imaging data pulled: 2019-08-29
code written: 2019-09-07
last ran: 2019-09-07
website: http://rpubs.com/navona/NM_SN_thresholds_4mm

Here we compare the effect of adjusting the threshold for generating binary maps that visualize potential voxels of the SN. (Note that the previous analysis here erred in that the radius of the reference ROI selection was 5mm; it should have been 4mm. Here, we re-run the same analysis with the correct reference ROI size.)


Method

For this analysis, we compare two example participants: one with ‘good’ data (SEN011), and one with ‘poor’ data (SEN015). In both cases, we analyzed the slice upon which the SN was the most visually apparent; for both participants, this was slice 11. For SEN011, the two reference ROIs were consistently placed at [x=231, y=261] and [x=190, y=261]. For SEN015, the placement was [x=236, y=249] and [x=191, y=249].

Chen’s formula is \(I{_{diff}}\) = \(I{_{SN_{voxel}}}\) - \(\mu\)\(_{REF}\) - \(x\) * \(\sigma\)\(_{REF}\). Chen sets \(x\) = 3; here, we review the difference between \(x\) = [1:6].

Summary table

\(x\) SI \(\mu\)\(_{REF}\) SI \(\sigma\)\(_{REF}\) SI \(\mu\)\(_{SN}\) SI \(\sigma\)\(_{SN}\) Voxels Volume CNR
SEN011
1 823.54 14.92 109582.0 9160.50 1054 421.6 6.14
2 823.54 14.92 109115.9 8776.00 847 338.8 6.89
3 823.54 14.92 108940.6 8683.94 743 297.2 7.88
4 823.54 14.92 109242.6 8576.84 633 253.2 8.64
5 823.54 14.92 109070.7 8482.68 560 224.0 9.18
6 823.54 14.92 108927.8 8455.84 504 201.6 9.59
SEN015
1 816.76 21.15 110465.5 9621.89 1590 636.0 4.84
2 816.76 21.15 109806.2 9274.51 1239 495.6 5.75
3 816.76 21.15 110573.9 9546.56 1096 438.4 6.33
4 816.76 21.15 110415.1 9140.72 830 332.0 7.14
5 816.76 21.15 110921.9 9253.74 688 275.2 7.71
6 816.76 21.15 111728.3 9510.72 551 220.4 8.26

Conclusion

We see that, as \(x\) increases, the SN is more uniquely defined (particularly for ‘bad’ data), but at the cost of voxels/volume.

Citation

Chen, Huddleston, Langley, Ahn, Barnum, Factor, Levey, & Hu. (2014). Simultaneous imaging of locus coeruleus and substantia nigra with a quantitative neuromelanin MRI approach. Magnetic Resonance Imaging, 32, 1301-1306.