The olfactory system shows a pronouced neuroplacticity. A decline in olfactory performance is accompanied by a volumetric reduction of the bulbus olfactorius as well as reduction in gray matter of olfactory related cortical areas¹. An improvement in previously hampered olfactory performance can be associated with a volumetric increase in these aforementioned regions.The purpose of this study was to investigate neuronal correlates by means of diffusion tensor imaging (DTI) for different olfactory recovery after sinus surgery.

This study involves 31 patients (51.98±12.2 yrs, 20 male, 11 female), which underwent pan sinus surgery. Olfactory performance tests aswell as MR imaging was performed on the same day once before and once approx. 3 months (98±9.4 d) after surgery. Olfactory function was determined birhinally using the “Sniffin’Sticks” as a combined score based on a test for butanol odor threshold, odor discrimination, and odoridentification (TDI score)². The MR imaging included anatomical scans (MP-RAGE, 1mm³) and structural scans (diffusion tensor imaging, 70 dw directions at b=1000 s/mm², 2.2 mm³ resolution, 6x b=0). MP-RAGE data were evaluated using the Freesurfer (version 5.2) cortical segmentation pipeline. The first step was cross-sectionally processing of all time points (two per subject). In the second step an unbiased template from all time points for each subject was created⁴. In the third step all time points were process longitudinally using the unbiased template created in step 2 foreach subject separately. From the white matter segmentation after step 3, we selected the parahippocampal white matter ROI and the rostral anteriorcingulate white matter ROI for the left as well as for the right hemisphere in order to perform probabilistic tractography between these two ROIs to extract the cingulum bundles. Probabilistic tractography was carried out using MRtrix3 (version 0.2.11) and non-negativity constrained super resolvedspherical deconvolution⁵ (cf. Fig. 1a). Three control points (cf. Fig. 1b), which labeled the begin, the end and the half-waypoint of each cingulum bundle were determined and normalized within each subject using the linear registration information from the longitudinal processing pipeline. The registration information between anatomical (MPRAGE) and structural (DTI) data within each MR study was obtained using bbregister⁶. The final length normalization between all subjects was performed by linearly transforming all mean tracts to the Tailairach spaceand placing perpendicular oriented cut-planes with respect to the mean orientation at the begin and the end of the tract. The normalized tracts were finally transformed back to the subject space. Changes in diffusion properties (e.g. FA, radial + parallel diffusion, ADC) between both time points were estimated by calculating the symmetric percentage change (SPC)³. Before extracting the SPC values along the cingulum paths data were smoothed using a spherical kernel (4.4 mm radius).Statistical differences between the groups of good performer (TDI diff after surgery > +10) and bad performed (TDI diff after surgery <+10) was observed for symmetric percentage change (SPC) of radial diffusion (c.f. Fig. 2) in the middle segment of the left cingulum. We observed, while less significant, similar results for the right cingulum. At this specific location good performer had increased radial diffusivity, while fractional anisotropy was decreased.To the best of our knowledge, this is the first study which investigated longitudinal changes in olfactory performance due to pan sinus surgery in combination with diffusion tensor imaging. Currently, the observed significant correlation between good and poor recovery of olfaction and diffusion measures are not well understood. The posterior cingulate cortex is not known to be directly involved in olfaction.