SENSAAS, a new atlas of the brain’s language areas

The GIN has just developed a new brain atlas of cortical areas of language integration. It will be particularly useful for cohort studies looking for the genetic bases of language regions or on the neuroanatomical changes associated with language pathology. 

32 multimodal areas of sentence processing activated and leftward asymmetrical during sentence production, reading and listening were identified by comparison with a reference task (word list production, reading and listening). The temporal correlations at rest between these 32 regions made it possible to detect their belonging to 3 networks. Among these networks, one, including 18 regions, contains the essential language areas, i.e. those whose lesion would cause an alteration in the understanding of speech.

In the line of the AAL atlas which allows the automatic anatomical localization of activations (Tzourio-Mazoyer 2002) and the AICHA atlas based on connectivity at rest (Joliot 2014), SENSAAS was developed from functional Magnetic Resonance Imaging (fMRI) obtained during both task-induced and resting-state acquisitions in 144 right-handed subjects from the BIL&GIN database (a database dedicated to the study of hemispheric specialization, Mazoyer 2016).

 SENSAAS is available to the scientific community: here.

Extracting the regions involved in language comprehension

Language is one of the first cognitive functions to be explored in functional neuroimaging, and 30 years of research have revealed the involvement of many cortical areas in various language activities (see the meta-analyses of Price 2012, Vigneau 2006). However, there are no operational atlas of language integration areas available to the community. To provide such an atlas, we have developed a method combining the analysis of activations during several language tasks with that of temporal correlations measured during the so-called “resting” state in the same individuals. We selected 144 right-handed participants to ensure that the language areas are located in the left hemisphere (being right-handed ensures that the language areas are located in the left hemisphere in 97% of cases). The verbal tasks used involved reading, listening and producing two levels of linguistic material: word lists and sentences. The individual difference maps between the activations obtained during sentence processing (high-level tasks) and those obtained during word list processing (reference task) were calculated to isolate the regions involved in understanding the sentences read, heard and produced.

Identifying networks and their hubs

We identified 32 regions that were both activated and leftward asymmetrical during reading, producing and listening to sentences (compared to word lists) among the 185 regions of the AICHA atlas (see figure).

Location of the 32 regions constituting SENSAAS. We can see the 3 networks identified across the 32 regions; in red the LANG network essential for sentence processing, in green the sentence processing network with a strong visual component, and in blue, the sentence processing network associated with episodic memory. F3t corresponds to the lower frontal gyrus or Broca area. STS3 and STS4 are the upper temporal furrows 2 and 3, corresponding to the Wernicke area.

With the acquisition of the fMRI of the resting-state of the same participants, we explored the network organization(s) of these 32 regions. This analysis of temporal correlations at rest made it possible to identify 3 networks, including a network comprising the essential language areas for sentence processing whose lesion causes aphasia (LANG network, in red in the figure). Graph analysis of LANG network showed that the inferior frontal gyrus – corresponding to the Broca area – and two posterior regions of the superior temporal sulcus – corresponding to the Wernicke area – are hubs, i.e. crucial centres for information exchange within LANG. We also found a relationship between the resting-state connectivity of LANG regions and the amplitude of their activations: the more a region integrates and transmits information, the larger its activation will be during language tasks.

The fact that resting connectivity is a marker of the recruitment of regions by language tasks opens the way to the study of language-related pathologies or to epidemiological studies of population neuroimaging in order to search for the genetic basis of language based solely on resting-state acquisition.

Complete reference

A SENtence Supramodal Areas AtlaS (SENSAAS) based on multiple task-induced activation mapping and graph analysis of intrinsic connectivity in 144 healthy right-handers. Labache L, Joliot M, Saracco J, Jobard G, Hesling I, Zago L, Mellet E, Petit L, Crivello F, Mazoyer B, Tzourio‐Mazoyer N. Brain Struct Funct 2018 Dec 07. 10.1007/s00429-018-1810-2

Contact : Nathalie Tzourio-Mazoyer

Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., Mazoyer, B., Joliot, M. (2002). Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage, 15:273-289.

Joliot M, Jobard G, Naveau M, Delcroix N, Petit L, Zago L, Crivello F, Mellet E, Mazoyer B, Tzourio-Mazoyer N (2015) AICHA: an atlas of intrinsic connectivity of homotopic areas. J Neurosci Methods 254:46–59.

Mazoyer, B., Mellet, E., Perchey, G., Zago, L., Crivello, F., Jobard, G., Tzourio-Mazoyer, N. (2016). BIL&GIN: A neuroimaging, cognitive, behavioral, and genetic database for the study of human brain lateralization. Neuroimage, 124:1225-1231.

Vigneau M, Beaucousin V, Hervé PY, Duffau H, Crivello F, Houdé O, Mazoyer B, Tzourio-Mazoyer N (2006) Meta-analyzing left hemisphere language areas: phonology, semantics, and sentence processing. Neuroimage 30:1414–1432

Price CJ (2012) A review and synthesis of the first 20 years of PET and fMRI studies of heard speech, spoken language and reading. Neuroimage 62:816–847