Accurate Localization of Brain Activity in Presurgical fMRI by Structure Adaptive Smoothing

• Published in: IEEE transactions on medical imaging 2008-04, Vol.27 (4), p.531-537
• Main Authors: Tabelow, K., Polzehl, J., Ulug, A.M., Dyke, J.P., Watts, R., Heier, L.A., Voss, H.U.
• Format: Article
• Language: English
• Subjects: Adaptive estimation; Algorithms; Biomedical imaging; Brain; Brain Mapping - methods; Brain Neoplasms - diagnosis; Brain Neoplasms - surgery; Epilepsy; functional magnetic resonance imaging (fMRI); Humans; Image Enhancement - methods; Imaging, Three-Dimensional - methods; Levee; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Neoplasms; neurosurgical planning; Pattern Recognition, Automated - methods; Preoperative Care - methods; Reproducibility of Results; Sensitivity and Specificity; Shape; Signal to noise ratio; Smoothing methods; spatial filtering; Surgery
• ISSN: 0278-0062; 1558-254X
• DOI: 10.1109/TMI.2007.908684

An important problem of the analysis of functional magnetic resonance imaging (fMRI) experiments is to achieve some noise reduction of the data without blurring the shape of the activation areas. As a novel solution to this problem, recently the propagation-separation (PS) approach has been proposed. PS is a structure adaptive smoothing method that adapts to different shapes of activation areas. In this paper, we demonstrate how this method results in a more accurate localization of brain activity. First, it is shown in numerical simulations that PS is superior over Gaussian smoothing with respect to the accurate description of the shape of activation clusters and results in less false detections. Second, in a study of 37 presurgical planning cases we found that PS and Gaussian smoothing often yield different results, and we present examples showing aspects of the superiority of PS as applied to presurgical planning.