Neural organization of speech production: A lesion-based study of error patterns in connected speech

Brielle C Stark; Alexandra Basilakos; Gregory Hickok; Chris Rorden; Leonardo Bonilha; Julius Fridriksson

Comparison of neural data between mutually exclusive groups. The goal of this study was to compare the results of a voxelwise lesion-symptom mapping (VLSM) analysis between the two task groups. A basic . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/544841 doi: bioRxiv preprint assumption regarding VLSM is that results will generalize beyond the specific participants included in the analysis. There are two possible challenges for this assumption when comparing two distinct groups of subjects: 1) that lesion distribution is significantly different between task groups, and 2) that the number of participants with lesion damage per parcel is significantly different between task groups. We addressed these assumptions, below, and provided a supplementary table (Supplementary Table 2 ) of lesion proportion damage and number of participants with damage to each parcel. 1) Lesion distribution was not significantly different between task groups, which we demonstrated at both the voxel-and parcel-level (gray-and white-matter parcels derived from the Johns Hopkins University [JHU] atlas, [Faria et al., 2012] ). There was no significant difference in proportion damage at the voxel-level between task groups (z-score range: -3.34 -4.10), which was calculated using a two-tailed Liebermeister test (p<0.05) and FDR correction for multiple comparison (Rorden et al., 2007) in all left hemisphere voxels where at least 10% of people had damage. Further, there was no significant difference in lesion damage to any parcel between task groups (z-score range: -1.80-2.579), which was calculated using a two-tailed t-test (p<0.05) and FDR correction for multiple comparison in all left hemisphere parcels where at least 10% of people had damage. Prior to multiple comparison correction, six parcels had a score of z>1.96 (therefore, where the PNT task group had more lesion proportion damage than the CS task group): a posterior section of the middle frontal gyrus (two-tailed z=2.579, p=0.0099), white matter (superior corona radiata) (two-tailed z=2.572, p=0.01), precentral gyrus (two-tailed z=2.517, p=0.012), postcentral gyrus (two-tailed z=2.127, p=0.033) and white matter (anterior limb of internal capsule) (two-tailed z=2.099, p=0.036). Figure 1B displays these parcels. However, given no statistical significance after multiple comparison correction, we stipulate that lesion distribution was not significantly different between task groups.