code written: 2019-09-30
last ran: 2019-12-04
website: http://rpubs.com/navona/NM_SN_methods
github: https://github.com/navonacalarco/CurAge/blob/master/scripts/analysis/00_NM_ROI_SN_methods.Rmd

Methods

Acquistion. All data were acquired on the Siemen’s 3T Prisma at ToNI, with 15 contiguous slices and 4 measurements at a voxel size of .4x.4x.3mm\(^3\).

Processing. Images from the 4 measurements were registered to the first image using a linear transform and averaged (FSL). The data was smoothed in MATLAB.

Segmentation of SN. I selected all slices that I judged to display the most visible hyperintensity in the anatomic region of the SN. Segmentation of NM images was performed in MATLAB, mostly in keeping with the methods delineated in the Chen paper, and with a script provided by Sofia Chavez. Modified code can be found in NM_ROI_SN.m. Our process was:

  1. Two reference ROIs (circles with diameter of 4mm) were placed adjacent to the SN. [FIGURE A]. The values from these two reference ROIs were averaged into a single reference ROI.

  2. Voxel intensities in the reference ROI were verified as approximately normally distributed.

  3. Signal intensity values were binarized according to thresholds calculated using the reference ROI \(\mu\)\(_{REF}\) (mean signal intensity) and \(\sigma\)\(_{REF}\) (standard deviation). For segmentation, voxels with a signal hyperintensity I > 3\(\sigma\)\(_{REF}\) were considered to be plausibly part of the SN [FIGURE B].

  4. Manual segmentation [FIGURE C] was performed to isolate the SN [FIGURE D] in accordance with its known anatomical characteristics.

Example images from SEN015, slice 9.

Example images from SEN015, slice 9.

Derived values

  1. Signal intensity. Signal intensity in the reference region and ROI reflects a biological property of the tissue (i.e., it is not calculated). However, it varies as a function of a number of scanner parameters, so we should not necessarily expect our values to match with data from other groups/scanners.

  2. Volume. We calculated volume as number of voxels x voxel size. Chen reports volumes of 338 - 1671mm\(^3\) (6 participants).

  3. CNR. CNR (contrast-to-noise ratio) was calculated as \(\mu(\dfrac{I_{SN_{voxel}} - I_{\mu_{REF}}}{I_{\sigma_{REF}}})\), where I is signal intensity. Chen reports a CNR range of approximiately 2.8 to 3.2 (6 participants).

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.