next up previous contents
Next: Scalar Maps Up: DTI Analysis and Fiber Previous: DTI Analysis and Fiber   Contents


Tensor Estimation and Smoothing




Image tensorcalculate Tensor estimation represents the pre-processing step before fiber tracking. It also allows you to visualize scalar maps such as Fractional Anisotropy (FA) or Apparent Diffusion Coefficient (ADC). You can perform this estimation by clicking the ``Estimate Tensors'' button. This will not display anything on screens. To display tensors as ellipsoids please see the next chapter concerning the Tensor Viewer Module. Once the estimation is done, you can display scalar maps to evaluate your results (see Sec. 3.4.2).

Image tensorsave You have the possibility to save the tensors independently (apart from the DTI-study). They are saved in a home made format called Inrimage and are compressed to save memory (extension .inr.gz).

Image tensoropen You will be able to load these tensors afterward, and use them to track fibers apart from any study or in another one.




Fig. 2.8 shows a part of the Processing panel where you can set some parameters for processing estimation. The Background Removal Threshold is the MRI signal threshold underwhich tensors will not be estimated (based on the B image). For instance you may move the slider down if the FA map (see Sec. 3.4.2) seems to contain too many empty pixels.

For noisy data, you may need to use the smoothing filter. It provides anisotropic tensor smoothing. For that you have to click ``Use Smoothing'' (see Fig. 2.8) and choose between the three different noise levels of the data (``High'' means highly noisy data). Then perform tensor estimation. You have to note that when using the smoothing filter, tensor estimation is very time consuming. Hence you may not choose this option when time is a limiting factor. Once the estimation is done, you should save the tensor field separatly (or with the study) for future use.


next up previous contents
Next: Scalar Maps Up: DTI Analysis and Fiber Previous: DTI Analysis and Fiber   Contents
Nicolas Toussaint 2007-06-22