Statistical issues in the analysis of neuroimages

Acute tryptophan depletion potentiates 3,4-methylenedioxymethamphetamine-induced cerebrovascular hyperperfusion in adult male Wistar rats

The serotonergic (5-hydroxytryptamine; 5-HT) dysfunction found in depression may affect not only brain function (mood) but also cerebrovascular control. Similar, but possibly occult, disturbances may also be induced by 3,4-methylenedioxymethamphetamine-induced neurotoxicity (MDMA, or "ecstasy"). Acute tryptophan depletion (ATD) is widely used to identify vulnerability to depression, and we hypothesized that repeated MDMA administration would increase the sensitivity of rats to this acute serotonergic challenge. In this study, male Wistar rats were injected with MDMA (20 mg kg(-1), twice daily for 4 days) and challenged 3 weeks later with ATD, induced by intragastric administration of a nutritional mixture with tryptophan (TRP) removed. Cerebral metabolism (CMRG) and blood flow (CBF) were measured in parallel groups of animals following ATD by using quantitative [(14)C]2-deoxyglucose and [(14)C]iodoantipyrine autoradiographic techniques, respectively. A significant reduction in paroxetine binding to 5-HT transporter sites in MDMA-treated rats indicated 5HT terminal depletion, whereas the plasma TRP/sum large neutral amino acids ratio was reduced by 40% following ATD. Under all experimental conditions, the normal close correlation between CBF and metabolic demand was maintained. However, a global analysis of all brain regions revealed a significant decrease in the overall ratio of CBF to CMRG after ATD in control animals, whereas a higher ratio was observed after ATD in the MDMA-treated group. This increase in blood flow relative to cerebral metabolism suggests an ATD-induced loss of cerebrovascular tone in MDMA-treated animals that could have pathophysiological consequences and might conceivably contribute to the behavioral dysfunction of depression.

Reduced cerebral blood flow but elevated cerebral glucose metabolic rate in erythropoietin overexpressing transgenic mice with excessive erythrocytosis

To examine the impact of excessive erythrocytosis on local cerebral blood flow (CBF) and cerebral glucose metabolic rate (CMR(glc)), we made use of our constitutively erythropoietin (Epo)-overexpressing transgenic mouse line (tg-6) that reach a mean hematocrit of 0.87. Compared with wild-type (wt) control siblings, CBF decreased by 44% in tg-6 mice, while upon hemodilution (tg-6-HD) to a physiologic hematocrit (e.g., 0.44) tg-6-HD mice returned the CBF to wt levels. Cerebral blood flow was determined in another transgenic mouse line that overexpresses human Epo in the brain only (tg-21): CBF increased by 17% compared with wt controls. However, oxygen delivery was similar in all four mouse groups tested (wt, tg-6, tg-6-HD and tg-21). Mean CMR(glc) was higher in tg-6 (+72%), tg-6-HD mice (+43%) and tg-21 (+22%) than in wt mice. Local CMR(glc) was higher in all 40 brain regions in tg-6 but only in 15 and 8 regions in tg-6-HD and tg-21 mice. These results show that prolonged increases in hematocrit did not alter cerebral oxygen delivery at a decreased CBF and increased CMR(glc). Hemodilution suggests that high blood viscosity is a cause of the decrease in CBF and partly of the increase in CMR(glc). Cerebral glucose metabolic rate may also be increased by a direct effect of Epo in the brain (tg-21 mice).

Statistical parametric mapping of hypoxic tissue identified by [(18)F]fluoromisonidazole and positron emission tomography following acute ischemic stroke

Positron emission tomography (PET) and the ligand [(18)F]fluoromisonidazole ((18)F-FMISO) have been used to image hypoxic tissue in the brain following acute stroke. Existing region of interest (ROI)-based methods of analysis are time consuming and operator-dependent. We describe and validate a method of statistical parametric mapping to identify regions of increased (18)F-FMISO uptake. The (18)F-FMISO PET images were transformed into a standardized coordinate space and intensity normalized. Then t statistic maps were created using a pooled estimate of variance. Statistical inference was based on the theory of Gaussian Random Fields. We examined the homogeneity of variance in normal subjects and the influence of normalization by mean whole brain activity versus mean activity in the contralateral hemisphere. Validity of the distributional assumptions inherent in parametric analysis was tested by comparison with a non-parametric method. The results of parametric analysis were also compared with those obtained with the existing ROI-based method. Variance in uptake at each voxel in normal subjects was homogeneous and not affected by mean voxel activity or distance from the centre of the image. The method of normalization influenced results significantly. Normalization by whole brain mean activity resulted in a smaller volume of tissue being classified as hypoxic compared to normalisation by mean activity in the contralateral hemisphere. The ROI-based method was subject to interobserver variability with a coefficient of variability of 16%. The volumes of hypoxic tissue identified by parametric and nonparametric methods were highly correlated (r = 0.99). These findings suggest that using a pooled variance and contralateral hemisphere normalisation, statistical parametric mapping can be used to objectively identify regions of increased (18)F-FMISO uptake following acute stroke in individual subjects.

Estimation of the number of "true" null hypotheses in multivariate analysis of neuroimaging data

The repeated testing of a null univariate hypothesis in each of many sites (either regions of interest or voxels) is a common approach to the statistical analysis of brain functional images. Procedures, such as the Bonferroni, are available to maintain the Type I error of the set of tests at a specified level. An initial assumption of these methods is a "global null hypothesis," i.e., the statistics computed on each site are assumed to be generated by null distributions. This framework may be too conservative when a significant proportion of the sites is affected by the experimental manipulation. This report presents the development of a rigorous statistical procedure for use with a previously reported graphical method, the P plot, for estimation of the number of "true" null hypotheses in the set. This estimate can then be used to sharpen existing multiple comparison procedures. Performance of the P plot method in the multiple comparison problem is investigated in simulation studies and in the analysis of autoradiographic data.

Selection of an adaptive test statistic for use with multiple comparison analyses of neuroimaging data

Statistical analysis of neuroimages is commonly approached with intergroup comparisons made by repeated application of univariate or multivariate tests performed on the set of the regions of interest sampled in the acquired images. The use of such large numbers of tests requires application of techniques for correction for multiple comparisons. Standard multiple comparison adjustments (such as the Bonferroni) may be overly conservative when data are correlated and/or not normally distributed. Resampling-based step-down procedures that successfully account for unknown correlation structures in the data have recently been introduced. We combined resampling step-down procedures with the Minimum Variance Adaptive method, which allows selection of an optimal test statistic from a predefined class of statistics for the data under analysis. As shown in simulation studies and analysis of autoradiographic data, the combined technique exhibits a significant increase in statistical power, even for small sample sizes (n = 8, 9, 10).

Mild and moderate hypothermia (alpha-stat) do not impair the coupling between local cerebral blood flow and metabolism in rats

BACKGROUND AND PURPOSE

The effects of hypothermia on global cerebral blood flow (CBF) and glucose utilization (CGU) have been extensively studied, but less information exists on a local cerebral level. We investigated the effects of normothermic and hypothermic anesthesia on local CBF (LCBF) and local CGU (LCGU).

METHODS

Thirty-six rats were anesthetized with isoflurane (1 MAC) and artificially ventilated to maintain normal PaCO(2) (alpha-stat). Pericranial temperature was maintained normothermic (37.5 degrees C, n=12) or was reduced to 35 degrees C (n=12) or 32 degrees C (n=12). Pericranial temperature was maintained constant for 60 min until LCBF and LCGU were measured with autoradiography. Twelve conscious rats served as normothermic control animals.

RESULTS

Normothermic anesthesia significantly increased mean CBF compared with conscious control animals (29%, P<0.05). Mean CBF was reduced to control values with mild hypothermia and to 30% below control animals with moderate hypothermia (P<0.05). Normothermic anesthesia reduced mean CGU by 44%. No additional effects were observed during mild hypothermia. Moderate hypothermia resulted in a further reduction in mean CGU (41%, P<0.05). Local analysis showed linear relationships between LCBF and LCGU in normothermic conscious (r=0.93), anesthetized (r=0.92), and both hypothermic groups (35 degrees C r=0. 96, 32 degrees C r=0.96, P<0.05). The LCBF-to-LCGU ratio increased from 1.5 to 2.5 mL/micromol during anesthesia (P<0.05), remained at 2.4 mL/micromol during mild hypothermia, and decreased during moderate hypothermia (2.1 mL/micromol, P<0.05).

CONCLUSIONS

Anesthesia and hypothermia induce divergent changes in mean CBF and CGU. However, local analysis demonstrates a well-maintained linear relationship between LCBF and LCGU during normothermic and hypothermic anesthesia.

Statistical limitations in functional neuroimaging. I. Non-inferential methods and statistical models

Functional neuroimaging (FNI) provides experimental access to the intact living brain making it possible to study higher cognitive functions in humans. In this review and in a companion paper in this issue, we discuss some common methods used to analyse FNI data. The emphasis in both papers is on assumptions and limitations of the methods reviewed. There are several methods available to analyse FNI data indicating that none is optimal for all purposes. In order to make optimal use of the methods available it is important to know the limits of applicability. For the interpretation of FNI results it is also important to take into account the assumptions, approximations and inherent limitations of the methods used. This paper gives a brief overview over some non-inferential descriptive methods and common statistical models used in FNI. Issues relating to the complex problem of model selection are discussed. In general, proper model selection is a necessary prerequisite for the validity of the subsequent statistical inference. The non-inferential section describes methods that, combined with inspection of parameter estimates and other simple measures, can aid in the process of model selection and verification of assumptions. The section on statistical models covers approaches to global normalization and some aspects of univariate, multivariate, and Bayesian models. Finally, approaches to functional connectivity and effective connectivity are discussed. In the companion paper we review issues related to signal detection and statistical inference.

Functional connectivity in depressive, obsessive-compulsive, and schizophrenic disorders: an explorative correlational analysis of regional cerebral metabolism

In order to investigate the changes in functional relationships between brain regions in three psychiatric disorders, an exploratory statistical analysis of the regional cerebral metabolic rates for glucose (rCMRglu) obtained with positron emission tomography (PET) was performed. Correlations between various rCMRglu were computed in control, depressive, obsessive-compulsive, and schizophrenic groups to determine whether alterations of the correlation pattern were found in the psychiatric groups. The absence of correlation between left and right frontal lobes was common to the three psychiatric groups studied. Depressive patients recovered a better frontal interhemispheric coupling after successful treatment and the alterations in the depressed state also involved the relation between the right temporal cortex and the right thalamus. Obsessive-compulsive patients had not only lateral frontal dysfunction but also alterations in the functional relationships between cortex and thalami. In schizophrenic patients, the modifications of regional cerebral metabolic correlations involved both anterior and posterior cortical regions. Thus, although the relationship between left and right frontal lobes was altered in three psychiatric disorders, the pattern of functional connectivity between frontal regions, posterior cortical and subcortical regions differed depending on the diagnostic group.

Serotonin in the regulation of brain microcirculation

Manipulation of brainstem serotonin (5-HT) raphe neurons induces significant alterations in local cerebral metabolism and perfusion. The vascular consequences of intracerebrally released 5-HT point to a major vasoconstrictor role, resulting in cerebral blood flow (CBF) decreases in several brain regions such as the neocortex. However, vasodilatations, as well as changes in blood-brain barrier (BBB) permeability, which are blocked by 5-HT receptor antagonists also can be observed. A lack of relationship between the changes in flow and metabolism indicates uncoupling between the two variables and is suggestive of a direct neurogenic control by brain intrinsic 5-HT neurons on the microvascular bed. In line with these functional data are the close associations that exist between 5-HT neurons and the microarterioles, capillaries and perivascular astrocytes of various regions but more intimately and/or more frequently so in those where CBF is altered significantly following manipulation of 5-HT neurons. The ability of the microvascular bed to respond directly to intracerebrally released 5-HT is underscored by the expression of distinct 5-HT receptors in the various cellular compartments of the microvascular bed. Thus, it appears that while some 5-HT-mediated microvascular functions involve directly the blood vessel wall, others would be relayed through the perivascular astrocyte. The strategic localization of perivascular astrocytes and the different 5-HT receptors that they harbor strongly emphasize their putative pivotal role in transmitting information between 5-HT neurons and microvessels. It is concluded that the cerebral circulation has full capacity to adequately and locally adapt brain perfusion to changes in central 5-HT neurotransmission either directly or indirectly via the neuronal-astrocytic-vascular tripartite functional unit. Dysfunctions in these neurovascular interactions might result in perfusion deficits and might be involved in specific pathological conditions.

Cognitive activation during PET: a case study of monozygotic twins discordant for closed head injury

Using positron emission tomography (PET), we studied physiological changes in a patient with frontal lobe damage resulting from closed head injury by assessing his regional cerebral blood flow (rCBF) pattern relative to that of his uninjured monozygotic (MZ) cotwin and against normal variability between cotwins in 10 comparably-aged, uninjured MZ twin pairs. rCBF was measured during the Wisconsin Card Sorting Test (WCST) and a sensorimotor control task. Differences between the index twins in rCBF activation (WCST-control) in regions determined on coplanar MRIs were compared to within-pair differences in the control group using the Z-statistic. Activation differences between the index twins extending beyond normal variability were found in two regions-relative to his uninjured cotwin, the injured twin showed less activation in the inferior portion of the left inferior frontal gyrus while showing greater activation in the left hippocampus (P < 0.04). This pattern contrasts with that of normal subjects during WCST performance of augmented rCBF in the prefrontal cortex and not in the hippocampus. These rCBF differences appear to reflect the utilization of different neural systems when performing a frontal-lobe task. The use of the hippocampus by the injured twin might reflect an attempt to compensate for compromised frontal lobe functioning.

Functional lateralization of the prefrontal cortex during traditional frontal lobe tasks

We attempted to identify brain regions functionally lateralized during cognitive tasks traditionally linked to the prefrontal cortex (PFC) by measuring regional cerebral blood flow with H2(15)O positron emission tomography (PET). Fourteen normal subjects were scanned six times while performing six different cognitive conditions comprising three task paradigms putatively sensitive to PFC integrity: the Wisconsin Card Sort (WCS), Delayed Response Alternation (DA), and the Spatial Delayed Response (SDR) Tasks, and three matched sensorimotor control tasks. Multivariate and repeated measures analyses indicated that for all three cognitive paradigms there were no significant hemisphere, hemisphere-by-condition, or hemisphere-by-region effects. However, with more liberal statistical comparison (paired t tests), the superior frontal gyrus showed lateralization during both the WCS and SDR tasks (both R > L). These results suggest that, although some asymmetries may be found using liberal analyses, there is less evidence of lateralized brain function during performance of these tasks of PFC function, than in language and motor systems. Implications for testing PFC function in neuropsychiatric groups are discussed.

Coupling between local cerebral blood flow and metabolism after hypertonic/hyperoncotic fluid resuscitation from hemorrhage in conscious rats

The effects of small volume hypertonic/hyperoncotic fluid resuscitation from hemorrhage on brain metabolism and blood flow were evaluated by autoradiographic techniques with high spatial resolution. The data were compared to fluid resuscitation with a volume equal to shed blood of isotonic 6% hydroxyethyl starch solution (HES) and a control group without hemorrhage and fluid resuscitation (n = 6 in each group). In conscious rats, volume-controlled hemorrhage for 30 min (30 mL/kg body weight, resulting in a blood loss of approximately 50% of the circulating blood volume) was followed by intravenous infusion of a hypertonic/hyperoncotic saline hydroxyethyl starch solution (HTHO; 7.5% saline/10% hydroxyethyl starch, 4.0 mL/kg body weight). Local cerebral blood flow (LCBF) and local cerebral glucose utilization (LCGU) were measured in 34 brain structures 2 h after fluid resuscitation by means of the quantitative autoradiographic iodo [14C]antipyrine and 2-[14C]-deoxy-D-glucose methods. Compared to the untreated control group, LCBF increased significantly in all brain regions analyzed after fluid resuscitation with HTHO (mean, +63%) or HES (mean, +56%). The increases in LCBF after fluid resuscitation were sufficient to restore cerebral oxygen delivery to the level calculated for the untreated control group. LCGU was unchanged after fluid resuscitation. The close relationship between LCGU and LCBF observed in the control group (r = 0.95) was preserved after hemorrhage and fluid resuscitation with HTHO (r = 0.97) and HES (r = 0.96), although the LCBF-to-LCGU ratio was reset to a higher level (1.5 mL/mumol in the control group and 2.7 mL/mumol after fluid resuscitation with HTHO and HES, P < 0.05). We conclude that the increase in LCBF compensates for the reduction of arterial oxygen content to maintain cerebral oxygen delivery. Therefore, "small volume resuscitation" appears to be as effective as resuscitation with large volumes of isotonic HES in meeting the circulatory and metabolic demands of the brain tissue within the first 2 h after fluid resuscitation from hemorrhage.

Commentary and opinion: I. Principal component analysis, variance partitioning, and "functional connectivity"

We briefly review the need for careful study of "variance partitioning" and "optimal model selection" in functional positron emission tomography (PET) data analysis, emphasizing the use of principal component analysis (PCA) and the importance of data analytic techniques that allow for heterogeneous spatial covariance structures. Using an [15O]water dataset, we demonstrate that--even after data processing--the intrasubject signal component of primary interest in baseline activation studies constitutes a very small fraction of the intersubject variance. This small intrasubject variance component is subtly but significantly changed by using analysis of covariance instead of scaled subprofile model processing before applying PCA. Finally, we argue that the concept of "functional connectivity" should be interpreted very generally until the relative roles of inter- and intrasubject variability in both disease and normal PET datasets are better understood.

[18F]FDG PET study in obsessive-compulsive disorder. A clinical/metabolic correlation study after treatment

BACKGROUND

We used [18F]FDG and PET in patients with obsessive-compulsive disorder (OCD) to evaluate cerebral metabolic involvement before and after treatment with serotonin-specific reuptake inhibitors.

METHOD

In 11 untreated, drug-free adults, regional cerebral metabolic rate for glucose (rCMRglu) was compared with that of 15 age-matched normal controls.

RESULTS

rCMRglu values were significantly increased in the cingulate cortex, thalamus and pallidum/putamen complex. After treatment a significant improvement in obsessive-compulsive symptoms on the Y-BOC scale (t = 3.59, P < 0.01) was associated with a significant bilateral decrease of metabolism in the whole cingulate cortex (P < 0.001). Clinical and metabolic data were significantly intercorrelated (Kendall's tau = 0.65; P < 0.01).

CONCLUSIONS

These findings indicate that OCD is associated with functional hyperactivity of a selected neuronal network and that treatment to reduce symptoms may have a selective neuromodulatory effect on cingulate cortex.

Difference of regional cerebral metabolic pattern between presenile and senile dementia of the Alzheimer type: a factor analytic study

Positron emission tomographic studies using metabolic ratios to represent the contrast between certain brain regions have shown differences of cerebral glucose metabolism between presenile and senile dementia of Alzheimer type (DAT). In this study, factors for regional variance of cerebral metabolism in 61 patients with probable Alzheimer's disease were obtained using principal component analysis (PCA). The difference between two age groups of DAT patients according to age at onset < or = 65 years (n = 30) and > 65 years (n = 31) were investigated. A PCA using nine pairs of cerebral regions normalized to cerebellum for reduction of intersubject variability identified five regional factors for cerebral metabolism: one each for the left and right temporo-parietal cortex, and the other three for frontal cortex, primary visual and sensorimotor cortex, and mesial temporal sensorimotor cortex, and mesial temporal cortex. The age groups differed significantly only in the factor for the right temporo-parietal cortex, with a more prominent metabolic reduction in presenile than senile DAT. Severity of dementia was correlated with the factors for the right and left temporo-parietal cortex. The effect of age at onset on glucose metabolism of the right temporo-parietal cortex was independent from that of dementia severity in DAT.

Inhibition of nitric oxide synthesis: effects on cerebral blood flow and glucose utilisation in the rat

The consequences of inhibition of nitric oxide synthesis on local CBF and glucose utilisation have been studied in the conscious rat using the specific nitric oxide synthase inhibitor Ng-nitro-L-arginine methyl ester (L-NAME; 30 mg kg-1 i.v.). Local CBF and glucose utilisation were assessed with the [14C]iodoantipyrine and the 2-deoxy-D-[14C]glucose autoradiographic techniques, respectively. L-NAME induced prolonged (> 3 h) reductions in local CBF throughout the CNS with concomitant increases in arterial blood pressure. For example, 1 h post L-NAME, CBF dropped from 79 +/- 4 to 45 +/- 1 ml 100 g-1 min-1 in cerebellum, from 76 +/- 4 to 47 +/- 2 ml 100 g-1 min-1 in medulla oblongata, and from 117 +/- 6 to 72 +/- 2 ml 100 g-1 min-1 in cortex. L-NAME produced sustained elevations (e.g., 46 +/- 2 mm Hg at 1 h after bolus administration) in mean arterial blood pressure throughout the period evaluated. Despite evidence implicating nitric oxide in neuronal signalling, L-NAME did not significantly influence CNS functional activity, as measured by local rates of glucose utilisation, in any neuroanatomical region examined. Consequently, the normal ratio of blood flow to glucose use throughout the brain was significantly reduced in the presence of L-NAME, although the hierarchy of blood flow levels in different neuroanatomical regions was preserved. These results are consistent with the involvement of nitric oxide in the tonic control of cerebral tissue perfusion.

High metabolic activity demonstrated by positron emission tomography in human auditory cortex in case of deafness of early onset

Glucose metabolism has been studied in the auditory cortex of human subjects with deafness of early onset, and compared to normal subjects with ears plugged. The metabolism in the auditory cortex and in the association auditory cortex was higher in deaf subjects than in normal subjects. This result is compared to similar observations that we made previously in the visual cortex of human subjects with blindness of early onset.

Local cerebral blood flow measured by xenon-enhanced CT during cryogenic brain edema and intracranial hypertension in monkeys

We developed a closed-skull model of freeze injury-induced brain edema, a model classically thought to produce vasogenic edema, and observed the natural course of changes in edema and blood flow using xenon-enhanced computed tomography (CT) in five rhesus monkeys before and for up to 6 h post insult. Intracranial pressure (ICP) gradually rose throughout the duration of the experiment. CT scans and CBF images permitted direct observation of the evolution of the lesion and revealed early ischemia in the periphery of the injury zone that progressed over time in association with edema. Frequency histogram analysis of local CBF (ICBF) demonstrated subtle but potentially important changes in distribution of ICBF between and within hemispheres at various times post insult. Changes in ICBF distribution were phasic and dissociated from increases in ICP in the latter stages of injury. The Xe/CT CBF method can be used to evaluate the effects of injury and therapy on CBF in this and other models of acute brain injury.

Effects of dorsal raphe nucleus stimulation on cerebral blood flow and flow-metabolism coupling in the conscious rat

In the present study, we have investigated the effects of an activation of the ascending serotonergic pathway on the cerebral blood supply to a number (63) of well-defined neuroanatomical structures. To this end, we have measured the local cerebral blood flow during electrical stimulation of the dorsal raphe nucleus. Measurement of regional blood flow was performed in the conscious rat through the use of the [14C]iodoantipyrine autoradiographic technique. Stimulation of the dorsal raphe nucleus induced increases (> 15% compared to control) in cerebral blood flow in 17 structures of which statistical significance (P < 0.05) was achieved in nine; raphe stimulation significantly decreased flow in three regions. The greatest increases (+71 and +46%) were found in the frontal sensorimotor and posterior parietal cortices. Other increases were noted in relay stations of the extrapyramidal and limbic systems. Stimulation induced a decrease in two regions of the primary auditory system and in the lateral habenular nucleus. These results show that activation of the serotonergic pathway in the conscious rat effects regional cerebral blood flow heterogeneously, differing from the widespread increase in glucose utilization that we previously observed using the same experimental paradigm. Statistical analyses indicated that activation of the dorsal raphe nucleus resulted in a global modification of the flow-metabolism ratio. Moreover, in 19 out of 31 regions analysed, this ratio is significantly altered as compared to control. The dichotomy between raphe-induced changes in flow and glucose-metabolism could be explained by one or both of two hypotheses; firstly there could be a direct serotonergic innervation of cerebral resistance vessels; secondly, during raphe stimulation it could be that glucose use is not the primary determinant of tissue perfusion.

A regional covariance approach to the analysis of functional patterns in positron emission tomographic data

This article provides a complete description of the subprofile scaling model (SSM) approach to the analysis of positron emission tomography (PET) data. The goals and assumptions underlying the development of SSM are outlined, and its strengths and weaknesses are discussed. It is demonstrated that all obtainable information about regional ratios can, in theory, be derived from the SSM regional covariance patterns. A general constraint on the ability to effectively remove global variation while identifying region-specific information about PET data sets is outlined and discussed within the SSM framework. Finally, an extension of the SSM technique to the generation of disease-specific covariance patterns is demonstrated for paraneoplastic cerebellar degeneration, the acquired immune deficiency syndrome dementia complex, and Parkinson's disease, and the importance of multidimensional descriptions of disease, such as may be obtained from PET data using SSM, is emphasized.

Discussion of PET workshop reports, including recommendations of PET Data Analysis Working Group

On May 1-2, 1989, a PET Data Analysis Working Group convened to consider positron emission tomography (PET) methodology and data analysis. The papers presented and the recommendations of the Group are reviewed. The Group recommended that a standard phantom of the human brain be used by different institutions to examine machine and data reconstruction PET variables. Interinstitutional comparisons could be aided by using a standard three-dimensional coordinate system. Deformations within individual diseased or atypical brains would require nonlinear as well as linear transformations to the standard space, using magnetic resonance images in register with the PET images. Methods for intersubject averaging of pixel-by-pixel or region-of-interest data, as well as appropriate statistical methods, need to be developed. PET data may first be exploratory and hypothesis-generating (with less stringent statistical theory), then later used to test hypotheses (with more stringent statistical criteria). Common databases, obtained by computer simulation models with known inherent structure, or directly by PET measurements on different groups, could be used to compare analytical and statistical methods among institutions.