Spatially Adaptive Variable Screening in Presurgical fMRI Data Analysis

Abstract

Accurate delineation of tumor-adjacent functional brain regions is essential for planning function-preserving neurosurgery. Functional magnetic resonance imaging (fMRI) is increasingly used for presurgical counseling and planning. When analyzing presurgical fMRI data, false negatives are more dangerous to the patients than false positives because patients are more likely to experience significant harm from failing to identify functional regions and subsequently resecting critical tissues. In this paper, we propose a novel spatially adaptive variable screening procedure to enable effective control of false negatives while leveraging the spatial structure of fMRI data. Compared to existing statistical methods in fMRI data analysis, the new procedure directly control false negatives at a desirable level and is completely data driven. The new method is also substantially different from existing false negative control procedures which do not take spatial information into account. Numerical examples show that the new method outperforms several state-of-the-art methods in retaining signal voxels, especially the subtle ones at the boundaries of functional regions, while providing cleaner separation of functional regions from background noise. Such results could be valuable to preserve critical tissues in neurosurgery.