code written: 2020-01-12
last ran: 2020-06-17

Description: This report summarizes Slicer segmentation success, for all tracts, for all participants. We have data for n=40 tracts. We expect to have data from n=41 tracts. We have data for n=445 out of an expected total of n=445 participants (note: I have not removed participants who are ineligible, etc., from this analysis, in an effort to derive the most fulsome understanding of segmentation success).

Table: Fiber summary.

Below, all data is collapsed across site. The count variable indicates the number of participants with data for a given tract, and percent indicates corresponding percentage. Note that the table does not show data for tracts missing across all participants: all participants are missing the following tracts: CPC – LR. In total, we have data for 29687 tracts out of a possible maximum 32040, i.e., 92.6560549%. Again, we see that the SLF I and II, and the IOFF, can not be reliably segmented. The right AF is also less than ideal.

Visualization: Complete missingness.

Below is a visual summary of missing data. Black indicates that the tract is missing entirely for the given participant. Again, tracts missing across all participants are not shown here: CPC – LR. We see that the SLFs remain problematic across all sites. We see that the IOFF is not problematic at CMH or ZHH, but is at all of the Siemens sites.

Table: Missing tracts by participant. A large number of participants are missing several tracts, as follows:

Tracts missing 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 27 51
Percent missing 0.00 1.39 2.78 4.17 5.56 6.94 8.33 9.72 11.11 12.50 13.89 15.28 16.67 18.06 19.44 20.83 22.22 23.61 25.00 27.78 29.17 30.56 31.94 37.50 70.83
Participant count 10 37 83 82 53 31 22 22 22 13 12 21 8 7 4 3 4 2 2 1 2 1 1 1 1

Visualization: Number of Fibers.

The following plot shows the number of fibers segmented per tract, for each of the n=445 participants, separated by hemisphere (n=74), and coloured by site / scanner. Again, tracts missing across all participants are not shown here: CPC – LR. We see small values and high variability across tracts and participants; however, results here are the most consistent yet.