Developmental dyscalculia (i.e., mathematical learning disability) is a neurodevelopmental disorder specific to arithmetic learning even with normal intelligence and age-appropriate education [1]. Its prevalence has been estimated around 3 – 13.8 % within child populations, and the difficulties tend to persist even in adulthood negatively affecting the academic career and daily life. Despite its importance, the neuronal underpinnings of dyscalculia have yet to be fully understood.

Neuroimaging studies have indicated some differences in brain structure and/or function in children with developmental dyscalculia [2-5]. In this study, we investigated the neocortical features (i.e., cortical thickness, cortical surface area and cortical volume) in dyscalculic children comparing them to their typically developing (normally achieving) peers using surface-based morphometry methods.

We recruited the sample from 2058 school-age children representing various socioeconomic backgrounds in Ankara, Turkey, using a population-based longitudinal approach. We included 12 children with developmental dyscalculia (eight females, mean age: 11.18±0.66 years) and 16 typically developing peers (nine females, mean age: 11.22±0.6 years) in the MRI part. The groups were well-matched in terms of age, gender, handedness, total intelligence quotient (WISC-R corrected for arithmetic), and reading scores. After familiarizing the participants with the MRI procedure in a mock scanner, we acquired the high-resolution anatomical T1-weighted images on a 3 Tesla Siemens Magnetom Tim Trio MRI scanner.

We performed whole brain surface-based morphometry analysis using the Freesurfer software v6.0 [6]. Each individual cortex was anatomically segmented and parcellated, and registered to the Freesurfer's "fsaverage" surface map. A 10-mm full-width half-maximum Gaussian spatial smoothing kernel was applied. We corrected the errors after checking the quality of the cortical segmentation and parcellation outputs using the software recommendations. Using a general linear model, we assessed cortical thickness, surface area, and volume measures with between-group contrasts. In the volume analysis, total intracranial volume was used as a covariate. We defined statistically significance level as p<.001 for vertex-level (cluster forming) threshold and p<.05 cluster-level threshold by running Monte-Carlo Simulations with 5000 iterations to correct for multiple comparisons in addition to Bonferroni correction across the two hemispheres.

We found significant group differences (p<.001 vertex-level threshold and p<.05 cluster-level threshold, corrected for multiple comparisons by Monte-Carlo Simulations) for cortical surface area and volume measures but not cortical thickness measures for any contrast. When compared to typically developing peers, dyscalculic children had lower cortical volume in the bilateral superior frontal lobe, including the precentral gyrus and the left supramarginal gyrus in the parietal lobe. Cortical surface area followed a similar pattern, with dyscalculic children having smaller cortical surface area than typically developing peers in the right superior and middle frontal lobes including the precentral gyrus, and the left supra marginal gyrus in the parietal lobe. There were no regions showing greater cortical volume or cortical surface area in the dyscalculia group.

Cortical volume reductions in the frontoparietal regions in developmental dyscalculia were replicated in our study [7-9]. Furthermore, we revealed that lower cortical volume in dyscalculia may be associated with smaller cortical surface area rather than thinner cortex. However, because we had relatively small sample size, further studies with larger sample size are necessary to confirm our findings.