Fuzzy Fibers: Uncertainty in dMRI Tractography - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 84

reproducibility [ 21 , 64 ]. Selecting streamlines from a whole brain tractography via
three-dimensional regions of interest [ 5 ] or semi-automated clustering [ 63 ]isan
alternative way to reproducibly extract fiber bundles.
When the same person places the seeds on a repeated scan, the resulting variability
is generally higher than when different observers follow a written protocol to place
seeds in the same data [ 13 ]. Within the same session, measurement noise is the main
limiting factor [ 14 ]. Between sessions, differences in exact head positioning and
other subtle factors increase the variability noticably [ 64 ].
Reproducibility suffers even more when repeating the measurement on a different
scanner [ 49 ]. Even a pair of nominally identical machines has produced a statistically
significant bias in Fractional Anisotropy [ 64 ]. Improving calibration between ses-
sions or scanners via software-based post-processing appears possible [ 64 ], but has
not been widely explored so far.
More time consuming measurement protocols generally afford better repro-
ducibility. Even though Heiervang et al. [ 21 ] report that the improvement when using
60 rather than 12 gradient directions was not statistically significant, Tensouti et al.
[ 61 ] report a clear improvement between 6 and 15 directions, which continues—at
a reduced rate—when going to 32 directions. Farrel et al. [ 16 ] use 30 directions and
demonstrate a clear improvement when averaging repeated measurements.
Finally, reproducibility depends on the tractography algorithm [ 61 ], its exact
implementation [ 9 ], as well as the methods used for pre-processing the data [ 34 ,
64 ] and their parameter settings. Given that the reproducibility of tractography will
be crucial for its wider acceptance in science and medicine, more work is needed
that specifically targets these problems.
8.5 Conclusion
Reproducibility of dMRI tractography is a fundamental problem that limits the accep-
tance of this technique in clinical practice and neuroscience research. Although some
effort has been made to include uncertainty information in the tractography results,
several open issues remain that need further investigation.
Probabilistic tractography is established, but visualization research has concen-
trated on deterministic streamline-based techniques, and few techniques have been
developed to visualize the information obtained by probabilistic methods. There are
several sources of uncertainty in the tractography visualization pipeline. However,
only a few of them have been explored, and if at all studied, they are often consid-
ered independently with no connection to each other. Techniques that investigate the
impact of parameters on the fiber tracking results and that aim to reduce the impact
of user bias through parameter selection have been investigated only recently. Model
selection and data preprocessing have hardly been studied with respect to their effects
on tractography results.
Techniques that allow the combined analysis of uncertainty from different sources
in the same framework, and that facilitate an understanding of their influence on the