, realize its potential for translating advances in neuroscience into clinical applications.

After highlighting Wndings and hypotheses emerging from recent functional neuroimaging studies in both chil-

1The order of authorship does not reXect on the relative contributions of the coauthors; both contributed equally to this work.

dren and adults, this chapter will examine some of the critical needs that still need to be addressed to move the Weld of pediatric neuroimaging forward. The second part of the chapter will delineate recent advances in technology (progress in widely used modalities and new techniques) as well as in methodology (study design and data analysis). Finally, the authors will outline future directions in neuroimaging research in child psychiatry.

Selected contributions of functional neuroimaging in child psychiatry

With respect to its potential for addressing questions of pathophysiology in psychiatry, functional neuroimaging research has and continues to be met with some skepticism. Often, Wndings seem only to conWrm hypotheses concerning human brain function previously demonstrated in animal studies or supported by clinical reports, rather than substantively extending our knowledge base. The validation of models in humans is, however, critical for the formulation of new hypotheses directly applied to humans. In addition, because new knowledge overlaps with familiar theories, its import tends to be lost. Although it is still too early to appreciate fully the gains of functional neuroimaging in child psychiatry, studies of adult and adolescent patients have begun to suggest important developmental diVerences and similarities across age groups in the pathophysiology of psychiatric disorders, as well as to elucidate the neural bases of the therapeutic eVects of medications.

Attention-deWcit hyperactivity disorder (ADHD) is recognized as a highly prevalent and potentially lifelong disorder and is probably the childhood psychiatric disorder most studied in children and adolescents using functional neuroimaging to date. Symptom patterns of ADHD are

known to change across the lifespan, from childhood into adolescence into adulthood. Accordingly, positron emission tomography (PET) studies now oVer provocative evidence suggesting an evolution of neural deWcits with age, both supporting and extending a priori hypotheses of the neuropathophysiology of this disorder. While attentionally challenged, adolescents show discrete reductions in glucose metabolism relative to controls, particularly in the left anterior prefrontal cortex (Zametkin et al., 1993; Ernst et al., 1994). Adults show widespread cerebral hypometabolism, which, however, aVects predominantly the left anterior prefrontal cortex (Zametkin et al., 1990). Together, these Wndings suggest that the global reduction in synaptic activity in adults may be secondary to an initial causal deWcit, whereas dysfunction in the left anterior prefrontal cortex may reXect a more primary defect. The most likely mechanism underlying the observed reductions in brain metabolism is a dampening of modulatory synaptic activity that originates in subcortical structures and predominantly aVects the prefrontal cortex.

PET studies using [18F]-Xuorodopa (FDOPA) also suggest age-related diVerences in the pathophysiology of ADHD (Ernst et al., 1999a). Adults with ADHD have shown an abnormally low FDOPA signal in the anterior prefrontal cortex (Ernst et al. 1998). In contrast, although also showing an emerging reduction of the FDOPA signal in the anterior medial frontal cortex, adolescents with ADHD have demonstrated an elevated FDOPA signal in the midbrain (substantia nigra and ventral tegmentum complex) (Ernst et al., 1999a). These Wndings provide the most direct evidence to date in support of the dopaminergic hypothesis of ADHD (Solanto, 1984; Levy, 1991; Castellanos, 1997; Ernst, 1998) and suggest a need for additional studies probing systematically the presynaptic and postsynaptic dopaminergic systems.

In schizophrenia, age-related similarities in neuroimaging Wndings support a common pathophysiology for the adult-onset and much rarer childhood-onset forms. As detailed earlier in this volume, studies using magnetic resonance spectroscopy (MRS) have revealed regionally speciWc reductions of N-acetylaspartate, suggestive of neuronal involvement, in the mesial temporo-limbic and prefrontal cortices in both childhood-onset and adult-onset schizophrenia. Abnormally low frontal activity (hypofrontality), one of the most consistent Wndings in PET/SPECT (single photon emission computed tomography) research in adult-onset schizophrenia, has also been reported in adolescents with early-onset schizophrenia.

Preliminary studies of anorexia nervosa using PET and SPECT suggest a role of the temporal lobe in both adults (Herholz et al., 1987) and adolescents (Gordon et al., 1997),

a role already proposed in the 1950s (Anand and Brobeck, 1951). Temporal lobe dysfunction is hypothesized to reXect a functional imbalance in the limbic system, leading to a disturbance in the hypothalamic±pituitary axis directly responsible for the clinical presentation of anorexia nervosa. However, neuroimaging Wndings also suggest that adult pathology involves additional structures including the caudate nucleus (Herholz et al., 1987) and parietal cortex (Delvenne et al., 1997).

In Tourette's disorder, metabolic, blood Xow, and, to a lesser extent, radioligand studies have implicated the basal ganglia portions of the cortico±striato±thalamo±cortical circuitry in the pathophysiology of this disorder. These Wndings may, in part, reXect secondary compensatory function (i.e., attentional processing associated with tic suppression) and require further investigation. While a study of the dopaminergic system of adult patients has failed to identify abnormalities (Turjanski et al., 1994), adolescents with Tourette's disorder have shown elevated accumulations of FDOPA in the left caudate and right midbrain (Ernst et al., 1999b). Together, these studies suggest the possibility of important developmental changes, again reinforcing the need for further studies that include children.

Other emerging Wndings have begun to localize the eVects of therapeutic medications in child psychiatry. Blood Xow studies suggest that stimulants may normalize prefrontal activity in ADHD (Teicher et al., 1996; Vaidya et al., 1998) but fail to normalize brain activity in other regions (i.e., striatum; Vaidya et al., 1998). Such paradigms oVer opportunities for localizing the sites of the therapeutic impact of pharmacologic and other treatments. In addition, this approach opens possibilities for exploring the neural substrates of age-related diVerences in drug response, such as diVerences in the eYcacy of antidepressant treatments of adolescent versus adult depression.

Critical needs in pediatric neuroimaging

In most domains of inquiry, research in children has lagged behind research in adults. This chapter proposes several areas of need speciWc to the investigation of children that are fundamental to the successful application of neuroimaging in child psychiatry. These areas include the collection of age-related normative data relevant to indices of brain function at rest and during task performance, the development of age-appropriate brain atlases, and the prioritization of research questions in this extraordinary burgeoning Weld. Finally, integrative models capable of bridging the gap between basic neuroscience and clinical

400M. Ernst and J. M. Rumsey

science are needed for hypothesis generation, data interpretation and theory building.

Normative models of structural, functional, and biochemical brain development

The analysis and interpretation of functional brain imaging data relies on its integration with structural data. The use of a common algorithm for stereotaxic normalization in functional imaging studies assumes that brain anatomy is comparable across individuals and groups. Systematic changes with age or disease introduce error in mapping function to structure and in averaging normalized data across subjects for group comparisons. Because existing atlases are derived from a single or at best a few individual brains, anatomic variability is largely ignored. Therefore, such atlases are more accurate for brain regions with low intersubject variability and for sites close to the landmarks of the reference system than for more variable regions (e.g., neocortex, asymmetrical perisylvian regions). To address these limitations, eVorts are underway to develop digital and probabilistic brain atlases. These atlases use data from large populations, which provide probabilistic information on variability in brain architecture, and base their landmarks on intersubject stability (Mazziotta et al., 1995; Thompson et al., 1996, 1997). Such atlases will provide a means of detecting, mapping and distinguishing ageand disease-related deviations in brain anatomy and a framework for integrating functional and anatomic data across subjects and imaging modalities.

Also needed is a better understanding of normal functional brain development. To date, few functional neuroimaging studies have directly examined the eVects of maturation in pediatric samples. Not only is it critical to understand how the brain develops in humans to elucidate pathology, but such data can also foster the development of mathematical models to apply in the analysis of brain imaging data from children. Such mathematical models can be used to correct for age-related diVerences, which can confound the eVects of variables under study, and help to reduce the number of subjects required for a given study. In fact, mathematical models of maturational changes, such as those in glucose metabolism, are under development (Muzik et al., 1999), but data needed to make these models fully operant are lacking.

Adequate normative models of functional brain development underlying speciWc neurobehavioral abilities of relevance to child psychiatric disorders are lacking. While activation studies have assumed a prominent role in deWning the neural circuitry involved in a host of neurocognitive and neurobehavioral functions in adults,

studies seeking to delineate their development in children are just emerging. For example, Luna et al. (1999) have begun to map the neural circuitry subserving response inhibition in children (ages 8 to 13 years), adolescents (ages 14 to 17 years), and adults using an antisaccade task. Preliminary results suggest that maturational improvements in voluntary inhibition are subserved by striato±thalamic and cerebello±thalamic modulation of higher-order fronto±parietal cortical networks.

Brain neurochemical development can also be modeled using functional neuroimaging and may help to elucidate the underlying biology of abnormalities identiWed using activation paradigms. However, neurochemical investigations are lagging behind activation studies because, so far, neurotransmitter studies can be conducted only with PET and SPECT technology. While the ethical issue of radiation exposure with these techniques remains a concern in studies of children, such studies are critical for the understanding of the neural mechanisms responsible for aberrant behavior and the subsequent development of rational therapeutic interventions.

Setting priorities in research questions

Which psychologic functions, neurotransmitter systems, and neural circuitry represent priorities for human brain mapping in child psychiatry? What role do they play in these disorders? Which of the available study designs are best suited to addressing them? These are all important questions that must be addressed.

Child psychiatric disorders are multifaceted, involving emotional and behavioral dysregulation and cognitive deWcits as well as underlying neurotransmitter and neuroendocrine disturbances. Therefore, their study can be approached from a variety of perspectives, using both activation paradigms and neurochemical methods. Challenges include the choice of the appropriate aVective, attentional, cognitive, or pharmacological probes. This choice depends on the nature of the processes hypothesized to be critically involved in a given disorder and the presumed primary or secondary role of the function under scrutiny. The investigation of relatively unexplored areas such as emotions and their role in learning (a manifestation of brain plasticity) may also represent a priority for neuroimaging research in child psychiatry. Neural models of emotional development and its interaction with cognitive development have the potential to revolutionize the Weld of psychopathology.

Nature of the process under study

DeWcits in executive functions, attention, and working memory are broadly implicated in child psychiatric disor-

ders, and, therefore, these domains constitute priorities for neuroimaging in child psychiatry. These processes are involved in emotional, cognitive, and behavioral self-regu- lation. Each of them is multifactorial and subsumes a variety of elemental functions, which presumably engage somewhat diVerent neural networks or impose diVerential demands on them. While impairments within these domains have been noted across a wide range of developmental neuropsychiatric disorders, their speciWc nature within and across disorders and their developmental precursors require further elucidation. For example, whereas deWcits in executive function characterize both autism and ADHD, their nature diVers between these disorders. DeWcits in vigilance and response inhibition are prominent in ADHD but not in autism, which instead is characterized by deWcits in the disengagement and shifting of attention (Pennington and OzonoV, 1996). Behavioral studies elucidating core deWcits associated with speciWc disorders thus provide for the rational selection of activation tasks for use in neuroimaging studies. While executive functions, attention, and working memory are thought to rely on fronto±subcortical brain circuits throughout human development, this remains a working hypothesis in need of testing in normally developing children and in those with psychiatric disorders. Circuit-speciWc probes capable of distinguishing disorders will help to further our knowledge of the diversity of ways in which key fronto±subcortical circuits are aVected in child psychiatric disorders.

Primary versus secondary role of de®cits

The choice of study tasks and conditions depends heavily on assumptions concerning the primary (causal) or secondary (resultant) role of identiWed deWcits in child psychiatric disorders. For example, whereas deWcits in the ability to understand other's intentions (ªtheory of mindº deWcits) characterize autism, these may be secondary and develop as a result of early deWcits in conjoint (social) attention, which in turn may stem from deWcits in the ability to shift attention rapidly and Xexibly from one object to another (e.g., from mother to objects in the environment). If this is the case, the identiWcation of the neural substrates of such secondary deWcits will be of limited value in elucidating the primary underlying neurobiology.

Forays into new areas

To date, neuroimaging has been used less in studies of emotional processing than in studies of cognitive functions, even though emotional disturbances are part of the clinical picture of most psychiatric disorders. Activation paradigms for examining the neural bases of emotion and

limbic function are, however, emerging. Functional imaging has not only conWrmed the role of limbic structures in human emotions but is also providing new evidence of deWcient habituation as a potential mechanism underlying psychopathology. For example, the passive viewing of facial expressions has been shown to elicit amygdalar activation in adult volunteers (Breiter et al., 1996). These responses, elicited in healthy individuals, show rapid habituation relative to nonlimbic brain regions involved in facial recognition (e.g., fusiform gyrus), thus providing a potentially useful index of physiologic adaptability. Abnormalities of resting amygdalar blood Xow in depression (Drevets, 1998) and task-related amygdalar blood Xow in anxiety disorders (Rauch et al., 1998) suggest the involvement of this limbic structure in psychiatric disorders. Hyper-reactivity or an inability to habituate to threat stimuli may characterize anxiety disorders, whereas dysfunction aVecting the ability to assess emotional stimuli may characterize depressive disorders or autism. Studies in children can clarify whether neural abnormalities involving limbic responsivity and habituation constitute strong predictors or early signs of disorder or whether they evolve over the course of illness.

Important relationships between cognitive and emotional development can be examined using paradigms assessing emotionally based learning. For example, imaging paradigms have begun to examine the role of the amygdala in conditioned fear acquisition and extinction in adults (LaBar et al., 1998). The circuitry underlying emotional learning has been elaborated in an event-related functional magnetic resonance imaging (fMRI) study in which faces were conditioned through pairing with an aversive tone (Buchel et al., 1998), eliciting responses in the amygdalae, anterior cingulate, and anterior insula, regions involved in emotional processing. The use of passive paradigms (e.g., facial expression paradigms) with young children may help to assess the integrity of primary emotional brain systems, whereas the use of conditioning paradigms may help to assess the impact of these systems (or lack thereof) on learning.

Integrative models

There is a need for the development of theoretical models of systems-level and large-scale neural networks that mediate and link the emotional/behavioral dysregulation and the cognitive deWcits concurrently present in child psychiatric disorders. One such model is that proposed by Damasio (1994), who views emotion and cognition as intimately linked in the brain and partners in the regulation of behavior. Within this model, primary emotions, which are

402M. Ernst and J. M. Rumsey

wired in at birth, are processed and categorized by the early sensory cortices and detected by limbic structures, including the amygdala and anterior cingulate. Secondary emotions develop as connections between neural representations of categories, objects, and situations are formed. These processes are supported by widely distributed neural networks, which include the prefrontal and somatosensory cortices. Therefore, emotions and ªfeelings of emotionº (which include both primary emotional and cognitive elements) are represented at many diVerent neural levels, including the neocortical level.

Damasio (1994) highlights the unique and privileged role of the prefrontal cortices in the integration of emotion and cognition. Receiving signals from all the sensory systems and from bioregulatory systems (e.g., brainstem neurotransmitter nuclei, basal forebrain, amygdala, anterior cingulate, hypothalamus), the prefrontal cortices have available to them knowledge of the external world, of innate biological regulatory preferences, and of prior and current body states. The prefrontal cortices mediate categorizations of the contingencies of personal life experiences and, thus, provide a basis for the projecting scenarios of future outcomes required for predictions and planning. Finally, the prefrontal cortices can directly aVect cerebral output (e.g., motor and chemical responses). The integrated functioning of these prefrontal systems allows humans to think ahead, predict outcomes, preemptively avoid danger, generalize, and respond Xexibly based on personal experience. Clearly, such a model can help to organize the diVerent phases of a research plan aimed at elucidating the origin and evolution of a particular disorder. However, such research can occur only if it is supported by reliable and sensitive technologies and methodologies.

Technologic advances

Enhancements of existing technology

Technologic progress entails the enhancement of the sensitivity and speciWcity of the physiologic signals, the implementation of the combined use of available complementary techniques, and the minimization of health hazards. Improvements of spatial and temporal resolution and developments in methodology continue to be at the forefront of technical eVorts in neuroimaging research. Current fMRI techniques provide spatial resolution as good as 2±5mm and temporal resolution of less than 1 s. The ultimate limits of temporal resolution are not yet known and depend on the signal-to-noise ratio of imaging

measurements and on the underlying variance in hemodynamic latencies. Variance in the delay and shape of the hemodynamic curve observed across brain regions and subjects may, in part, reXect delayed neuronal processing, opening possibilities for further improvements in temporal resolution (see Rosen et al., 1998). Higher magnetic Weld strengths, improvements in radiofrequency receiver coil technology, and greater sophistication in data analytic methods oVer the hope of further improvements in sensitivity and spatiotemporal resolution.

In addition, the combined use of diVerent functional neuroimaging techniques can facilitate the dissection in time and space of the neural processes that mediate behavior and cognition. For example, both fMRI and PET maps related to speciWc tasks can be further analyzed using event-related potential and magnetoencephalographic techniques to deWne the underlying temporal dynamics of distinct regional activations more exactly (George et al., 1995; Thatcher, 1996; Rosen et al., 1998). However, multimodal neuroimaging is diYcult to implement because of the constraints imposed by cost, time, technical feasibility, and instrument availability; consequently its use to date is limited. Nonetheless, the development of methods for integrating imaging modalities is expected to contribute to coherent models of the neural systems underlying normal and aberrant emotional and behavioral development.

Advances in fMRI are emphasized because of the paucity of use of new PET or SPECT techniques with children. The development of MRI as a functional imaging tool presents minimal health hazards when appropriate caution is exercised. Its lack of ionizing radiation allows the repeated study of both adults and children. Practical considerations imposed by comfort and movement, rather than radiation dose, have now become the most relevant limiting factors for subjects when using fMRI.

However, it is also important to note that the development of increasingly sensitive PET and SPECT scanners is reducing the doses of radioactive tracer necessary to obtain reliable images. These doses, which are already minimal, may eventually reach a level that will cease to raise health concerns. In addition, alterations of experimental design can minimize the dosages required for studies of children (Zametkin et al., 1993). As already mentioned, PET and SPECT are the only techniques to date that can provide reliable measures of neurobiochemical systems, particularly neurotransmitter activities. The development of ligands and of mathematical models to extract the physiological information of the behavior of these ligands constitutes the major eVorts in PET and SPECT research.

Although fMRI presents critical advantages over PET and SPECT in activation studies, fMRI has two relative disadvantages. While motion artifacts pose a major problem in fMRI studies of children, they are less of an issue with PET or SPECT. The other relative disadvantage of fMRI is the inability to measure neural activity associated with behavior outside of the scanner, in contrast to PET or SPECT where the uptake of the tracer can often occur outside of the scanner. This may present a signiWcant advantage for subjects who cannot easily be trained to remain still and to tolerate the conWnes of fMRI magnets. This also makes it possible to sedate severely impaired children for scanning only after tracer uptake.

New techniques

The range of functional imaging modalities and the tools available for use in conjunction with functional neuroimaging continue to expand. One promising new modality with exquisite temporal resolution is optical imaging (Frostig et al., 1990). Optical imaging makes use of the transmission of infrared light through the head and brain to measure localized scattering and absorption changes in optical reXectance. These changes in optical reXectance are inXuenced by the concentrations of oxygenated and deoxygenated hemoglobin, which are dependent on local neural activity (Malonek and Grinvald, 1997). The temporal resolution of this technique reaches the picosecond range; yet the spatial resolution can exceed 1 cm. In addition, measures are limited to superWcial cortical regions.

Of particular interest is the potential use of this technique in association with electrical recordings to clarify relationships between neuronal activity and hemodynamic changes. Optical imaging has been used in humans to detect visual evoked activity in the occipital cortex. Gratton et al. (1997) found that event-related optical signals elicited in occipital cortex (with latencies of approximately 100ms) colocalized with the signals obtained with fMRI. The temporal and spatial resolution of the optical imaging signal makes it a promising technique for assessing the time course of activity in localized superWcial cortical brain areas.

Another relatively new tool that can enhance the capabilities of functional neuroimaging is transcranial magnetic stimulation (TMS). TMS can be used in combination with functional neuroimaging to map the substrates of simple to complex behaviors and neuronal connections. This tool permits the stimulation of discrete brain areas and can be employed to disrupt function, to enhance cortical excitability and (potentially) function, and to determine the functional connectivity between

brain regions (Pascual-Leone et al., 1999). The disruption of task performance through focal TMS stimulation can provide evidence that a stimulated region is critical for performance. However, future applications of this new tool in studies of children will require additional evaluation of safety.

Recent applications of this technology to the study of plasticity have included the eVects of practice on cortical network representations (Classen et al., 1998) and the cortical reorganization of visual cortex in blind subjects (Cohen et al., 1997). In the latter application, midoccipital TMS interfered with Braille reading but not with speech in early-blind subjects, thus indicating the intimate involvement of this region in their reading. Stimulation of somatosensory cortex, but not occipital cortex, disrupted the ability of sighted subjects to identify embossed letters tactilely, whereas the opposite pattern (a double dissociation) was seen in blind subjects, further suggesting cross-modal plasticity as a mechanism of functional compensation.

The combined use of TMS with PET oVers a novel approach to mapping functional connections in humans. To determine patterns of functional connectivity without requiring the subject to engage in a speciWc task, Paus et al. (1997) stimulated selected cortical regions while simultaneously measuring local and remote changes in resting regional cerebral blood Xow. Stimulation of the left frontal eye Weld provided signiWcant correlations between the number of TMS pulse trains and blood Xow in the visual cortex of the superior and medial parieto-occipital regions, consistent with the known anatomic connectivity of the monkey frontal eye Weld. Because no task was involved, potential confounds associated with task parameters and performance variables were avoided. Given the importance of plasticity in brain maturation, the use of this tool may be particularly well suited for studying connectivity in developmental neuropsychiatric disorders.

Clinical therapeutic applications of TMS in psychiatry are being investigated. Brain regions that are hypometabolic in a psychiatric disorder may be stimulated to enhance cortical excitability with potentially therapeutic results. Particularly promising is a role for TMS in the treatment of depression (Kirkcaldie et al., 1997; George et al., 1999). Recent work by Keenan et al. (1999) has demonstrated enhancement of the ability to recognize one's own face (one aspect of self-awareness) by repetitive TMS delivered to the right prefrontal cortex of normal adults. While it is too early to evaluate realistically the range of potential therapeutic applications, such Wndings raise the possibility of future clinical applications within child psychiatry.