4. Electrophysiological and Neuroimaging Techniques in Neuropsychology

The actual brain waves occurring during the artifact can be rejected when compared to typical brain waves associated with the type of evoked potentials. There are distinctive patterns associated with auditory and visual evoked potentials. These will be discussed in more detail.

In summary, brain wave differences have been found in selective and sustained attention tasks in children with ADHD, particularly when complex tasks are utilized. There is emerging evidence that the subtypes may differ in brain electrical activity, with children with ADHD without hyperactivity symptoms showing the largest amplitude differences from normal controls. If children with ADHD have different patterns of selective attention that can be measured through ERPs, it may well be possible to demonstrate changes in these differences following treatment. For example, children with ADHD without hyperactivity were found to have a larger than normal response to rare stimuli, which may be interpreted to mean that they overreact to novel stimuli. This finding is similar to the behavioral observation that ADHD children tend to be very attentive to rare and/or novel situations, tasks, and experiences. The findings of the processing negativity study indicate that it is almost impossible for these children to inhibit their responses. Therefore, not only do children with ADHD without hyperactivity show better attention to novel stimuli, but they also find it extremely difficult to ignore or inhibit responses to these rare occurrences. Thus, differences in how the brains of children with ADHD (ADHD with hyperactivity and ADHD without hyperactivity) respond to the environment may provide information for the development of interventions as well as furthering our understanding of the underpinnings of these disorders.

Computed tomography (CT) allows for the visualization of the brain anatomy to determine the presence of focal lesions, structural deviations, and tumors. A CT scan uses a narrow X-ray beam that rotates 360 degrees around the area to be scanned. Each CT slice is acquired independently, and that slice can be repeated if a movement artifact occurs. Scanning time is 15–20 minutes per slice. The resulting data are transformed by Fourier analysis into gray scale to create the image. Figure 4.3 illustrates a CT scan image. One advantage of a CT scan is the short acquisition time. Another is the ability to repeat single slices if movement or other artifacts occur. Limitations of CT scans are the relatively poor resolution of gray and white matter structures. Moreover, CT slices are obtained in the axial plane, which limits visualization of the temporal lobes and posterior fossa (Filipek & Blickman, 1992). You may recall from Chapter 2 that the pediatric population is susceptible to tumors of the posterior fossa, which cannot be easily visualized by CT scanning. Although within acceptable ranges, radiation is used in CT scans in order to produce the images.

Magnetic resonance imaging (MRI) procedures offer noninvasive investigation of neuroanatomical structures in a living brain. MRIs visualize brain tissue at about the level of postmortem studies, with clarity superior to that of CT scans. Figure 4.4 presents a representative MRI scan. The MRI consists of a large magnet with magnetic field strengths up to 3.0 Tesla for use with humans, and up to 7.0 for animals. Tesla is an indication of the magnet strength.

The patient lies on a movable table, which goes inside a doughnut hole-shaped opening. A coil covers the area, to be scanned with an opening over the face in the use of the head coil. To understand how MRI works, it is important to briefly review the physics behind it. In a magnetic field, hydrogen photons align in the same direction as the field. A radio-frequency pulse will deflect these photons into an angle predetermined by the clinician. Once the pulse ceases, the photons return to their original alignment through a series of ever-relaxing and slowing circles (Gazzaniga, Ivry, & Mangun, 2002). Altering the rate, duration, and intensity of the radio frequency pulses allows for the differing visualization of the brain. An MRI is actually several complementary sequences, each an average of 7–10 minutes in length.

In clinical use, routine MRI scans utilize TI- and T2-weighted sequences. T-weighted scans provide visualization generated photons involved in the pulse, whereas T2-weighted scans involve the interaction of the neighboring photons (Cohen, 1986). T1-weighted scans provide excellent anatomical detail with myelin and structural abnormalities. In contrast, T2-weighted scans are sensitive to water content in the tissues and are used to determine the extent of the lesions.

Because MRIs are noninvasive and do not use radiation, they are fairly risk-free. MRIs are expensive and are generally used only when there is suspicion of brain abnormality. They are more sensitive than CTs for locating lesions and, when a contrast agent such as gadolinium is used, the tumor can be differentiated from surrounding swelling.

In the past, neuropsychological assessment was used to try to localize brain tumors, lesions, etc. The MRI has supplanted neuropsychological assessment for this process in cases where tumors or lesions are suspected. Neuropsychological assessment is used for intervention planning and for determining how the person solves the difficulty. Neuropsychological assessment continues to be sensitive to subtle damage or damage at the microcellular level that the MRI is unable to pinpoint.

MRI technology now allows us to obtain pictures of the brain from living children and assess the impact of treatment on the brain's development (Filipek et al., 1989). Because it may be possible to detect neuroanatomical differences before the behavioral problems emerge in response to school requirements, investigating brain measures may allow us to begin studying children at a much younger age and provide earlier interventions.

Dyslexia Research by Hynd, Semrud-Clikeman, Lorys, Novey, and Eliopulos (1990) and Semrud-Clikeman, Hynd, Novey, and Eliopulos (1991) utilizing MRI scans found differences specific to dyslexia in the neuroanatomical regions involved with language processing. These findings were replicated in several subsequent studies (Duara et al., 1991; Jernigan, Hesselink, Sowell, & Tallal, 1991; Larsen, Hoien, Lundberg, & Odegaard, 1990). Additional findings in these studies indicated no differences in the cerebral hemispheric area or posterior areas in dyslexic subjects. These findings strongly implicate areas thought to be important in language processing. No differences were found in brain size, so reported regional differences more strongly implicate the likelihood of a neurodevelopmental process active during gestation of these regions. Abnormal development is thought to occur somewhere between the 24th and 28th week of gestation.

Fine, Semrud-Clikeman, Keith, Stapleton, and Hynd (2006) studied children with dyslexia within families at risk for learning problems. Findings indicated that better readers show larger midsagittal regions in the middle of the corpus callosum compared to poorer readers. In addition left and right volumes of the corpus callosum were symmetrical and not related to the diagnosis of a reading disability.

For a regional difference to be predictive, it needs to be unique to a specific disorder. The consistent finding of smaller splenial (the posterior portion of the corpus callosum) measurements in ADHD may be unique to ADHD (Hynd et al., 1990; Semrud-Clikeman et al., 1994). Using a carefully diagnosed sample of ADHD with overactivity symptoms and no codiagnoses, compared to matched non-disabled control subjects, Semrud-Clikeman et al. (1994) were able to demonstrate differences through MRI analysis in the posterior regions of the corpus callosum, with particularly smaller splenial regions for the ADHD group. No differences were found in the anterior regions of the corpus callosum, in callosal length, or callosal total area. When response to medication was evaluated, a trend also was found that may suggest structural differences for the ADD/H sample based on medication response. Subjects who did not respond to methylphenidate, but did respond to desipramine or imipramine, tended to have a smaller corpus callosum splenial measure compared to subjects who were positive methylphenidate responders or control subjects. These results indicate that medication response may be partially mediated by fewer connections between posterior regions of the brain. Given the small number of subjects, the area of medication response and its relationship to brain structure needs further study.

A study by Semrud-Clikeman et al. (1996) explored the utility of using selected brain morphometric indices to predict group membership for children who were developmental dyslexic, ADHD, and a typically developing group. Six brain regions were selected a priori for inclusion in a discriminant function analysis (right and left plana temporale length, right and left insula length, left and right anterior width). This analysis classified participants with an overall accuracy at 60 percent with the best accuracy found for persons with developmental dyslexia and controls. When chronological age and FSIQ were added to the discriminant analysis, the overall classification accuracy rose to 87 percent with the misclassified participants assigned to one of the clinical groups. This preliminary study supports the hypothesis that selected brain structures can reliably discriminate groups by developmental disorder.

A structural MRI study by Filipek et al.(1997) found the frontal regions to be smaller volumetrically in children with ADHD compared to normal controls, with greater differences in the right frontal region. Moreover, the region of the brain including the caudate head and other anterior basal ganglia were also found to be volumetrically smaller in the ADHD sample than in the controls. Analyzing the white matter in the frontal area yielded significantly smaller volume in frontal right white matter volume in the ADHD sample. The finding of similar cerebral hemispheric and ventricular volumes in the ADHD cohort without enlarged external cerebrospinal fluid spaces indicates that the underlying pathophysiology is not likely the result of degeneration or atrophy. Rather differences in specific hemispheric regions implicate a neurodevelopmental process which alters neural system configuration, particularly in the right hemisphere, in children with ADHD. Differences have also been found in the corpus callosum, with smaller regions variously reported in the genu and the splenial regions (Giedd et al., 1994; Hynd, Marshall, & Semrud-Clikeman, 1991; Semrud-Clikeman et al., 1994).

The caudate regions have been found to be smaller in several studies in children with ADHD (Castellanos et al., 1994; Hynd et al., 1993; Semrud-Clikeman, 2006; Semrud-Clikeman, Pliszka, Lancaster, & Liotti, 2006).The caudate is intimately involved in the dopaminergic system; a system in which methylphenidate is believed to correct imbalances. Moreover, lesions to the caudate in adults and animals have resulted in behaviors which are very similar to those seen in hyperactive children (Posner & Rothbart, 2007).

These results relate directly to the clinical finding that hyperactivity is a common symptom in caudate infarcts (Caplan, Schmahmann, & Kase, 1990). Hynd et al. (1995) suggested that such a difference may be related to lower levels of neurotransmitter being relayed to the frontal lobes, resulting in compromised levels of complex attentional skills. In contrast, Castellanos et al. (1994) found the right caudate to be smaller and symmetrical with the left caudate in their ADHD sample, while the control sample evidenced R > L caudate volumes. Castellanos et al. (1994) also found that the volume of the caudate decreased significantly with age for the normal controls, with no difference in the ADHD group. Filipek et al. (1997) found smaller volumetric measurements in the left anterior caudate of the ADHD sample compared to normal controls. This finding is consistent with the results of Hynd et al. (1995). Symmetrical caudate volume was found for the ADHD group due to a smaller than expected left caudate. The control group possessed left greater than right caudate volume. Moreover, there was a significant medication effect related to smaller left caudate volume with the ADHD sample who responded favorably to methylphenidate showing the smallest left caudate volume, followed by the sample who did not respond to methylphenidate, and with the normal control group showing the largest left caudate volume.

A more recent study controlling medication history found no asymmetry in the caudate (Semrud-Clikeman et al., 2006). However, findings indicated that the caudate was larger bilaterally in groups of children with ADHD and with and without a medication treatment history, compared to a group of children who were typically developing. There were no laterality differences in the volume of the caudate.

Although there are conflicting findings as to the direction of the caudate asymmetry, the studies to date have found that there are structural differences in the caudate regions of ADHD children compared to normal children. The caudate is intimately involved in the dopaminergic system, a system in which methylphenidate is believed to correct suspected imbalances. Moreover, lesions to the caudate in adults and animals have resulted in behaviors that are similar to those seen in hyperactive children (Posner & Raichle, 1994). Further study is needed to determine the role of the caudate in ADHD as well as the contributions of structural differences in this disorder.

Findings indicate that the anterior cingulate may differ between children with ADHD and controls. Those children without a history of stimulant treatment were found to show smaller volumes in the right ACC compared to children with a history of stimulant treatment and with controls (Semrud-Clikeman et al., 2006). The children with a history of stimulant treatment of at least one year (most of them four years or more) showed anterior cingulate volume equivalent to that of the controls. This is an important finding because the anterior cingulate is a structure that appears to be crucial for directing attention as well as being aware of successful or unsuccessful solutions to a problem. If it is possible for the brain to adapt to input from the environment when appropriate interventions are provided, structural changes indicate that such interventions actually allow neural changes to be implemented that may assist with the processing of information.

These studies are preliminary and must be interpreted with caution because of small numbers and differing methodologies. However, their results suggest that neurodevelopmental anomalies may characterize the brains of children with ADHD and that deviations in structure may be associated and related to prenatal deviations in cellular migration and maturation (Geschwind & Galaburda, 1985; Hynd & Semrud-Clikeman, 1989). The use of three-dimensional MRI as well as subgrouping ADHD may provide further information about the neurodevelopmental underpinnings of this disorder. Investigators have hypothesized that the brain areas implicated in the ADHD:combined subtype are in the anterior regions involved in motor activity modulation, whereas ADHD:predominately inattentive is hypothesized to be related to central-posterior regions affecting attention (McBurnett, Pfiffner, & Frick, 2001; Waldman, Lilienfeld, & Lahey, 1995).

These hypotheses have not been tested empirically, and MRI results may well show anatomical differences among the subtypes. As there appear to be differences in behavior between the subtypes, it is likely that the subtypes differ neurologically. Moreover, the finding that the subtypes of attention-deficit disorder respond differentially to medication further indicates that differing brain mechanisms may be implicated in the subtypes (Millich, Ballentine, & Lynam, 2001). Therefore, since the MRI studies to date have mostly utilized subjects with ADHD:combined subtype, the next step would be to look at subjects with differing expressions of ADHD to determine if neuroanatomical differences exist between the subtypes.

Functional imaging has found less activation in the dorsolateral regions of the frontal lobes as well as in the ACC (Pliszka et al., 2006). FMRI research in ADHD has primarily focused on inhibitory control. FMRI studies have shown that healthy adults engage the right dorsolateral prefrontal cortex (DLPFC) on inhibition tasks, particularly for successful inhibitions (Garavan, Ross, Murphy, Roche, & Stein, 2002; Rubia, Smith, Brammer, & Taylor, 2003).

While the anterior cingulate cortex (ACC) is also involved when inhibition is successful, it is involved more strongly for unsuccessful inhibitions. After unsuccessful inhibitions, activation of the ACC may increase prefrontal cortex activity on future trials, thus increasing cognitive control after errors (Kerns et al., 2004). After unsuccessful inhibitions, healthy children show more activation in the ACC (Pliszka et al., 2006) or the posterior cingulate cortex (Rubia, Smith, Brammer, Toone, & Taylor, 2005), compared to children with ADHD.

Findings for children with ADHD in the frontal cortex have been more inconsistent, with studies showing both increased (Durston, 2003; Schulz et al., 2004; Vaidya et al., 1998) and decreased activity (Rubia et al., 1999; Rubia et al., 2005) relative to controls during inhibitory tasks. Recently, Pliszka et al. (2006) obtained fMRI on children with ADHD (both medication naïve and those with a history of chronic stimulant treatment) to healthy controls while they performed a stop signal task. ADHD children had increased right PFC activity relative to controls during both successful and unsuccessful inhibitions. During unsuccessful inhibitions, healthy controls showed greater ACC and left ventrolateral PFC (VLPFC) activation than ADHD children. Studies of children with ADHD on social understanding types of tasks using fMRI, volumetric MRI, or DTI are not present in the literature.

To date, most of the research has either used samples of children with ADHD: combined subtype or without specifying the subtype. It is rare that children with ADHD: predominately inattentive type have been studied, and no studies were identified with children with ADHD: PI using functional neuroimaging. We were unable to locate any published studies utilizing DTI in children with ADHD: PI. The one study we located utilized children with ADHD: C with some on medications and others not (Ashtari et al., 2005). Findings were of decreased fractional anisotropy (FA) in the areas of the left cerebellum, right striatal region, and in the right premotor areas. It will be of interest to study children with ADHD: PI using fMRI and DTI to determine whether these findings also apply to the PI subtype.

Structural imaging in autistic spectrum disorder (ASD) has found a variety of cerebral anomalies. Some findings have indicated differences in the corpus callosum, particularly in thinning and in the midsagittal area and in white matter density in persons with ASD (Chung, Dalton, Alexander, & Davidson, 2004; Hardan, Minshew, & Keshavan, 2000; Vidal et al., 2006). These differences have been linked to difficulties with language processing and in working memory (Just, Cherkassky, Keller, & Minshew, 2004; Koshino et al., 2005). Structural imaging findings show increased brain size in children with ASD, specifically enlargement in the regions of the parietal, temporal, and occipital lobes found for children (Courchesne, Carper, & Akshoomoff, 2003; Piven, Arndt, Bailey, & Andreasen, 1996). The enlargement is due to greater volumes of white matter, but not gray (Filipek et al., 1997). For adolescents and adults with autism, increased brain size was not found, but increased head circumference was identified (Aylward, Minshew, Field, Sparks, & Singh, 2002). Findings have also implicated the caudate in children with ASD. Sears et al.(1999) found increased volume bilaterally in the caudate of adolescents and young adults with HFA. Caudate enlargement was found to be proportional to the increased total brain volumes also identified. To determine whether these findings were specific to the sample studied, the results were replicated with a different sample of participants with ASD. Thus, it will be of interest to evaluate the caudate volume in children and adolescents to determine whether the finding of larger caudates bilaterally are present in younger participants but not in older participants. Previous studies have found that the caudate volume in typically developing children decreases at puberty (Giedd, 2004). For the Sears et al. (1999) study this decrease did not occur. Increased caudate volumes were also not present for the ADHD population (Semrud-Clikeman et al., 2006).

In typically developing adults, right-sided activation has been found for the FG, the STG, and the amygdala when viewing social faces (Schultz et al., 2003). These findings seem to corroborate the hypothesis that the right hemisphere is more heavily involved in social processing than the left (Winner, Brownell, Happe, Blum, & Pincus, 1998). Studies of adults with high functioning autism (HFA) or Asperger Syndrome (AS) have found differences in these regions using structural and functional imaging with finding less activation or smaller volumes (Abell et al., 1999; Baron-Cohen et al., 1999). Some studies have found underactivation in the prefrontal areas in participants with ASD (Schultz et al., 2000) with higher activation in controls particularly in the left prefrontal regions. Significant metabolic reduction has been found in the anterior cingulate gyrus, an area important for error monitoring, and in the frontostriatal networks involved in visuospatial processing (Ohnishi et al., 2000; Silk et al., 2006). The amygdala and striatal regions have been implicated in processing of emotional facial expressions (Canli, Sivers, Whitfield, Gotlib, & Gabrieli, 2002; Killgore & Yurgelun-Todd, 2005; Yang et al., 2002). Studies involving the amygdala have found that patients with bilateral lesions show difficulty in judging emotional facial expressions (Bechara, Damasio, & Damasio, 2003). Faces that were negative in nature were found to produce slower activation than those which were neutral in controls (Monk et al., 2003; Simpson, Öngür, Akbudak, Conturo, & Ollinger, 2000). Positive stimuli was found to increase activity in the amygdala with happy expressions resulting in faster response times (Canli et al., 2002; Pessoa, Kastner, & Ungerleider, 2002; Somerville, Kim, Johnstone, Alexander, & Whalen, 2004).

Hare, Tottenham, Davidson, Glover, and Casey (2005) found longer latencies for negative facial expressions when the amygdale was activiated in normal adult controls, while activation of the caudate was present when the participant was avoiding positive information. These findings point to the neural networks that underlie the processing of positive and negative emotions. It would appear crucial to determine how emotional cues may influence behavior that interferes with this processing, as is frequently seen in children and adults with ASD. Similarly, Ashwin, Wheelwright, and Baron-Cohen (2006) studied the amygdala in adults with HFA and AS using fMRI when viewing fearful (high and low fear) and neutral faces. Findings indicated that the control sample showed more activation in the left amygdala and left orbitofrontal regions, compared to the HFA/AS group when viewing the fearful faces. In contrast, the HFA/AS group showed more activation in the anterior cingulate and the superior temporal cortex when viewing all of the faces.

Specific areas implicated in ASD using fMRI include the superior temporal gyrus (STG) (Allison, Puce, & McCarthy, 2000; Baron-Cohen et al., 1999) and the amygdala (Amaral, Price, Pitkanan, & Carmichael, 1992; Carmichael & Price, 1995). When a child with ASD is asked to view faces, functional imaging has found less activation in the right fusiform face gyrus (FG) (Critchley et al., 2000; Dierks, Bolte, Hubl, Lanfermann, & Poustka, 2004; Pierce, Muller, Ambrose, Allen, & Courchesne, 2001; Schultz et al., 2001; Schultz, Romanski, & Tsatsanis, 2000). The fusiform face gyrus appears to be selectively engaged when the person views faces. Children with ASD appear to pay less attention to the face and may not activate this region in the same manner as typically developing children (Schultz et al., 2003).

Differences in activation have not been solely found for viewing of faces. Just et al. (2004) found more cortical activation in children with HFA when processing verbal material compared to a control group. This finding may indicate that more brain resources are needed to process verbal information in children with ASD. This inefficiency may slow down their processing and take resources away from understanding the intent or speaker’s prosody or interpreting of facial expressions. Kana, Keller, Cherkassky, Minshew, and Just (2006) found that children with HFA show poorly synchronized neural connectivity when asked to process material using their imagination. Moreover, the children with ASD were also found to use parietal and occipital regions associated with imagery for tasks that had low imagery as well as high imagery requirements. Controls only utilized these regions for the high imagery tasks.

A finding of difference in activation when participants with ASD view faces is important for our understanding of ASD. Animal studies have found that the temporal cortex responds to the perceptual aspects of social stimuli (Hasselmo, Rolls, & Baylis, 1989; Perrett & Mistlin, 1990). Human studies have also found a relationship between the perception of eyes and mouths in the superior temporal cortex (Adolphs, 2001; Haxby, Hoffman, & Gobbini, 2000). The superior temporal lobe may be involved in perceiving social stimuli with the connections to the amygdala and frontal lobes for interpretating these stimuli. Findings from participants with ASD have repeatedly shown that these areas are important to process facial expressions.

There have been few studies using DTI in children with ASD. Alexander et al. (2007) used DTI with ASD children and found reduced volumes in the corpus callosum and reduced fractional anisotropy (FA) of the genu, splenium, and total corpus callosum in children with ASD. Medication status and the presence of comorbid diagnoses were not found to have an effect on the DTI measures. Similarly, Boger-Megiddo et al. (2006) scanned young children with ASD and found that the corpus callosum area was disproportionately small compared to total brain volume for these children, compared to typically developing children. Children with a more severe form of ASD showed the smallest corpus callosal area. Future research needs to utilize quantitative imaging to more fully capture the relation between the corpus callosum connectivity and ASD. Taken together these findings suggest that children with ASD may be inefficient in their use of neural networks and that their networks show poor connectivity compared to same-aged peers.

Positron emission tomography (PET) imaging can provide a direct measure of cerebral glucose metabolism. Zametkin et al. (1990) studied adults with ADD and hyperactivity (ADD/H) through the use of PET with 5–6 mm resolution. A radioactive tracer injected into the subjects was partially metabolized by the neurons and emitted radiation images through the PET scanner. These 25 ADD/H adults had onset of ADD/H in childhood and were also parents of ADD/H children. Subjects with conduct disorder or those using stimulant medication in childhood were eliminated from the sample. Cerebral glucose metabolism (CGM) was measured in 60 regions and the ADD/H group was found to have a CGM approximately 8 percent lower than the controls, with 50 percent of the regions showing significantly lower metabolism than the controls (p <0.05). The regions with significant hypometabolism were in the cingulate, right caudate, right hippocampal, and right thalamic regions. In addition, reduced metabolism was found in the left parietal, temporal, and rolandic structures for the ADHD subjects. Statistical difficulties are present in this study in that 60 t-tests were used for analysis, introducing the possibility of inflated significant results due to experiment-wise error. Zametkin has written additional information about his study and, when his results were studied using the Bonferonni method of correction, the areas that continued to be significant were in the superior prefrontal and premotor regions (Zametkin & Cohen, 1991).

In contrast to PET scans, single photon emission tomography (SPECT) is a direct measure of regional cerebral blood flow (rCBF) with neuronal glucose metabolism inferred from the rCBF. Thus SPECT is a vascular measure, whereas PET is a neuronal measure. Results from PET and SPECT studies may not be directly comparable because of this difference in acquisition. Lou, Hendriksen, and Bruhn (1984) utilized SPECT in 13 children with dysphasia and ADHD. Only two of these children had pure ADHD. The remaining 11 had variations of dysphasia, mental retardation, and visual-spatial delays, although two did not have a diagnosis of ADHD. All children were attending a school for children with learning disorders. The contrast group was selected from the siblings of these children. Given that Lou et al. (1984) did not provide information about the functioning of these children and of the finding of substantial learning and attentional problems in siblings of ADHD children, it may be that the contrast group was not a true control group. Despite these methodological problems, the findings from this study are intriguing. All children with ADHD (n = 11) were found to have less blood perfusion in the white matter of the middle frontal regions, including the region of the genu of the corpus callosum.

Hypoperfusion was also found in the caudate region. The caudate region was inferred from the scan, as the caudate nucleus is smaller than the 17 mm resolution of the scanner. Therefore, the hypoperfusion may also have involved the lateral wall of the anterior horn of the lateral ventricle as well as the caudate, in addition to other areas of the basal ganglia. When methylphenidate was administered to the ADHD group, increased flow was found in the central region of the brain most likely encompassing the basal ganglia region. Theoretically these findings make sense. Frontal regions have feedback loops into the caudate. However, the primary output for the caudate is the thalamus, and it is not clear why the thalamus was not found to be hypoperfused also. The thalamic region was found to have lower perfusion in only three of the children; two with mental retardation and one without ADHD.

Lou, Henriksen, Bruhn, Borner, and Nielson (1989) expanded their study to include 19 additional subjects, six of whom were identified as “pure ADHD.” The remaining 13 had ADHD as their primary diagnosis, along with other neurological deficits. Of these 13, three had mild mental retardation, nine had dysphasia, and six had visuospatial problems. The control group was the same as in the initial study. In this follow-up study, the mesial frontal region did not show significant hypoperfusion as was found in the 1984 study. Instead, hypoperfusion in the right striatal area encompassing the anterior corpus callosum, internal capsule, and part of the thalamus in addition to the caudate was found in the children with ADHD. Hyperperfusion was found in the occipital region for the group with ADHD and in the left anterior parietal and temporal regions. Methylphenidate normalized perfusion in the left striatal area but not the right, as well as in the association regions in the posterior region of the brain.

In a 1991 letter to the New England Journal of Medicine, Lou stated that statistical analysis was not performed on the SPECT scans in the initial 1984 study. Scans were analyzed through “visual analysis,” and when the prefrontal region measures were statistically analyzed with corrections for multiple analyses performed, the regional difference was not statistically different. Therefore, Lou et al.'s (1989) statistically significant finding of hypoperfusion in the striatal region and hyperperfusion in the posterior regions may implicate these regions in ADHD. With statistical correction Zametkin et al. (1990) reported that the superior prefrontal and premotor areas continued to show significant differences. Further study with controlled statistical procedures, as well as subjects without comorbid diagnoses, would clarify knowledge in this area, and it is hoped that these researchers will pursue this end.