N.F. Ramsey

Functional Magnetic Resonance Imaging Brain Mapping in Psychiatry: Methodological Issues Illustrated in a Study of Working Memory in Schizophrenia

Functional magnetic resonance imaging (fMRI) is a potential paradigm shift in psychiatric neuroimaging. The technique provides individual, rather than group-averaged, functional neuroimaging data, but subtle methodological confounds represent unique challenges for psychiatric research. As an exemplar of the unique potential and problems of fMRI, we present a study of 10 inpatients with schizophrenia and 10 controls performing a novel “n back” working memory (WM) task. We emphasize two key design steps: (1) the use of an internal activation standard (i.e., a physiological control region) to address activation validity, and (2) the assessment of signal stability to control for “activation” artifacts arising from unequal signal variance across groups. In the initial analysis, all but one of the patients failed to activate dorsolateral prefrontal cortex (DLPFC) during the working memory task. However, some patients (and one control) also tended to show sparse control region activation in spite of normal motor performance, a result that raises doubts about the validity of the initial analysis and concerns about unequal subject motion. Subjects were then matched for signal variance (voxel stability), producing a subset of six patients and six controls. In this comparison, the internal activation standard (i.e., motor activation) was similar in both groups, and five of six patients, including two whom were neuroleptic-naive, failed to activate DLPFC. In addition, a tendency for overactivation of parietal cortex was seen. These results illustrate some of the promise and pitfalls of fMRI. Although fMRI generates individual brain maps, a specialized survey of the data is necessary to avoid spurious or unreliable findings, related to artifacts such as motion, which are likely to be frequent in psychiatric patients.

Development of a functional magnetic resonance imaging protocol for intraoperative localization of critical temporoparietal language areas.

The aim of this study was to evaluate the use of functional magnetic resonance imaging as an alternative to intraoperative electrocortical stimulation mapping for the localization of critical language areas in the temporoparietal region. We investigated several requirements that functional magnetic resonance imaging must fulfill for clinical implementation: high predictive power for the presence as well as the absence of critical language function in regions of the brain, user‐independent statistical methodology, and high spatial accuracy. Thirteen patients with temporal lobe epilepsy performed four different functional magnetic resonance imaging language tasks (ie, verb generation, picture naming, verbal fluency, and sentence comprehension) before epilepsy surgery that included intraoperative electrocortical stimulation mapping. To assess the optimal statistical threshold for functional magnetic resonance imaging, images were analyzed with three different statistical thresholds. Functional magnetic resonance imaging information was read into a surgical guidance system for identification of cortical areas of interest. Intraoperative electrocortical stimulation mapping was recorded by video camera, and stimulation sites were digitized. Next, a computer algorithm indicated whether significant functional magnetic resonance imaging activation was present or absent within the immediate vicinity (<6.4mm) of intraoperative electrocortical stimulation mapping sites. In 2 patients, intraoperative electrocortical stimulation mapping failed during surgery. Intraoperative electrocortical stimulation mapping detected critical language areas in 8 of the remaining 11 patients. Correspondence between functional magnetic resonance imaging and intraoperative electrocortical stimulation mapping depended heavily on statistical threshold and varied between patients and tasks. In 7 of 8 patients, sensitivity of functional magnetic resonance imaging was 100% with a combination of 3 functional magnetic resonance imaging tasks (ie, functional magnetic resonance imaging correctly detected all critical language areas with high spatial accuracy). In 1 patient, sensitivity was 38%; in this patient, functional magnetic resonance imaging was included in a larger area found with intraoperative electrocortical stimulation mapping. Overall, specificity was 61%. Functional magnetic resonance imaging reliably predicted the absence of critical language areas within the region exposed during surgery, indicating that such areas can be safely resected without the need for intraoperative electrocortical stimulation mapping. The presence of functional magnetic resonance imaging activity at noncritical language sites limited the predictive value of functional magnetic resonance imaging for the presence of critical language areas to 51%. Although this precludes current replacement of intraoperative electrocortical stimulation mapping, functional magnetic resonance imaging can at present be used to speed up intraoperative electrocortical stimulation mapping procedures and to guide the extent of the craniotomy.

Language lateralization in schizophrenia, an fMRI study

Anatomical studies have shown that cerebral asymmetry is reduced in schizophrenia. Functional asymmetry appears to be reduced also, as was shown with dichotic listening studies. These studies, however, have not revealed whether reduced lateralization is the result of decreased language activity of the left hemisphere or whether it is the consequence of increased language-related activity in the right hemisphere. To elucidate this, we examined hemispheric dominance for language processing by means of functional MRI. Twelve schizophrenic patients and twelve healthy controls were scanned while they were engaged in a verb-generation and a semantic decision task. Activation was measured bilaterally in the frontal, temporal and temporo-parietal language areas, and a laterality index was derived from activity in these regions of interest in the left and the right hemispheres. Clinical symptoms were rated at the time of scanning. The results indicate that language processing is less lateralized in patients than in controls (a mean laterality index of 0.35 versus 0.63, respectively, difference p<0.01). Analysis of variance of the extent of activity, i.e. numbers of active voxels, revealed a significant hemisphere by group interaction (F(1,22)=11.2, p<0.001), which was due to increased activation in the right hemisphere of the patients (post hoc t-test p<0.05). We found no evidence of reduced activity in the left hemisphere. Further analysis of clinical symptoms rated prior to scanning revealed that decreased language lateralization was associated with more severe hallucinations (r=-0.54, p<0.05). We postulate that decreased language lateralization in schizophrenia may result from failure to inhibit the right hemisphere.

fMRI-Determined Language Lateralization in Patients with Unilateral or Mixed Language Dominance According to the Wada Test

Due to the reported variability of the language laterality index (LI) across fMRI studies, reliable distinction between patients with unilateral and mixed language dominance is currently not possible, preventing clinical implementation of fMRI as a replacement for the invasive Wada test. Variability of the LI may be related to differences in experimental and control tasks, and statistical methodology. The goal of this study was to improve detection power of fMRI for hemispheric language dominance by using a combined analysis of four different language tasks (CTA), that has previously shown more reliable and robust Lls in groups of normal volunteers than individual task analyses (see Ramsey et al). The CTA targets brain areas that are common to different language tasks, thereby focusing on areas that are critical for language processing. Further advantage of the CTA is that it is relatively independent of specific task and control conditions. 18 patients with typical (i.e., left-sided, n = 11) and atypical (i.e., right-sided or mixed, respectively, n = 3 and n = 4) language dominance according to the Wada test underwent fMRI (groups respectively denoted as WadaL, WadaR, and WadaM patients). Statistical methodology (including thresholding of activity maps) was fixed to assure a user-independent approach. CTA yielded better results than any of the individual task analyses: it was more robust (on average 2.5 times more brain activity was detected due to its higher statistical power) and more reliable (concordance for WadaL, WadaM and WadaR patients was respectively 10/11 (91%), 3/4 (75%), and 2/3 patients (67%)). Overall, a significant correlation was observed between frontal and temporoparietal LIs. Remarkably, brain activity for WadaM patients was significantly lower than for WadaL or WadaR patients, and a dissociation in lateralization was observed between frontal (right-sided) and temporoparietal (left-sided) activity in three of four patients. Of the individual task analyses, the verb generation task yielded best results for patients with unilateral language dominance (same concordance as CTA). However, in contrast to CTA results, the verb generation task was unable to identify WadaM patients (concordance in one of four patients). In conclusion, the CTA is a promising approach for clinical implementation of fMRI for the prediction of hemispheric language dominance.

Reproducibility of fMRI-determined language lateralization in individual subjects.

This study investigated within-subject test-retest reproducibility (i.e., reliability) of language lateralization obtained with fMRI. Nine healthy subjects performed the same set of three different language tasks during two fMRI sessions on separate days (verb generation, antonym generation, and picture naming). A fourth task analysis was added in which the three tasks were analyzed conjointly (combined task analysis, CTA). The CTA targets brain areas that are commonly used in different language tasks, aiming more selectively at language-critical structures. The number of active voxels (i.e., robustness) and calculated lateralization index (LI) were compared across sessions, tasks, subjects, and two a priori defined volumes of interest (classical language regions versus whole hemisphere) for a wide range of statistical thresholds. Robustness and reliability strongly varied between task analyses. The CTA was a robust detector of language-related brain activity, in contrast to the single task approaches. The CTA and verb generation task allowed for reliable calculation of the LI. Higher thresholds yielded a clear increase in left lateralization, which was largest when calculated from active voxels in classical language regions.

Working memory capacity in schizophrenia: a parametric fMRI study

Impaired working memory (WM) function in schizophrenia has been associated with abnormal activation of the dorsolateral prefrontal cortex (DLPFC). It is, however, not clear whether abnormal activation is a sign of DLPFC pathology, or a correlate of poor performance. We address this question by examining activity in the WM brain system at different levels of task difficulty. A parametric fMRI paradigm is used to examine how the WM system responds to increasing load. A parametric fMRI design with four levels of a spatial N-back task was used to examine the relationships between working memory load, functional output (performance) and brain activity in 10 schizophrenic patients on atypical antipsychotic medication and to compare these to 10 healthy controls. In spite of increasingly poor performance in schizophrenic patients, activity increased normally in DLPFC and inferior parietal cortex bilaterally and in anterior cingulate, with increasing load. At 3-back, activity dropped in DLPFC in comparison with controls, but not in the other regions. The results indicate that peak activation of the WM-system is reached at a lower processing load in schizophrenic patients than in healthy controls. As a decline of DLPFC activity at high processing loads in itself is not abnormal, WM dysfunction in schizophrenia appears to be the result of an impaired functional output of the whole WM system, causing elevation of the effective burden imposed by WM tasks.

Combined analysis of language tasks in fMRI improves assessment of hemispheric dominance for language functions in individual subjects.

Recent advances in functional neuroimaging techniques have prompted an increase in the number of studies investigating lateralization of language functions. One of the problems in relating findings of various studies to one another is the diversity of reported results. This may be due to differences in the tasks that are used to stimulate language processing regions and in the control tasks, as well as differences in the way imaging data are analyzed,in particular the threshold for significance of signal change. We present a simple method to assess language lateralization that allows for some variation of tasks and statistical thresholding, but at the same time yields reliable and reproducible results. Images acquired during a set of word-comprehension and -production tasks are analyzed conjointly. As opposed to the use of any one particular task, this combined task analysis (CTA) approach is geared toward identifying language regions that are involved in generic language functions rather than regions that are involved in functions that are specific to a single task. In two experiments CTA is compared to single-task analysis in healthy right-handed males. In a third experiment left-handed males were examined. Results indicate that CTA: (1) improves detection of language-related brain activity in individual subjects and (2) yields a high language laterality index (LI) in right-handed males with a small variance across subjects. The high LI matches the strong left-hemisphere dominance for language that is typical for these subjects as reported in neuropsychological and clinical tests in other studies. In the left-handed subjects dominance was found either in the left (n = 4) or the right (n = 1) hemisphere or was absent (n = 3). The LI derived from CTA is more consistent across statistical thresholds for significance of signal change in fMRI analysis than in individual-task analysis. Also, the CTA results are very similar to those obtained with conjunction analysis of the same data.

Language lateralization in female patients with schizophrenia: an fMRI study

Gender differences in schizophrenia are among the most consistently reported findings in schizophrenia research. However, the biological substrate underlying these gender differences is still largely unknown. Differences in language lateralization between men and women may underlie some gender differences in schizophrenia. In previous functional imaging studies, language lateralization was found to be decreased in male schizophrenia patients as compared to healthy males, which was due to enhanced language activation of the right hemisphere as compared to the healthy males. It could be hypothesized that decreased language lateralization in schizophrenia is gender specific, i.e. decreased lateralization in male patients and normal lateralization in female patients. To test this hypothesis, language activation was measured in 12 right-handed female patients with schizophrenia and 12 healthy females, and compared to findings in 12 male patients and 12 male controls of an earlier study. Language lateralization was significantly lower in the female patients (0.44) as compared to the female controls (0.75), which was due to increased activation of the right-sided language areas (patients: 19 voxels; controls: 8 voxels), while left hemisphere activation was similar in patients and controls. When these data are compared to the male patients and controls, both patient groups had lower lateralization than their healthy counterparts, but there was no difference between male and female patients. In both sexes, decreased lateralization resulted from increased right hemispheric language activation, which suggests a failure to inhibit nondominant language areas in schizophrenia. These findings indicate that lower language lateralization in women is not likely to underlie gender differences in schizophrenia.

Phase Navigator Correction in 3D fMRI Improves Detection of Brain Activation: Quantitative Assessment with a Graded Motor Activation Procedure

Motion poses severe problems for BOLD fMRI, particularly in clinical studies, as patients exhibit more involuntary movements than controls. This study focuses on the merits of a motion correction technique incorporated in multishot fMRI scans, so-called phase navigator correction. The technique entails real-time assessment and off-line elimination of signal fluctuations caused by subject motion. The purpose of this study was to quantify and characterize the effect of this type of improvement on 3D fMRI brain activity maps. For imaging, the 3D PRESTO method was used, with a relatively simple finger opposition task. The followed strategy was guided by the notion that application of any fMRI imaging tool in clinical studies requires several qualities, such as high and spatially homogeneous sensitivity to brain activity, and low sensitivity to motion. A graded motor activation protocol in 10 healthy subjects revealed that image stability was improved by approximately 20%, by the use of phase navigator correction. As a result, sensitivity for task-related BOLD signal change was enhanced considerably in the brain activity maps. Implications for use of this fMRI technique in patient studies are discussed.

Language area localization with three-dimensional functional magnetic resonance imaging matches intrasulcal electrostimulation in Broca's area

In this study, intraoperative electrocortical stimulation mapping (ioESM), the current gold standard for the localization of critical language areas, is compared with functional magnetic resonance imaging (fMRI) in a 14‐year‐old girl with medically intractable epilepsy caused by a tumor in the region of Brocas area. Prior to the operation, four different fMRI tasks that target inferior frontal language areas were applied. Prior to the resection, ioESM as well as fMRI detected no language areas at the exposed cortical area. After removal of the tumor, a unique opportunity presented itself, where ioESM could be performed in the depth of a now exposed and intact gyrus. One specific locus that was indicated to be a critical language area by multiple‐task fMRI was targeted. IoESM selectively confirmed the location of this language area to within an estimated 3 mm. We propose that the combined use of different fMRI tasks increases the sensitivity and specificity for the detection of essential language areas.