Elsevier

NeuroImage

Volume 24, Issue 2, 15 January 2005, Pages 323-331
NeuroImage

Influence of fMRI smoothing procedures on replicability of fine scale motor localization

https://doi.org/10.1016/j.neuroimage.2004.08.042Get rights and content

Recent publications analyzing the influence of spatial smoothing on fMRI brain activation results demonstrated that smoothing may artificially combine activations from adjacent though functionally and anatomically distinct brain regions and that activation from large draining vessels may be smoothed into neighboring neuronal tissue. To investigate whether functional localizations may be artificially shifted by the smoothing procedure we performed replicability measurements. Localization centers of motor hand activations achieved during different conditions (isolated hand movements and simultaneous hand and chin movements) were compared with respect to smoothing effects. The voxel with the highest probability to represent a true positive activation was localized with a non-smoothed and a standard 4 × 4 × 6 mm smoothed correlational data analysis technique. Results show an increase of motor center aberrations between measurements by about 100% due to data smoothing indicating a statistically significant decrease in localization replicability.

Introduction

In primary motor cortex and other essential cortical areas, the functional significance of neuronal tissue may change within millimeters. Neurosurgical literature described that a mean change in resection margins by about 1 cm may decide between no or permanent postoperative functional deficits and the transfer of a reversible to a permanent deficit is associated with a mean shift of resection margins by only about 1 mm (Haglund et al., 1994). In addition, fine scale somatotopic investigations using fMRI demonstrated that representation centers for different finger movements may be only 2–3 mm apart (Beisteiner et al., 2001, Hlustik et al., 2001, Indovina and Sanes, 2001). As a consequence, a localization change of fMRI activations due to spatial smoothing might be problematic. Since recent studies analyzing the influence of spatial smoothing on fMRI brain activation results (Fransson et al., 2002, White et al., 2001) imply that such artificial localization changes may occur, we were interested whether smoothing procedures—as often used in clinical studies (Lawrie et al., 2002, Rocca et al., 2002, Staffen et al., 2002)—affect the localization of functional centers. Extraction of functional centers refers to previous publications on hand motor localization (Beisteiner et al., 2000) and Broca language area localization (Rutten et al., 1999) which indicate that within extended representation areas single voxels with the largest probability for a true positive may be defined which correlate best with intraoperative cortical stimulation. Knowledge about localization of such a functional center is important for clinical studies, since it represents the area of the highest risk for a postoperative functional deficit when damaged.

We examined the replicability of the hand motor center with a non-smoothed and a smoothed data analysis technique using standard smoothing values. To test the center stability under difficult circumstances—namely with varying overall activation patterns—two different movement conditions were used. Isolated hand movements were compared with simultaneous hand and chin movements. Chin movements are of clinical relevance for overt speech studies and are prone to produce head motion artifacts which might influence motor center replicability differently with regard to the amount of data smoothing.

In extension to previous studies (e.g., White et al., 2001), this study presents the first investigation of smoothing effects on fine scale localization using replicability measurements. To allow inferences about clinical relevance, two patients have been included and data are analyzed individually. Anatomical localization errors (White et al., 2001) are avoided by analyzing functional activations without normalization directly on original EPI images.

Section snippets

Subjects/patients

11 healthy right handed subjects (mean age: 25.3 years; 7 male/4 female) and two right handed patients (mean age: 37.5 years; 1 male/1 female) participated in this study. Patient FI suffered a recurrence of a left frontoparietal astrocytoma II detected during routine MR diagnostics 3 months after first operation. Initially, the patient had presented with an epileptic seizure but was seizure free postoperatively. The postoperative local defect measured 2.8 cm in diameter. Patient HP suffered a 7

Results

Analysis of head motion behavior showed generally low motion effects and no difference between conditions. The mean time course of head motion over all runs is shown in Fig. 1 for the subject with the smallest and in Fig. 2 for the subject with the largest overall head motion. Kolmogorv–Smirnov tests indicated normally distributed data and t tests revealed no significant differences between condition related head motion for neither subject and neither motion form (translation, rotation).

Discussion

Recent publications analyzing the influence of spatial smoothing on fMRI brain activation results (Fransson et al., 2002, White et al., 2001) demonstrated that smoothing may artificially combine activations from adjacent though functionally and anatomically distinct brain regions and activation from large draining vessels may be smoothed into neighboring neuronal tissue. This implies that functional localizations may be artificially shifted by the smoothing procedure. To investigate this issue

Acknowledgments

We want to acknowledge important general support by Prof. Lueder Deecke, Department of Clinical Neurology, Medical University of Vienna, Head Ludwig Boltzmann Institute for Functional Brain Topography and by Prof. Siegfried Trattnig, Department of Radiology, Medical University of Vienna. This study was supported by the Austrian Science Foundation (P15102).

References (18)

There are more references available in the full text version of this article.

Cited by (0)

View full text