Dynamic Causal Tractography Analysis of Auditory Descriptive Naming: An Intracranial Study of 106 Patients

Abstract

Humans understand and respond to spoken questions through coordinated activity across distributed cortical networks. However, the causal roles of network engagements alternating across multiple white matter bundles remain understudied at the whole-brain scale. Using intracranial high-gamma activity recorded from 7,792 non-epileptic electrode sites in 106 epilepsy patients who underwent direct cortical stimulation mapping, we constructed an atlas visualizing the millisecond-scale dynamics of functional coactivation and co-inactivation networks during a naming task conducted in response to auditory questions. This atlas, termed the Dynamic Causal Tractography Atlas, identified functional coactivation patterns within specific time windows that were most strongly associated with stimulation-induced language and speech manifestations (p-value range: 2.5 × 10^-5 to 6.6 × 10^-14; rho range: +0.54 to +0.82). The atlas revealed that no single intra-hemispheric fasciculus was consistently engaged in all naming stages; instead, each fasciculus supported specific stages, with multiple distinct major fasciculi simultaneously contributing to each stage. Additionally, this atlas identified the specific linguistic stages and fasciculi where handedness effects became evident. Our findings clarify the dynamics and causal roles of alternating, coordinated neural activity through specific fasciculi during auditory descriptive naming, advancing current neurobiological models of speech network organization.

Keywords: broadband high-frequency activity; diffusion tensor imaging (DTI) tractography; epilepsy surgery; intracranial electroencephalography (EEG) recording; language.

Figure 1.

Spatial distribution of intracranial electrode sampling. (A) The figure shows the number of electrodes whose artifact-free, nonepileptic intracranial EEG data were available for measurement of task-related cortical high-gamma dynamics. (B) Regions of interest (ROIs), as defined in the Desikan-Killiany atlas (Desikan et al., 2006), are presented. In Supplementary Table 2, the full names corresponding to the given abbreviations of the ROIs are provided. To generate these images, we used FreeSurfer software (https://surfer.nmr.mgh.harvard.edu/fswiki/CorticalParcellation).

Figure 2.

Functional coactivation and co-inactivation during auditory naming. Snapshots illustrate the dynamics of functional coactivation and co-inactivation occurring during an auditory naming task. Orange and yellow streamlines: intra-hemispheric and inter-hemispheric functional coactivation. Blue and green streamlines: intra-hemispheric and inter-hemispheric functional co-inactivation. (A) 220 ms after stimulus onset (association with auditory hallucination). (B) 450 ms after stimulus onset. (C) 60 ms after stimulus offset (association with receptive aphasia). (D) 380 ms before response onset (association with expressive aphasia). (E) 50 ms before response onset (association with speech arrest). (F) 430 ms after response onset (association with face sensorimotor symptoms). For a comprehensive overview of the network dynamics, please refer to Supplementary Video 1. L: left. R: right.

Figure 3.

Probability map of stimulation-induced language-related symptoms. (A) Auditory hallucinations. (B) Receptive aphasia. (C) Expressive aphasia. (D) Speech arrest. (E) Face sensorimotor symptoms.

Figure 4.

Association between functional coactivation and stimulation-induced symptoms. Each plot displays Spearman’s rho, indicating the strength of the correlation between the probability of a given stimulation-induced manifestation and the mean functional coactivation intensity over time bins. (A) Auditory hallucination. (B) Receptive aphasia. (C) Expressive aphasia. (D) Speech arrest. (E) Facial sensorimotor symptoms. Horizontal bars indicate significant correlations based on a Bonferroni-corrected p-value of less than 0.05.

Figure 5.

Dynamics of functional coactivation through each fasciculus. The bar height represents the proportion of functional coactivation within each fasciculus for a given time bin, while the bar color reflects the average intensity of functional coactivation within each fasciculus. (A, B) Left intra-hemispheric pathways. (C, D) Right intra-hemispheric pathways. (E) Inter-hemispheric pathways. Supplementary Figure 5 presents the dynamics of functional co-inactivation.

Figure 5.

Dynamics of functional coactivation through each fasciculus. The bar height represents the proportion of functional coactivation within each fasciculus for a given time bin, while the bar color reflects the average intensity of functional coactivation within each fasciculus. (A, B) Left intra-hemispheric pathways. (C, D) Right intra-hemispheric pathways. (E) Inter-hemispheric pathways. Supplementary Figure 5 presents the dynamics of functional co-inactivation.