Refined analysis of complex language representations by non-invasive neuroimaging techniques

Abstract

Objective. The determination of language lateralisation is important for patients with medically intractable epilepsy or a brain tumour near the language areas to avoid the risk of post-surgical language deficits. The aim of this study was to evaluate the clinical usefulness of near-infrared spectroscopy (NIRS) to identify language lateralisation compared with functional MRI (fMRI) and magnetoencephalography (MEG) in multiple language tasks.Methods. We investigated 28 patients whose language dominance was evaluated by the Wada test. fMRI, MEG and NIRS were performed to investigate language representation. All patients were as'ked to read three-letter words silently for fMRI and MEG (Kana reading) and to write words beginning with a visually presented letter (word generation) for NIRS. The laterality index was calculated to assess language lateralisation in each investigation.Results. In 24 cases (85.7%), of which two investigations showed the same laterality, the results had perfect concordance with the Wada test. In patients with left dominance, the sensitivity and specificity of fMRI, MEG and NIRS was 95.0% and 62.5%, 100% and 87.5%, 75.0% and 87.5%, respectively. In three patients with right lateralization, only NIRS showed a significant increase of oxygenated-haemoglobin in the right inferior frontal region, indicating right dominance.Conclusion. We established a method to determine language lateralisation by co-utilising fMRI, MEG and NIRS with high reliability. NIRS recognised atypical language representation, in addition to fMRI and MEG. While fMRI, MEG and NIRS are not currently as accurate as the Wada test in determining language lateralisation, this non-invasive and repeatable method has great potential as an alternative to the Wada test in time following further research and refinement of these techniques.