PET imaging uses a method similar to that used for SPECT but with different radioactive tracers and more sophisticated detection equipment, which has improved with new technologies (Figures 6–8 and 6–9). As with SPECT, the physics behind PET limit its resolution, which is approximately 4 mm on high-quality scanners.
Thus, PET images are much clearer and show greater anatomical detail than SPECT images. As in the proce-dure used with SPECT, a radiotracer is injected into the patient intravenously. As it decays, a positron is released.
After collision with an electron, two photons are pro-duced that travel away from each other in a straight line at the speed of light. The photons are detected on oppo-site sides of the PET scanner simultaneously, and a com-puterized calculation is performed to pinpoint where in the brain the original positron was located. A record of these detections is made and can be transformed by a computer into a tomographic image (Figure 6–10).
Because two photons must be detected at the same instant to be “counted,” the technique reduces errors in detec-tion. As with MRI and SPECT, coronal, sagittal, and axial views are available. The images can be visually inter-preted but more commonly are analyzed statistically using various software programs.
Tracers
Like SPECT, PET requires the injection of a radioactive tracer but, because of differences in the tracers used, can image either CBF or metabolism. Fluoride 18 (18F) fluo-rodeoxyglucose (FDG) is the most commonly used tracer for clinical PET scans. It is taken into cells via the glucose transport mechanism, after which it is phosphorylated into FDG-6-phosphate. Because it is not a substrate for the glycolytic process, the FDG-6-phosphate remains trapped in the cell. Thus, scans with FDG produce a mea-sure of glucose metabolism rather than blood flow. A scan performed with FDG generally takes approximately 30–
40 minutes. The oxygen 15 (15O) tracer is more com-monly used in research; Table 6–3 provides an overview of FDA-approved PET radiotracers. Because of the short half-life, 15O scans are performed within a few minutes of when the patient is correctly positioned and in the scan-ner. The resolution obtained with 15O tracer is inferior to that obtained with FDG. However, the use of short-acting
isotopes permits repeat studies in the same subject in a short period. This circumstance is useful if a cognitive activation paradigm, such as performance of a verbal memory task, is to be compared with scans done in other states, such as motor activations (e.g., finger tapping) (Figure 6–11).
Practical Considerations
The method used for PET is similar to that used for SPECT, but the scan must take place within a few min-utes or seconds of the injection because of the differing properties of the isotopes used in PET. As with SPECT, for most clinical purposes a resting whole-brain scan is ordered. Depending on the tracer used, the time of the scan is 2–40 minutes, during which time the patient must remain still in the scanner. FDG, the most com-monly used PET tracer in clinical studies, requires a 30-to 40-minute scan. As with SPECT, sedation may be given after isotope injection if the patient is extremely anxious or unable to remain still while lying supine ing the scan. In many centers, headholders are used dur-ing PET scanndur-ing to keep the patient’s head in a stable position. Headholders can be constructed of thermo-plastic and individually fitted to the patient’s head. Alter-natively, they may be made of foam rubber or other soft material placed around the head to prevent motion. The degree of stabilization gained must be weighed against the amount of discomfort caused to the patient, espe-cially if he or she is claustrophobic or uncomfortable being somewhat restrained.
Indications
There are no clinical guidelines for use of PET in TBI at this time. As with SPECT, PET scans are often obtained when brain injury is suspected but not seen on structural studies or when structural studies do not indicate damage extensive enough to explain a patient’s deficits.
Limitations
PET scans generally cost $2,000 for a clinical study. In comparison, SPECT scans are $800–$1,000 at most cen-ters. The higher price of PET is due to several factors, including the advanced technology used in PET scanners compared with that used in SPECT scanners. For certain short-half-life isotopes, such as 15O, the isotope must be made onsite, limiting its use to centers that have a cyclo-tron (another expense). Thus, PET is not available at many institutions.
Overview of Abnormal Findings in Other Psychiatric Disorders
As with SPECT, PET is used in the evaluation of many neurological disorders. The most common clinical uses are in the assessment of patients with epilepsy, central nervous system malignancies, and cerebrovascular acci-dents. However, in acute cerebrovascular accident, SPECT results have been shown to reflect abnormalities not seen with FDG-PET (Henkin 1996). PET is also use-F I G U R E 6 – 8 . Procedure for obtaining a PET scan.
The patient receives an intravenous injection of the radioactive tracer while lying in a darkened room. After 20–30 minutes are spent in the darkened room to allow the tracer to distribute through the brain, the patient is ready to be scanned (A). Scanning usually begins within 1 hour of tracer injection and requires 30–45 minutes to complete. A headholder is often used to prevent head motion (B).
Source. Pictures courtesy of CTI Molecular Imaging, Inc.
ful, in some cases, in helping differentiate between differ-ent types of demdiffer-entia. The ability of PET to detect per-fusion changes consistent with AD may be superior to that of SPECT, with studies reporting sensitivity of 87%–
94% and specificity of 85%–96% (Hoffman et al. 1996;
Mielke and Heiss 1998; Van Heertum et al. 2000).
PET has also been used for research studies of head-ache. Flow reduction has been seen in migraine headache with and without auras (Bednarczyk et al. 1998; see Au-rora and Welch 2000 for a review), although hyperperfu-sion of cortical regions and brainstem have also been re-ported in studies of migraine without aura (see Cutrer et al. 2000 for a review). Studies with PET suggest that clus-ter headaches may be associated with activation of the hy-pothalamus (May et al. 1999).
Research has been conducted in evaluation of pain with PET. According to studies primarily with nonpatient volunteers, the brain regions most consistently found to be associated with varying types of pain perception in-clude the contralateral insula and anterior cingulate, bi-lateral thalamus and premotor cortex, and the vermis of the cerebellum, with magnitude of neuronal response in-creasing as level of pain is modulated upward (see Casey 1999 for a review). Hypothalamic and periaqueductal gray activation associated with pain perception has also been reported in other PET work with nonpatient volun-teers (Hofbauer et al. 2001; Hsieh et al. 1996).
The use of PET in the evaluation of other psychi-atric conditions has yet to be demonstrated. PET stud-ies of patients with depression have shown prefrontal cortex flow and metabolic changes, which may resolve with treatment (Goodwin 1996). Some PET studies of patients with obsessive-compulsive disorder have shown increased metabolism in the caudate and/or or-bitofrontal cortex (Baxter et al. 1987, 1988), although not all study results are consistent with these (Swedo et al. 1989). In schizophrenia, imaging studies suggest frontal metabolic and flow deficits (Andreasen et al.
1996; Liddle et al. 1992) and also have begun to dem-onstrate differences between patients with positive symptoms and those with more predominant negative symptoms (Lahti et al. 2001). Receptor ligand studies, similar to those described with SPECT, have also been conducted with PET for the study of psychiatric ill-nesses. In particular, work characterizing dopamine re-ceptor change has been extremely important, especially in the study of schizophrenia (see Verhoeff 2001 for a review).
Limited PET investigations have been conducted in patients with psychogenic disorders. Hypometabolism in the caudate, putamen, and right precentral gyrus was found in one study of somatization disorder and somato-form disorder (Hakala et al. 2002). Reduced frontal acti-vation was seen in three patients with limb weakness (Spence et al. 2000). In a single case study with PET, ac-tivations were produced during hypnotic paralysis similar to those observed with psychogenic paralysis (Halligen et al. 2000).
Overview of Abnormal PET Findings in TBI
PET has been used in several studies of TBI patients to assess many measures, including evidence of functional abnormalities in patients who have normal structural scans, prognosis, correlations between post-TBI behav-ioral disorders and brain injury, and correlations between neuroanatomical damage and neuropsychological test–
F I G U R E 6 – 9 . PET imaging then and now.
Axial PET images of brain acquired in 1983 and 2002. Note the significant improvement in resolution since the 1980s.
Source. Pictures courtesy of CTI Molecular Imaging, Inc.
performance deficits. Most of these studies involve small numbers of patients, making conclusions based on the data problematic. Use of PET for cognitive activation studies to look at neuroplasticity after TBI and for exam-ination of neuropathological changes in these patients are two promising applications for PET. In general, the scope of the clinical studies with PET is smaller than those with SPECT, but research applications of PET may ultimately prove to be more fruitful.
Studies Using PET and Structural Imaging In contrast to research with SPECT, little work has been done to assess whether PET is more accurate than struc-tural imaging in assessment of lesions in TBI patients.
Because PET can provide other data in addition to blood flow information, one might expect differing use of PET in prediction of outcome.
The limited work thus far suggests that, like SPECT, PET may be helpful in assessment of patients with TBI who have normal structural imaging but behavioral prob-lems or cognitive deficits. Studies using FDG (glucose metabolism) or cobalt 55 (cell death) have indicated that PET provides additional information beyond that avail-able from structural imaging (Fontaine et al. 1999; Jansen et al. 1996; Langfitt et al. 1986; Rao et al. 1984; Ruff et al.
1994; Umile et al. 2002). In all of these studies, more le-sions were present on PET. In some of these studies, the authors suggest that these abnormalities correlated with behavioral and cognitive complaints. However, as with SPECT, a causal link between a specific lesion seen on functional imaging and behavioral changes seen in a pa-tient is difficult to assess.
Other work has questioned whether the more exten-sive information obtained from PET is actually clinically F I G U R E 6 – 1 0 . Serial axial fluoride 18 fluorodeoxyglucose PET images of a normal adult brain.
Letter/number combinations below each image refer to brain slice order.
Source. Picture courtesy of CTI Molecular Imaging, Inc.
useful in the management of TBI patients. Worley et al.
(1995) examined PET results compared with CT or MRI data in 22 children and adolescents with severe TBI who were followed through a rehabilitation program. They concluded that PET was not more helpful than standard structural imaging in prediction of outcome after TBI in
children. In a more recent study, Bergsneider et al. (2001) found that FDG-PET was not useful in following func-tional recovery from moderate and severe TBI, because the correlation between change in metabolism on follow-up PET and recovery from neurological damage was weak. Their PET findings did suggest that metabolic re-T A B L E 6 – 3 . U.S. Food and Drug Administration–approved, commonly used tracers/ligands for PET Tracer/radioligand Parameter measured Comments
18F Glucose metabolism Commonly used in clinical studies; longer half-life than 15O means only one scan may be acquired in each scanning session.
15O Blood flow Short half-life means that multiple scans may be collected in one session with a subject; commonly used for cognitive research studies with cognitive activation paradigms.
13N Blood flow Used in cardiac assessment.
55Co Calcium Provides indications of areas where cell death is occurring.
11C Dopaminergic system Research use to study receptors.
F I G U R E 6 – 1 1 . Current PET imaging capabilities.
Three-dimensional (3D) reconstruction of PET results (A). 3D imaging improves appreciation of the extent of functional abnormal-ities. Neurotransmitter systems may also be imaged with PET. Presynaptic dopamine terminals can be labeled with [18F]fluorodopa (B). Dopamine D2 receptors can be labeled with [11C]N-methylspiperone (C).
Source. Pictures courtesy of CTI Molecular Imaging, Inc.
covery begins approximately 1 month after moderate or severe TBI, a concept that may have implications for the timing of pharmacological or rehabilitational interven-tions after TBI.
Some PET findings may indicate new directions for interventions post-TBI. Bergsneider et al. (1997) sug-gested that the apparent hyperglycolysis may be secon-dary to excitotoxicity or ischemia. Yamaki et al. (1996) studied CBF, oxygen ejection fraction, and cerebral me-tabolism for both oxygen and glucose in three patients with acute, severe, diffuse TBI. Their findings suggested that persistent anaerobic glycolysis (which may indicate excitotoxicity) is a predictor of poor outcome. PET find-ings suggest that hyperglycolysis (an influx of ions into cells that have not suffered irreversible damage) occurs af-ter TBI (Bergsneider et al. 1997), possibly because more energy (i.e., glucose) is needed to pump out the ions and restore homeostasis (Hovda 1996). Coles and others (2004) also discuss the use of oxygen extraction fraction studies in TBI for evaluation of ischemic burden and newer methods for its determination. Other uses for PET in investigation of pathophysiology following TBI have also been proposed in recent work (Hattori et al. 2003;
Hattori et al. 2004; Wu et al. 2004). These findings may encourage new interventions/treatment for severe acute TBI, such as diminution of persistent excitotoxicity.
PET consistently shows abnormalities not seen on structural imaging, especially in cases of mild TBI. How-ever, the actual clinical usefulness of this information has not been proven. PET has not been found to be useful in assessment of recovery but has suggested new avenues for research into early interventions (Bergsneider et al. 1997;
Yamaki et al. 1996).
Studies Using Behavioral Measures
Few studies have focused on the use of functional imaging to assess patients with behavioral symptoms after TBI.
Given the changes seen on PET scans of patients with primary psychiatric illness, one might expect some corre-lation between PET data and post-TBI behavioral prob-lems. Starkstein et al. (1990) used FDG-PET to evaluate patients with mania after TBI. Three patients who had only subcortical damage on structural imaging were scanned during mania; they showed right lateral basitem-poral hypometabolism, implicating right-sided damage in the development of mania. Fontaine et al. (1999) also reported a relationship between behavioral disorders in severe TBI and mesial prefrontal and cingulate metabolic abnormalities. Further work with detailed behavioral information and psychiatric diagnosis is needed in this area before use can be assessed, although these
prelimi-nary studies suggest that PET studies in TBI may enhance research into neuroanatomical underpinnings of psychiatric symptoms.
Studies Using Neuropsychological Assessments Several groups have compared PET results in TBI patients with results from their performance on neuropsychological tests, with varying results. Some studies found good corre-spondence between areas of abnormality on PET and neu-ropsychological test deficits (Fontaine et al. 1999; Langfitt et al. 1987; Rao et al. 1984; Ruff et al. 1994). On the other hand, the pattern of deficits on neuropsychological testing has not been shown to predict PET location of lesions (Jansen et al. 1996; Umile et al. 2002).
A single study has compared PET and SPECT di-rectly for assessment of neuropsychological deficits in TBI (Abu-Judeh et al. 1998). SPECT scanning demon-strated frontal and parietal perfusion, concurring with neuropsychological test results, whereas FDG-PET re-sults indicated normal glucose metabolism. The authors suggested that, at least in mild TBI, vascular compromise due to injury may cause SPECT findings of flow loss, al-though the normal glucose metabolism indicates that un-derlying tissue is still viable. Although this example illus-trates the possibility that different information from the two modalities could be complementary, no work has been done to apply this finding clinically to date.
Activation Studies
Performance of a behavioral task during a scan, called a cognitive activation paradigm, may be helpful in studying the function of particular cognitive domains. In the larg-est PET activation study to date, Gross et al. (1996) com-pared FDG-PET results from 20 patients with mild TBI to those of noninjured control subjects. All subjects were scanned while performing a simple continuous perfor-mance task (i.e., press button when “zero” appears on screen). The authors concluded that even mild TBI may produce abnormalities both on neuropsychological test performance and behaviorally and that cerebral metabo-lism may be affected. They also noted that performance of an activation task during scanning may have affected brain activity, because patients with more damage may need to exert more effort to perform the task, which could be reflected in metabolic change. Similar results were seen in a study by Levine and others (2002), which exam-ined brain activation differences in six moderate to severe TBI patients, with greater brain activation in the TBI patients relative to matched noninjured control subjects during performance of a cued recall task. The authors
suggested this may be due to brain reorganization in response to diffuse axonal injury, possibly indicating compensation.
Studies Using Other PET Tracers
As with SPECT, PET ligand studies are becoming an important tool for research. Although these techniques have not yet been used to study TBI, they may well pro-vide the most important future contributions from PET.
Potentially, radioactive ligands (e.g., raclopride, which is used to study dopaminergic transmission) could provide information on disruption of receptor types, intracellular messengers, and proteins after TBI. Ligands are also available, but less widely so, for research use in investiga-tions of serotonergic, acetylcholinergic, and other neu-rotransmitter systems.
Recommendations
As of this writing, PET does not have a large role in eval-uation of TBI. In very select cases in which more exacting localization of lesions is important, PET may be helpful, although correlation of specific lesion location with func-tion is often problematic. Otherwise, the lower cost and greater availability of SPECT make it the best functional assessment in those select cases in which functional imag-ing may enhance evaluation of TBI. As with SPECT, PET may sometimes be useful in detection of lesions in cases in which behavioral symptoms or cognitive deficits are present in the patient with no apparent structural injury. PET is generally superior to SPECT for use in research studies of cognitive function and brain injury because of its finer resolution. Use of 15O as a PET tracer allows investigators to perform several studies on a patient in one session, which is important when studying cognition. PET may have a role in investigation of patho-physiology of TBI. Most important, it may be useful in determining whether pathophysiological events after TBI are dynamic in nature and, if so, when the optimum time for intervention is. In the future, PET scanning may also be a technique for the study of putative mechanisms of cellular damage after TBI, including excitotoxicity and changes in neurotransmitter systems.