gms | German Medical Science

Objective: Functional connectivity analysis of resting-state fMRI (fcrs-fMRI) has been shown to be a robust non-invasive method for localization of functional areas and networks throughout the brain on an individual level. However, its broader clinical use for preoperative planning is impeded by the fact that existing analysis software is designed for brain researchers. We developed a prototypical interactive real-time visualization tool to enable neurosurgeons to explore functional areas and distributed networks, and evaluated its usability.

Methods: fcrs-fMRI and anatomical data were acquired from a healthy participant and two tumor patients by measuring changes in BOLD signal without specific task demand. Our implementation of a triplanar view for the preprocessed data builds upon LIPSIA, an established brain research software. The resulting real-time exploration of functional connectivity allows for integrated selection of a seed point of interest, the computation of voxelwise correlation, and the visualization of functionally connected areas. The tool was tested by four neurosurgeons previously unacquainted with the method. All users were asked to identify four prominent cortical networks (sensorimotor, language, dorsal-attention, and default-mode) in all cases after a brief introduction to their normal appearance using a schematic diagram. The resulting correlation maps were visually compared with results obtained by an expert in fcrs-fMRI analyses.

Results: With the exception of the language network, the neurosurgeons were able to consistently reproduce the expert's reference networks in the healthy and the pathological brains. The speed of feedback allowed for dragging the seed-point through the data in an exploratory fashion, while simultaneously observing the resulting correlations. This immediate feedback helped to pinpoint functional networks intuitively and quickly by maximizing recognizable correlation patterns. It took the neurosurgeons on average less than three minutes to identify a functional network, and they were able to report disturbances of the networks including tumor related displacement of the sensorimotor network.

Conclusions: Real-time interactive software is a promising tool to explore functional connectivity networks. It has the potential to allow for non-invasive presurgical exploration of well-known networks (e.g. sensorimotor and language), as well as for networks less typically described in neurosurgical practice (e.g.dorsal-attention, and default-mode network).