Advanced lesion symptom mapping analyses and implementation as BCBtoolkit

Abstract

Background: Patients with brain lesions provide a unique opportunity to understand the functioning of the human mind. However, even when focal, brain lesions have local and remote effects that impact functionally and structurally connected circuits. Similarly, function emerges from the interaction between brain areas rather than their sole activity. For instance, category fluency requires the associations between executive, semantic, and language production functions.

Findings: Here, we provide, for the first time, a set of complementary solutions for measuring the impact of a given lesion on the neuronal circuits. Our methods, which were applied to 37 patients with a focal frontal brain lesions, revealed a large set of directly and indirectly disconnected brain regions that had significantly impacted category fluency performance. The directly disconnected regions corresponded to areas that are classically considered as functionally engaged in verbal fluency and categorization tasks. These regions were also organized into larger directly and indirectly disconnected functional networks, including the left ventral fronto-parietal network, whose cortical thickness correlated with performance on category fluency.

Conclusions: The combination of structural and functional connectivity together with cortical thickness estimates reveal the remote effects of brain lesions, provide for the identification of the affected networks, and strengthen our understanding of their relationship with cognitive and behavioral measures. The methods presented are available and freely accessible in the BCBtoolkit as supplementary software [1].

Figure 1:

Category fluency performance (mean performance with 95% confidence intervals [CIs]) for patients with (dark gray) or without (light gray) disconnection of each tract of interest. The green intervals indicate the range of controls’ performance corresponding to 95% CIs. * P < 0.05.

Figure 2:

Areas directly disconnected by the lesion that significantly contributed to a decreased score on category fluency task (referred to as “disconnected areas” in the manuscript). A) Representative slices from disconnectome maps computed for category fluency performance; blue clusters indicate group average low performance and red clusters indicate high performance. B) Brain areas contributing significantly after correction for multiple comparisons. C) Category fluency performance (mean performance with 95% confidence intervals [CIs]) for patients with (dark gray) or without (light gray) disconnection of each of the examined cortical regions. The green interval indicates performance in matched controls with 95% CIs. Abbreviations: A, anterior group of thalamic nuclei; LH, left hemisphere; IPs, intraparietal sulcus; MFg, middle frontal gyrus; pars Op., frontal pars opercularis; PreSMA, presupplementary motor area; RH, right hemisphere; VA, ventral anterior; VLa, ventrolateral anterior; VLp, ventrolateral posterior. * P < 0.05 Bonferroni-Holm corrected for multiple comparisons.

Figure 3:

Areas classically activated with functional magnetic resonance imaging (P < 0.01 corrected for False Discovery Rate) during fluency (pink) and categorization (cyan) tasks. Areas involved in both fluency and categorization are highlighted in dark blue. Abbreviations: A, anterior group of thalamic nuclei; IPs, intraparietal sulcus; MFg, middle frontal gyrus; PreSMA, presupplementary motor area; VA, ventral anterior; VLa, ventrolateral anterior; VLp, ventrolateral posterior.

Figure 4:

Functional networks involving the identified disconnected areas, as defined by resting state functional connectivity. Top panel, Main cortical networks involving the disconnected areas revealed by a principal component analysis. Bottom left panel, Principal component analysis of the raw functional connectivity result. Bottom right panel, Strength of the functional connectivity for patients with (dark gray) or without (light gray) involvement of the functional network. Abbreviations: CO, cingulo-opercular network; CS, cortico-striatal network; VFP, ventral fronto-parietal network. * P < 0.05; **, P < 0.01.

Figure 5:

Dimensional relationship between cortical thickness measured in rs-fMRI disconnected networks and category fluency. Note that regression lines (in black) and intervals (mean confidence intervals in red) are for illustrative purposes since we performed a rank-order correlation. Abbreviations: IPs, intraparietal sulcus; MFg, middle frontal gyrus.