16 landmarks of the mouse brain have been validated as fiducials for registration to WHS
Marina Sergejeva (Institute of Pharmacology, FAU Erlangen), Rembrandt Bakker (Radboud University Nijmegen), André Gaudnek (Institute of Pharmacology, FAU Erlangen), Yuko Okamura-Oho (Brain Research Network and RIKEN Advanced Science Institute), Jyl Boline (Informed Minds), Andreas Hess (Institute of Pharmacology, FAU Erlangen)
We need convenient methods that permit researchers to register their own data to WHS. Existing automatic registration processes are rather complex. In addition, we propose an easier and faster approach: registration based on well recognizable brain landmarks (LMs) or fiducials.
Here we present a set of LMs validated as fiducials, as they were reliably identified • by different individuals (anatomy specialists and novice) • in different MR imaging modalities (T1, T2, T2*) • in various specimens • by different cutting directions • by different image resolutions.
On coronal MR datasets (T1, T2, T2*, 256x256x128, 80 μm) from an adult C57BL/6J male we defined an initial set of LMs recognizable in all 3 modalities and rendered descriptions how to find them. 15 guessers identified these LMs according to the descriptions on datasets from two C57BL/6J males, visualized in ImageJ as coronal slices. The probability of finding them, mean values for x, y and z coordinates and deviations from the mean were calculated for every LM. Finally, we excluded LMs with a deviation of more than 1,5 voxels in the x and y directions in both animals and ended up with 16 potential fiducials. Their average deviations were: 1,0 (x), 0,6 (y) and 1,5 (z), the probability of finding was > 95%.
Further, we located these 16 LMs on the canonical WHS datasets, and presented them in the web-based atlasing tool Scalable Brain Atlas (http://scalablebrainatlas.incf.org/WHS10). WHS images have a four times higher resolution and a different inclination than ours, but despite the differences all LMs were well identifiable according to our definitions. This supports their validity as fiducials.
We also evaluated “the classical” LMs, Bregma and Lambda derived from the skull 3D µCT datasets coregistered to brain MRI datasets of five mice. We found that positions of these LMs with respect to brain anatomy vary considerably between the mice. The largest distance between Bregma z-positions was 1,2 mm, and between Lambda z-positions - 1,68 mm. Thus, we cannot accept these two LMs as fiducials.