Sidebar: Scanning protocols at two sites of Chinese language processing study

Clinicians have used the Wada test for the past four decades to assess the hemispheric dominance of language function.¹ The Wada test is highly invasive and distressing to patients, requiring intracarotid administration of sodium amytal.^1,2 Progress in functional brain imaging techniques,³ however, has generated interest in the use of conventional MRI systems to precisely localize human brain function, including language processing. Functional MRI (fMRI) has consequently been proposed as a noninvasive alternative to the Wada test.^4,5

We have investigated the suitability of fMRI to localize cortical areas associated with Chinese language processing. The same experiment was conducted using two different clinical MRI systems in separate institutions to examine the reproducibility and enhance the reliability of the study. Participants were asked to perform synonym and homophone judgment tasks that have been studied extensively and are easy to perform. They require operation of two basic components of word reading: semantic and phonological processes.

Four men and four women, all of whom were healthy and between 23 and 36 years of age, consented to take part in the study. All participants were right-handed native Chinese speakers recruited from either the Graduate School of the Chinese Academy of Sciences in Beijing or the University of Minnesota in Minneapolis in the U.S. One of the subjects participated in the same experiment at both locations. We performed MRI on a 1.5-tesla GE Signa Horizon in Beijing, and on a 1.5-tesla Siemens Vision in Minneapolis (see table). All subjects were asked to identify word pairs with similar meanings (synonym judgment) and similar pronunciation (homophone judgment). Passively viewing a crosshair in the center of the screen served as baseline control task.

A series of fMR images were collected while participants completed eight to 12 blocks of linguistic tasks (10 word pairs presented per task) alternating with the baseline control task. Data analysis was then performed using Analysis of Functional Neuroimages software.⁶ The functional images were spatially registered to check and correct for motion artifacts. The images were coregistered to the three-dimensional structural images and normalized according to standard coordinates.⁷

Time series data were analyzed by correlating the time course for each voxel with the ideal trapezoidal reference waveform corresponding to the timing of linguistic tasks. Voxels meeting or exceeding a correlation coefficient of 0.40 (p<0.00001) were considered to be reliably associated with the tasks. The activated voxels were then superimposed on the anatomic images to produce activation maps. Finally, a student t-test was used to further examine the possible differences in brain activation while homophone and synonym judgment tasks were performed.

Results from all eight participants showed that the tasks activated areas of the cerebral cortex, mainly on the left hemisphere, including Broca’s area and Wernicke’s area. Activation was also observed in the bilateral supplementary motor area (SMA), extrastriate visual areas, and ventral temporal cortex, all of which are related to language. Activation patterns showed little difference between data collected at each of the two centers.

Data from the subject who took part in the investigation at both sites revealed activation of an extensive network during linguistic processing, including Broca’s area, Wernicke’s area, the SMA, bilateral ventral temporal cortex, and the extrastriate cortex (Figure 1). Images produced on the Siemens scanner, however, showed more activated voxels in Broca’s and Wernicke’s areas than did images from the GE machine. The data from the Siemens scanner also highlighted two additional activation areas in the inferior frontal cortex and superior temporal gyrus on the right hemisphere, mirroring the traditional Broca’s and Wernicke’s areas.

Comparison of the two sets of data for all participants showed no substantial difference between the time course of signal intensity changes in four primary activated areas (Broca’s area, Wernicke’s area, left extrastriate visual areas, and left ventral temporal cortex). We then generated activation maps, one corresponding to the homophone judgment task and another to the synonym judgment task, to investigate the possible difference between brain activation during the two linguistic tests. No essential difference was found. The student t-test also failed to reveal significant differences in brain activation when the two tasks were compared, while the time taken to respond to each task proved to be virtually identical (Figure 2).

Wider Implications

The high spatial resolution and noninvasive nature of fMRI have contributed to its wide use in neuroscience research. The potential of fMRI is now being realized in clinical practice as well. The laterality of language function assessed by fMRI activation, for example, is believed to be consistent with that determined by Wada testing.^4,8 And as we have shown, fMRI can also detect brain activation during Chinese word processing. In addition to determining the laterality of language function, fMRI allows better precision and may become a widely accepted clinical tool for localizing functional brain areas in patients.

Traditional neuropsychological studies in brain-damaged patients and more recent neural image studies in normal volunteers have demonstrated left hemispheric dominance in subjects processing alphabetic languages.^9,10 Several authors have proposed that the reverse is true during Chinese word processing and that the right hemisphere is dominant, given the prominent pictorial nature of Chinese characters.^11,12 Yet our own study indicates a strong left hemisphere dominance during Chinese word processing, in accordance with studies of word processing in Western languages. This finding remains consistent with previous investigations that showed strong left hemisphere activation in subjects listening to and reading Chinese words or sentences.^13-15

The similarity in brain activation between subjects performing semantic tasks and those engaged in phonological processing was surprising. It conflicts with the traditional view that semantic and phonological processes are two basic operations of word reading.¹⁶ It is possible that both semantic and phonological systems are automatically and spontaneously engaged during word reading, whether or not specific instruction is provided. This argument is supported by work showing that the mere presence of words prompts widespread brain activation, with or without explicit linguistic demands.¹⁷

The almost identical brain activation shown in data acquired on the two different scanners highlights the reproducibility and reliability of our research.¹⁸ Differences in brain activation exhibited by the subject who participated in experiments at both trial locations may be explained as a learning effect. Activation in Broca’s and Wernicke’s areas has been shown to decrease dramatically when subjects read English words that they are familiar with.¹⁹ Our subject read the same Chinese words during each of the two experiments, suggesting that the same could be true in the case of Chinese word processing.

Acknowledgment

This article was adapted from a poster presented at the joint meeting of the International Society for Magnetic Resonance in Medicine and the European Society of Magnetic Resonance in Medicine and Biology, held in Glasgow, Scotland, in April 2001.

References

1. Wada J. Intra-carotid injection of sodium amytal for the lateralization of cerebral speech dominance. J Neurosurg 1960;17:266-282.
2. Hollway V, Liegeois F, Baldeweg T, et al. Language lateralization using fMRI in children undergoing Wada testing. Neuroimage 2000;11:S290.
3. Weng XC, Ding YS, Volkow ND. Imaging the functioning human brain. Proc Natl Acad Sci USA 1999;96:11073-11074.
4. Desmond JE, Sum JM, Wagner AD, et al. Functional MRI measurement of language lateralization in Wada-tested patients. Brain 1995;118:1411-1419.
5. Brockway JP. FMRI may replace the Wada test for language lateralization. Neuroimage 2000;11:S277.
6. Cox RW. AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 1996;29:162-173.
7. Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain. New York: Thieme, 1988.
8. Binder JR, Swanson SJ, Hammeke TA, et al. Determination of language dominance with functional MRI: a comparison with the Wada test. Neurology 1996;46:978-984.
9. Petersen SE, Fiez JA. The processing of single words studied with positron emission tomography. Annu Rev Neurosci 1993; 16: 509-530.
10. Fiez JA, Petersen SE. Neuroimaging studies of word reading. Proc Natl Acad Sci USA 1998;95:914-921.
11. Hatta T. Recognition of Japanese kanji in the left and right visual fields. Neuropsychologia 1977;15:685-688.
12. Tzeng O, Hung D, Cotton B, et al, Visual lateralization effects in reading Chinese characters. Nature 1979;282:499-501.
13. Li E, Weng XC, Han Y, et al. Asymmetry of brain functional activation: fMRI study under language and music stimulation. Chin Med J 2000;113:154-158.
14. Tan LH, Spinks JA, Gao JH, et al. Brain activation in the processing of Chinese characters and words: functional MRI study. Hum Brain Mapp 2000;10:16-27.
15. Chee M, Tan E, Thiel T. Mandarin and English single word processing studies with functional magnetic resonance imaging. J Neurosci 1999;19:3050-3056.
16. Petersen SE, Fox PT, Posner MI, et al. Position emission tomographic studies of the cortical anatomy of single-word processing. Nature 1988;331:585-589.
17. Price CJ, Wise RJ, Frackowiak SJ. Demonstrating the implicit processing of visually presented words and pseudowords. Cereb Cortex 1996;6:62-70.
18. Tegeler C, Strother SC, Anderson JR, et al. Hum Brain Mapp 1999;7:267-283.
19. Raichle ME. Practice-related changes in human brain functional anatomy during no-motor learning. Cereb Cortex 1994;4:8-26.

Sidebar

Scanning protocols at two sites of Chinese language processing study

Beijing (GE Signa Horizon)

• Conventional axial T1-weighted images acquired with spin-echo pulse sequence (TR 500/TE 14, FOV 24 × 24 cm², slice thickness 6 mm, gap 1 mm, matrix 256 × 192, number of excitations = 2)
• 124 functional images per slice acquired, using single-shot gradient-echo echo-planar imaging (EPI) (TR 2000/TE 40/ flip angle 90°, slice thickness 6 mm, gap 1 mm, matrix 64 × 64)
• 64 continuous slices covering entire brain, acquired with fast spoiled GRASS pulse sequence (TE 2.2/flip angle 30°, slice thickness 2.5 mm, gap 0, matrix 256 × 256) for subsequent 3-D reconstruction and spatial normalization

Minneapolis (Siemens Vision)

• 2-D anatomical images acquired using FLASH pulse sequence (FOV 24 × 24 cm², slice thickness 7 mm, gap 0, matrix 256 × 256)
• 167 images per slice acquired, using a T2*-weighted EPI sequence (TR 2000/TE 55)
• 3-D FLASH sequence used to obtain 3-D anatomic data set (resolution 0.9 × 0.9 × 1.5 mm³)