NeuroLang: Representing Neuroanatomy with Sulcus-Specific Queries Antonia Machlouzarides-Shalit, Nikos Makris, Gaston Zanitti, Valentin Iovene, Guillaume Lemaitre, Guillaume Favelier, Demian Wassermann

To cite this version:

Antonia Machlouzarides-Shalit, Nikos Makris, Gaston Zanitti, Valentin Iovene, Guillaume Lemaitre, et al.. NeuroLang: Representing Neuroanatomy with Sulcus-Specific Queries. Organization of Human Brain Mapping, Jun 2020, Montreal, Canada. ￿hal-02879734￿

HAL Id: hal-02879734 https://hal.archives-ouvertes.fr/hal-02879734 Submitted on 24 Jun 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. NeuroLang: Representing Neuroanatomy with Sulcus-Specific Queries Antonia Machlouzarides-Shalit1, Nikos Makris2, Gaston Zanitti1, Valentin Iovene1, Guillaume Lemaitre1, Guillaume Favelier1, Demian Wassermann1 1 Parietal Team, CEA, Inria Université Paris-Saclay, France 2 Psychiatry Neuroimaging Laboratory, BWH, HMS,Boston, United States Contact - [email protected] http://team.inria.fr/parietal/

Objective: To identify and label subject-specific sets of sulci. We developed NeuroLang, a query-based mapping language which labels sulci according to the spatial relationships to primary sulci. We assess sensitivity of our sulcus-specific queries and prevalence of tertiary sulci in a population.

(open in internet Poster explainer video:provider for hyperlink) Tertiary sulci have high variability in their existence and 1 Introduction https://youtu.be/bIgX-80mKKw morphology, while their location remains relatively constant. They are usually omitted from template atlases or else grouped with Sulci may vary greatly in morphology, while their relative 1 locations to primary sulci are what define them. their neighbouring gyri . Tertiary sulci can have relationships with cognitive functions, Template atlases use this characteristic to find landmarks of 2 the brain and wrap around it to label the same set of sulci for comparative neuroanatomy or cytoarchitectonic boundaries . any subject. 2 Mindboggle3 was used to extract an average of 33 sulci per Materials & Method hemisphere in 52 subjects of the Human Connectome Project. NeuroLang has 36 sulcus-specific queries which identify and label sulci on an individual level, designed in relation to - 6 subjects each subject's primary sulci using the Destrieux atlas to ensure reliability.

20 sulci had Destrieux atlas counterparts (secondary sulci) and 16 were omitted from the Destrieux atlas or included as part of their surrounding gyrus (tertiary sulci). Fig 1. Example of the 33 (LH) and 37 (RH) unlabelled folds in subject 212823 of the Human Connectome Project, extracted using mindboggle3.

Seconday queries 3 Results (Fig. 2) Left Q_cingulate, 0.88 Right Q_cingulate, 0.94 1 0.75 Queries were assessed 0.5 Q_anterior_occipital, No corresponding Destrieux sulcus by the proportion of the 0.25 Q_anterior_occipital, R_S_oc_temp_lat most common Q_callosomarginal, R_S_central x=-40 x=-14 x=-10 x=8 x=10 x=12 0 Q_callosomarginal, No corresponding Destrieux sulcus Destrieux match with Left Q_paracingulate, 0.85 Right Q_paracingulate, 0.83 Q_collateral, R_S_orbital_lateral the query result. Q_collateral, No corresponding Destrieux sulcus Some query results had Q_inferior_frontal, No corresponding Destrieux sulcus Q_inferior_temporal, R_S_temporal_inf no match with a Q_inferior_temporal, L_S_temporal_inf Destrieuxhttp://team.inria.fr/parietal/ sulcus, x=-40 x=-14 x=-10 x=10 x=34 x=38 Q_intralingual, No corresponding Destrieux sulcus possibly due to the fact LeftL Q_lunate, 0.81R L R L R RightL Q_lunate, 0.19R L R L R Q_intraparietal, No corresponding Destrieux sulcus Q_jensen, No corresponding Destrieux sulcus that the extracted sulci Q_jensen, L_S_interm_prim_Jensen had a wider variety in Q_middle_frontal, No corresponding Destrieux sulcus Q_middle_frontal, R_S_front_middle shape and size. Q_middle_frontal, L_S_front_inf Hemisphere y=-83 y=-45 y=-15 y=-89 y=-87 y=-85

Q_occipitotemporal, L_S_temporal_inf Left LeftL Q_inferior_occipital,R L 0.81 R L R RightL Q_inferior_occipital,R L 0.98 R L R Q_occipitotemporal, R_S_temporal_inf Right Q_olfactory, No corresponding Destrieux sulcus Fig 3. Q_orbital_H_shaped, No corresponding Destrieux sulcus Probability Q_postcentral, L_S_central maps of the Query Name, Most Common Match Q_postcentral, R_S_central z=-26 z=0 z=24 z=-32 z=-6 z=18 Q_precentral, L_S_front_sup Tertiary queries results of Left Q_superior_parietal, 0.85 Right Q_superior_parietal, 0.15 Q_precentral, R_S_precentral_sup_part L R L R L R L R L R L R (Fig. 3) some of the Q_subparietal, L_S_subparietal Q_subparietal, R_S_subparietal Queries were tertiary Q_superior_frontal, L_S_front_sup assessed by success queries, Q_superior_frontal, R_S_front_sup of their locations in thresholded Q_superior_rostral, No corresponding Destrieux sulcus z=14 z=32 z=50 z=20 z=46 z=70 Q_superior_temporal, L_S_temporal_sup the probability maps. LeftL Q_rhinal, 0.77R L R L R RightL Q_rhinal, 0.9R L R L R at 0.1, and Q_superior_temporal, R_S_temporal_sup Examples from each Top the 0.0 0.2 0.4 0.6 0.8 lobe are shown, with Proportion proportion Fig 2. Bar plot of results for secondary queries. Each the proportion of of subjects query is labelled, next to the Destrieux sulcus which was subjects with a result. z=-32 z=-30 z=-6 z=-38 z=-36 z=-12 with results. most often matched with the result of the query.

4 Discussion & Conclusions NeuroLang includes the identification and labelling of lesser- labelled sulci which can contribute to the uniqueness of a brain. We present a new method for the labelling of a subject- specific atlas of sulci, with varying sets of sulci according to Individually, subject-specific sulci sets may shed more light on individual cortical organisation. structure-function relationships3.

NeuroLang is intended as a complement to current On the population level, tertiary sulci statistics may aid in template-based methods for brain mapping. understanding the evolution of the human brain4.

References [1] Destrieux, C., Fischl, B., Dale, A. and Halgren, E. (2010). ‘Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature.’ NeuroImage, 53(1), pp.1-15. [2] Rademacher, J., Caviness, V., Steinmetz, H. and Galaburda, A., 1993. Topographical Variation of the Human Primary Cortices: Implications for Neuroimaging, Brain Mapping, and Neurobiology. Cerebral Cortex, 3(4), pp.313-329. [3] Klein A, Ghosh SS, Bao FS, Giard J, Hame Y, Stavsky E, Lee N, Rossa B, Reuter M, Neto EC, Keshavan A. (2017) Mindboggling morphometry of Acknowledgements: human brains. PLoS Computational Biology 13(3): e1005350. [4] Armstrong, E., Zilles, K., Curtis, M. and Schleicher, A., 1991. Cortical folding, the lunate sulcus and the evolution of the human brain. Journal of This work acknowledges the support of ANR NeuroRef and Human Evolution, 20(4), pp.341-348. ERC-StG NeuroLang OHBM 2020, Virtual PARIETAL - INRIA - FRANCE