Participants:
This retrospective study pooled data of healthy participants across several prospectively recruiting.
All participants aged 16 years and over gave informed consent,
and consent was provided by the parent or guardian for participants aged under 16 years.
97 healthy control volunteers who were free from major neurologic,
psychiatric or medical disorders were included.
Imaging acquisition:
All participants underwent MRI using a 3T GE scanner (Discovery MR750,
GE Healthcare,
Milwaukee,
WI) with 32-channel head coil.
The head was stabilized with inflated foam padding.
The 3D spoiled gradient recalled (SPGR) neuromelanin-sensitive T1-weighted images with magnetization transfer.
30 axial image slices were obtained on each individual with parallel alignment to the anterior callosum-posterior callosum line.
Image processing:
In-house Matlab-based semi-automated approaches were used to compute 1) the suprathreshold hyperintense volume of the neuromelanin-rich region (the delineation of ROIs was illustrated in Figure 1-A) and 2) the contrast to noise ratio (CNR).
Referring to a recent histologic study with a similar age range of our adult cohort,
we chose a threshold that yielded the closest approximation of NM-rich volume with this expected pigmented volume using only the healthy controls with ages from 17 to 90 years [9].
Threshold signal intensity (Thresig) was calculated as follows:
Thresig = BGmean + (multipliers × SDBG),
where BGmean is the mean background signal intensity and SDBG is the standard deviation of the background signal intensity.
The suprathreshold signal intensity volumes were then achieved by using a range of multipliers (one to five SD of the background ROIs with 0.25 SD increment) in adults that aged above seventeen [9].
To further adjust for variations in the midbrain volume across the wide age range,
we measured the midbrain volume of all the selected slices by using another in-house Matlab script.
To calculate the contrast to noise ratio (CNR) of bilateral SNs,
two ROIs were placed in the visually brightest appearing anterior and posterior SN on each side on three consecutive slices (Figure 1-B).
One larger background ROI of 6 mm radius at each slice was chosen near the midline of the pons.
We then divided the mean intensity of each ROI by the average signal of this background ROI at pons for each slice to obtain the mean CNR of the left and right SN,
and the anterior and posterior SN.
We measured the CNR using the following equation.
CNR = (SNsig - BGsig)/BGsig,
where SNsig is the intensity of semi-automatically defined ROIs in the SN and BGsig the intensity of the background ROI.
Statistical analysis:
Statistical tests were performed by using IBM-SPSS for windows (version 21.0; IBM,
Armonk,
NY).
Both linear and polynomial regression analysis were performed to assess age-associations.
To determine whether the midbrain volume influences the change of NM-rich volume,
two regression analyses were performed with and without the midbrain volume.
Since sex is not completely matched in different age groups ,
we also tested whether sex is a confounding factor using multiple regression analysis.
For subregion difference in SN (anterior vs.
posterior),
we conducted univariate analyses of variance with least-squares differences post hoc comparison.
Where data was not normally distributed,
we applied non-parametric tests.
The significance was defined at alpha=0.05.
Values are given as mean±standard deviation.