Date written: 2020-05-26
Last ran: 2020-05-27

Description: Here, we review one SPINS participant, SPN01_CMH_0135 in detail, at every step along processing, from (1) raw data preprocessing, to (2) Slicer tractography , to (3) use of the ORG atlas. This participant is a healthy control and was randomly chosen from those who passed QC and had low in-scanner motion.

Links:
Review 1: Preprocessing
Review 2: Slicer tractography
Review 3: ORG atlas

[1] Create a mask

Description. Here, we use Slicer DiffusionWeightedVolumeMasking to create a mask for tractography. As shown, the mask is true to anatomy. We used only default parameters. The code and available arguments can be reviewed here (step 1). We used Slicer version 4.8.1 on Linux.

mask

b0

[2] Fit the tensor

Description. We then performed diffusion tensor estimation with Slicer DWIToDTIEstimation. We chose weighted least squares, which weights the DWI in accordance with noise characteristics, as has previously been done in the lab. We don’t see any obvious issues with tensor fit. [Red: right-left; green: anterior-posterior; blue: inferior-superior.] The code and available arguments can be reviewed here (step 2). We used Slicer version 4.8.1 on Linux.

tensor fit

tensor fit .gif

comparison dtifit

[3] Perform tractography


[a] Command line

Description. Next, we performed whole brain tractography with Slicer TractographyLabelMapSeeding. Based on prior analyses in the lab and our comparative testing, we used default parameters in our last group analysis and this analysis of SPN01_CMH_0135, with two exceptions. First, we adjusted stopping value (the distance in millimeters below which measurements will be dropped), from the default of .25 to .1. Second, we turned on the --useindexspace argument, which a former script indicated Dr. O’Donnell had suggested. Though not excellent, our opinion was that this tractography appears to be ‘ok’ and not unlike outputs seen from other studies at other times, on the CAMH GE scanner. The code and available arguments can be reviewed here (step 3). We used Slicer version 4.10.2 on Mac.

[b] Parameter comparison

Description. Next, via the command line, we adjusted the permitted argumented allowed by TractographyLabelMapSeeding. Below, we show adjustments for 7 allowed arguments (6 continuous, 1 binary), for a total of 20 comparisons. In all cases, the default setting is shown in the middle, and we show a lower and higher value to the left and right, respectively. As indicated above, we used default parameters for all arguments with the exception of stopping value and turning on --useindexspace. As above, the default code and list of parameters can be reviewed here.

stopping value

start threshold

minimum length

maximum length

stopping curvature

step length

index space

fiber length histogram

[c] GUI method 1 of 2

Description. Next, as a sanity check, we ran tractography in the Slicer GUI (cf. command line), using the Tractography ROI Seeding module (Modules > Diffusion > Tractography > Region based > Tractography ROI selection). Again, we used all default parameters, with the exception of Stopping Threshold, which we modified from .25 to .1 (in the GUI, the --useindexspace argument is already turned on). As expected, our outputs were identical.

[d] GUI method 2 of 2

Description. As a further sanity check, we ran tractography in the Slicer GUI via a second method, this time using the Tractography Seeding module (Modules > Diffusion > Tractography > Region based > Tractography ROI selection). Again, we used all default parameters, with the exception of Stopping Threshold, which we modified from .25 to .1 (in the GUI, the --useindexspace argument is already turned on). Against expectation, Michael discovered that these output were not identical! In the case of SPN01_CMH_0135 as well as a small set of other SPINS participants tested, Tractography Seeding seems to give rise to “fuller” tracts. Michael noted the same pattern in his comparison of PNS participants. Michael has brought this issue to the attention to Fan, who confirmed via a GitHub issue that outputs across the two modules are expected to be identical. We are awaiting further resolution from Fan. For now, we will use both tractograms for subsequent white matter analysis.

static

gif

[4] Conclusion

We await direction from Fan regarding the displayed difference between the command line and the Tractography ROI Seeding GUI, and the Tractography Seeding GUI. However, it may be the case that the outputs from the command line / Tractography ROI Seeding are sufficient, as they appear comparable to past outputs from CAMH scans acquired on the GE scanner.