Mapping functional connectivity in patients with brain lesions

Abstract

Objective: The spatial distribution of functional connectivity between brain areas and the disturbance introduced by focal brain lesions are poorly understood. Based on the rationale that damaged brain tissue is disconnected from the physiological interactions among healthy areas, this study aimed to map the functionality of brain areas according to their connectivity with other areas.

Methods: Magnetoencephalography recordings of spontaneous cortical activity during resting state were obtained from 15 consecutive patients with focal brain lesions and from 14 healthy control subjects. Neural activity in the brain was estimated using an adaptive spatial filtering technique. The mean imaginary coherence between brain voxels was then calculated as an index of functional connectivity.

Results: Imaginary coherence was greatest in the alpha frequency range corresponding to the human cortical idling rhythm. In healthy subjects, functionally critical brain areas such as the somatosensory and language cortices had the highest alpha coherence. When compared with healthy control subjects, all lesion patients had diffuse or scattered brain areas with decreased alpha coherence. Patients with lesion-induced neurological deficits displayed decreased connectivity estimates in the corresponding brain area compared with intact contralateral regions. In tumor patients without preoperative neurological deficits, brain areas showing decreased coherence could be surgically resected without the occurrence of postoperative deficits.

Interpretation: Resting state coherence measured with magnetoencephalography is capable of mapping the functional connectivity of the brain, and can therefore offer valuable information for use in planning resective surgeries in patients with brain lesions, as well as investigations into structural-functional relationships in healthy subjects.

Figure 1

Frequency spectrum and spatial distribution of imaginary coherence (IC). A. The mean (± standard error of mean) imaginary coherence across all connections between 20×20×20 mm voxel pairs is plotted for all healthy subjects (N=14) and for all patients with brain tumors that did not contain functional tissue (N=10). Note the peak in the alpha frequency range. Tumor tissue was associated with significantly lower functional connectivity estimates in the alpha frequency range than non-tumor tissue of lesion patients (p=0.011) and than brain tissue of healthy controls (p=0.029). B. The average functional connectivity map of all 14 healthy subjects is superimposed on a 3-dimensional individual brain. An arbitrary threshold of >90% of the maximum IC value was used to visualize the regions with largest connectivity estimates. The Broca area and the right visual cortex are opened to demonstrate the high functional connectivity of these critical areas.

Figure 2

Functional maps obtained with magnetic source imaging or intraoperative cortical mapping as well as L-images and P-images of 4 patients with brain tumors are superimposed over their 3D-rendered individual brain. A. Twenty-five year old woman with a central paresis of the right foot due to an astrocytoma WHO grade III that infiltrated the left medial sensorimotor cortex. Note that the L-image displays a corresponding decrease in functional connectivity in the sensorimotor cortex of the right foot. B. The L-images of these 3 tumor patients without pre-surgical functional deficits indicate functional disconnection (in blue) of different proportions of the corresponding tumor tissue (graded 0–2, with 0 indicating smallest proportion with disconnection). In agreement with the L-images and the clinical status, functional cortex was mapped outside of disconnected (blue) areas by MSI and cortical mapping in all patients. In addition, L-images predicted the functional status after radical surgery: whereas patient 6 suffered from post-surgical sensible deficits in the left arm and leg, no deficits were observed in patient 1 and 9. P-images show diffuse or scattered areas with significantly lower connectivity estimates than a healthy control population, but these areas are unrelated to tumor location and brain regions with functional deficits.

Figure 3

Bar plots illustrating the percentage of patients without critical tissue in vicinity of the tumor and without functional deficits after tumor resection, in relation with the functional disconnection score derived from L-images.