3918

Which lateralization index and which language task should we use in lateralization test?
Kayako Matsuo1, Kazufumi Akiyama1, and Yasushi Kaji2

1Department of Biological Psychiatry and Neuroscience, Dokkyo Medical University, Tochigi, Japan, 2Department of Radiology, Dokkyo Medical University, Tochigi, Japan

Synopsis

It is crucial to select a good language task in fMRI to examine the language lateralization. We examined the concordances of 6 lateralization indices (LIs) using 4 language tasks. We confirmed that our previously developed LI, AveLI, showed a better concordance of index values across 4 tasks. Also, the verb generation task showed a better concordance across the 6 LIs than other tasks. In addition, this task showed a plausible ratio of rightward asymmetry (5 out of 16 left/both-hand preferred). Thus, we concluded to conduct the verb generation task and compute AveLI as a robust index of the language lateralization.

INTRODUCTION

Preoperative examinations of neurosurgery often require the testing of language asymmetry/lateralization because about 25% of left-hand preferred people have a rightward asymmetry in language although many others have a leftward asymmetry 1. We developed a new lateralization index (LI) of fMRI, AveLI, to conquer computational problems in a conventional LI method 2. Another problem was which language task we should use. In this study, we conducted 4 language tasks of fMRI and focused on the following 2 topics: (1) which LI method showed a better concordance across task types and (2) which language task showed a better concordance across LI types. For the former, we previously found a better concordance between 2 language tasks in AveLI than in other types of LIs 2. It might mean that AveLI was a task-independent index and provided a robust index regardless of task types. For the latter, a better concordance of a task type across LI types would indicate a robustness of the task regardless of LI methods.

METHODS

Thirty-three people participated in this study (female 14, age 19 to 71, average 36.8 years old). Seventeen (female 9) were right-hand preferred people (handedness score 100 3) whereas 16 (female 5) were left-hand/both-hand preferred (=< 86.7). We used Siemens MAGNETOM Prisma 3T to conduct fMRI with the following parameters: multi-band factor 2, TE/TR 22/1250 ms, FOV 192, 64 x 64, 46 slices, slice thickness 3.6mm (3 plus gap 0.6) and 226 volumes per run. Four runs were conducted to perform the following 4 language tasks: Narration listening, Verb generation, lexical decision for Auditory word sound and lexical decision for Visual word display. SPM12 (University College London) and MATLAB (The MathWorks, Inc.) were used to analyze the imaging data. After a standard SPM processing, the following 6 LIs were computed using a left and a right regions of interest (ROIs) of opercular and triangular parts of the inferior frontal gyrus.

  1. AveLI: compute LIs by using all voxel values inside the ROIs as thresholds. An LI is computed by the formula (Left – Right) / (Left + Right) using intensity values of the above-threshold voxels. AveLI is the average of all these LIs.
  2. AveLI_v: same algorithm as AveLI but use voxel number instead of intensity values.
  3. baseLI: compute an LI using all positive (>0) intensity values with no weighting.
  4. baseLI_v: same as baseLI but use voxel number.
  5. p001unc: use a threshold of p<0.001, uncorrected, and compute (Left – Right) / (Left + Right) using intensity values.
  6. p001unc_v: same as p001unc but use voxel number (conventional method).

We computed Kendall’s coefficient of concordance (W) to examine rankings of individuals’ LIs in a task by a LI type to compare (1) LI types by seeing how the 4 tasks showed a concordance within an LI type and (2) task types by seeing how the 6 LI types showed a concordance within a task type.

RESULTS

We observed a considerable variation of LI values across LI types as well as task types (Fig. 1). Two subjects had no above-threshold voxels for p001unc and p001unc_v in one of lexical decisions. We put 0 (bilateral) for these data. Some other subjects had no above-threshold voxels in one side of ROIs (left or right), indicating 1 or -1, which resulted in having a same rank. The averaged rank was assigned to these data in computing the W. The orders of Ws (df=32) were as follows (p<0.001 for all). (1) AveLI 0.638, AveLI_v 0.635, baseLI 0.607, p001unc 0.598, p001unc_v 0.595, baseLI_v 0.498. (2) Verb 0.916, Narration 0.911, Auditory 0.843, Visual 0.842.

DISCUSSION

As in our previous study using 2 tasks 2, an overall high concordance of AveLI across 4 tasks was confirmed. However, the LI values of individuals considerably varied across 4 tasks even in AveLI. As for the task types, the verb generation task showed the highest concordance among all, which indicated that this task showed robust lateralization values regardless of LI types (Fig. 1). In addition, Figure 1 showed other favorable features in verb generation in that; (1) no changes in left to rightward asymmetry across LI types within an individual and (2) only 5 rightward asymmetry individuals (all from left/both-hand preferred), the ratio of which (31.3%) was similar to that in a previous study 1.

CONCLUSIONS

The findings suggested that the verb generation task was favorable in the examination of the language lateralization and so was the use of AveLI for the computation of the index at least for the anterior language cortex corresponding to Broca’s area.

Acknowledgements

No acknowledgement found.

References

1. Knecht S, Dräger B, Deppe M, et al. Handedness and hemispheric language dominance in healthy humans. Brain. 2000;123(12):2512-2518.

2. Matsuo K, Chen SH, Tseng WY. AveLI: a robust lateralization index in functional magnetic resonance imaging using unbiased threshold-free computation. J Neurosci Methods. 2012;205(1):119-129.

3. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9(1):97-113.

Figures

Variations of index values by LI types as well as by task types. Each line indicates each individual. The same color indicates the same individual across the four tasks. The vertical axis indicates the LI values. The value 1 indicates the most leftward lateralization (i.e., no above-threshold voxels in the right ROI) whereas -1 indicates the most rightward.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3918