Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Inflated? Using different visualisation software packages influences the identification of sulcal patterns in relation to reading
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Yifan Yang1,2, Kate Nation1, Kate Watkins1,2; 1Department of Experimental Psychology, University of Oxford, 2Centre for Integrative Neuroimaging (OxCIN), University of Oxford
Introduction: A key aim of cognitive neuroscience is to understand the link between variations in brain structure and cognitive function. In the domain of reading, previous studies suggest that interruption in the left posterior occipitotemporal sulcus (pOTS) is a prenatally determined predictor of better reading ability. These gyral gaps would increase local white-matter connectivity, potentially conferring a computational advantage. We recently evaluated the effect of pOTS morphology on word-reading in a large sample (N = 153) of children and adolescents. However, we noted a much higher incidence of interruptions in our sample compared with several previous studies, and that the incidences reported in those studies were also very variable. Notably, the three studies using FreeSurfer reported overall higher interruption rates (on average 58.3%) and significant relationships between sulcal interruption and better reading, while the six using BrainVISA had lower interruption rates (on average 40.6%) and did not always find a significant relationship between sulcal interruption and better reading. Yet, no study so far has compared the two approaches or explained this difference. Methods and results: To investigate whether using different software may influence the identification of sulcal interruptions, we processed our own sample of brain scans (N = 153) using FreeSurfer and BrainVISA and manually classified the sulcal patterns. Importantly, we found that sulci were visualised differently in the two packages. BrainVISA failed to display deep intrasulcal interruptions that were visible in FreeSurfer. This difference means that sulci with these deeper interruptions are classified as ‘continuous’ in BrainVISA, but as ‘interrupted’ in FreeSurfer, thus resulting in a higher incidence of sulcal interruptions in the latter. In our sample, 99 (64.7%) sulci were classified as ‘interrupted’ in FreeSurfer compared with 73 (47.7%) in BrainVISA (χ2(1) = 8.30, p = .004). Specifically, interruptions in 21 (13.7%) cases were identified by FreeSurfer but not BrainVISA. We replicated the relationship between the presence of a sulcal interruption and better reading when using FreeSurfer (ΔR2 = 2.80%, p = .03). However, this relationship was not observed with BrainVISA classifications (p = .44). There were other differences between the two software packages in sulcal labelling that will be explored in future analyses. Discussion: Based on these observations, we suggest that different visualisations of intrasulcal structures contribute to the discrepancy in the reported frequencies of sulcal interruptions. It is important to understand how the depth of these gyral gaps affects local connectivity and, in turn, how this supports reading. Alongside this, we call for clarification on definitions of sulcal features to improve transparency and reproducibility.
Topic Areas: Reading, Morphology