Key question:

-Are there differences in volume between
men and women.

- And if there are differences. Is this
explained by age?

WM volumes of the SCG will be the continues variable.
Due to potential differences between the WM volumes of men and women, the
effect of gender will be assessed.

When the 93 manual parcellations for the subjects are
done it is of importance to see if data are normally distributed and if there
is homogeneity of variances. This will be analysed by a histogram and a
residual plot. The results of the Kolmogorov Smirnov and the Shapiro-Wilk will
be used to assess normality. If these assumptions hold, an independent samples
t-test will be performed. If not, a non-parametric test such as the
Mann-Whitney test will be executed or the data will be transformed, when
appropriate. For example, with a log transformation depending on the skewness
of the data.

In addition, it is of importance to asses if there are
significant differences in the volumes of the SCG between the LH and RH. For
this a paired t-test will be used. If there is no difference between the SCG of
the LH and RH the data will be collapsed to form data of one individual.

Based on the results of the above described analyses,
it will be determined how the dataset will be treated. If there are significant
differences between de LH and RH, the datasets for each hemisphere will be
analysed separately of each other. If the datasets of the LH and RH do not
differ, the datasets can be treated as one dataset.

Furthermore, because of the possibility of differences
in age between male and female this will also be assessed with an independents
t-test. Afterwards, if the necessary assumptions hold, a regression model will
be constructed containing age and gender as explanatory variables or a
stratified analysis on age groups will be performed.

Please do the above described for the 93 subjects done
by myself.

In the single
datafile you can recognize the 93 subjects done by myself by my initials YTT in
the name of the datafile: example
sub-001_ses-1_acq-wb_mod-t1map_mask-scg_hem-l_rat-ytt_calc-bin_out.nii.gz

Then repeat the same
but with adding the remaining 14 subjects.

The remaining 14 subjects were excluded
in the previous analysis for my part of parcellations but can be included for a
second analysis by adding the 14 single others, recognizable by the initials of
my supervisor AAX:

example sub-002_ses-1_acq-wb_mod-t1map_mask-scg_hem-l_rat-aax_calc-bin_out.nii.gz

PLEASE! Check the usual
assumptions of the data first and also write something about this (so the
distribution of the data, sphericity, outliers and all the usual stuff)

I already had did some analyses concerning the outliers but lost the
files unfortunately and I can already report that if I wrote it down correctly,
in the single datafile for the Left Hemisphere the volumes for scanid-066 and
scanid-095 are outliers and should be removed cause they are human errors made
by me (did not save these files correctly). For the right hemisphere sub-038
information is missing so this right hempishere should also be excluded.
Furthermore, for the right hemisphere the data for the following participants,
scanid-065, scanid-086 & scanid-014 should also be outliers but can be
included in the analysis because they are not human errors, their subcallosal
gyrus just happens to be big. But please do the analyses again to see if you
get the same outliers or different ones? Let me know ASAP if there are any
other outliers so I check the brain volumes files immediately in another
software named FSLeyes to see if this are human errors and should be excluded.

Very important! These are the analyses I want to have done. Please
with tables.

1.
Are there gender difference in the volumes of the SCG in the 93
subjects done by me YTT (in the filename). So, every scanid in the single file except the following 002, 007, 012, 017,
028, 031, 035, 043, 044, 047, 055, 059, 065, 092 if I’m correct.

2.
Are there gender differences in the volumes of the SCG in the 107
subjects done by me (YTT) and AAX).

The above analyses are quite simple and you
can find the data in the file named ‘single’. The following analyses are
conjunctions. You can find them in the datafiles named ‘conjunctions’. Please check again the same same assumptions
again described in the black colored text above.

3.
Is there a gender difference if we make an mean for every hemisphere
for every subject (Segvolume+Refvolume = outcome, outcome divided by 2 for
every participant/hemisphere). So, we can say we made the sample more
trustworthy by adding multiple raters in to the analysis. Afterwards please
analyse if there is an effect for gender.

I will try to explain what this file consists
of but I don’t know the exact details to be honest cause my supervisor made
this rapport for me:

1.
Segvolume: This is the
‘primary’ volume. It doesn’t really mather but if you want to know how this
variable existed. The parameter for Segvolume is very simple. If rater AAX did
the segmentation include this segmentation in Segvolume if AAX did not do the
segmentation, use the segmentation of YTT. The thought behind this is that AAX
(my supervisor) has more experience in parcellations of the SCG then me,
therefore here parcellation is probably of higher quality and if possible,
should always be included if hers is not available I am so to speak the second
preferable rater cause it’s my research.

2.
Refvolume: After the
segmented volumes were determined we need another segmentation of the SCG to
make conjunctions. For this variable the parameter is also simple. The code
first checked if there is a segmentation available from me as a rater (YTT) if
this was not available a third rater (also a student as me) came in to play
with the initials VHX. So when Segvolume was done by AAX the chosen Refvolume
was done by YTT. But if YTT already was chosen for the afore mentioned
Segvolume different rater should be choosed and this was a third rater (VHX)
his segmentations were used as Refvolume to make conjuctions with. This was
done to calculate the following variables.

3.
Volume
difference. I really don’t understand this variable and don’t know which
analyse she used herefore. For example: Sub-002 has the following Segvolumes
and Refvolumes.

SegVolume,sub-002_ses-1_acq-wb_mod-t1map_mask-scg_hem-r_rat-aax_calc-bin_out.nii.gz,-,-,165.9548

RefVolume,-,sub-002_ses-1_acq-wb_mod-t1map_mask-scg_hem-r_rat-vhx_calc-bin_out.nii.gz,-,181.77373

=

Volume_difference,sub-002_ses-1_acq-wb_mod-t1map_mask-scg_hem-r_rat-aax_calc-bin_out.nii.gz,sub-002_ses-1_acq-wb_mod-t1map_mask-scg_hem-r_rat-vhx_calc-bin_out.nii.gz,-,0.087025344

I don’t know which analyse type she used here to come to the
following volume difference. Maybe for now we should ignore this part unless
you seem to know which analyse type she used? Also I don’t know how this should
give any information about my key questions?

4.
Dice overlap. This one is
very important! This analyse tells how good inter-rater agreement is. How higher the overlap how better the
segmentations of AAX VS. YTT or YTT VS. VHX how better the inter-rater
agreement. If you want more info about the dice score check:

https://en.wikipedia.org/wiki/S%C3%B8rensen%E2%80%93Dice_coefficient

It’s important that you analyse the Mean and
SD of this dice score.

5.
Average surface
distance

What I found about this:

What I understand of this is that the average surface distance is
the Mean Error between de distance of the contours of the mask (segmentation of
the SCG) and the voxels. So how lower the value of this error how better the
inter-rater agreement. Please also do an analyse on the mean and sd of this and find an useful article of this and cite
this while telling something about the mean of this error.

6.Possible outcomes could also be explained by another factor,
namely age. Please also check this for me.

7. Last but not least! Please write everything down that you do. So
which tests you did and which participants you excluded or not.

Get Higher Grades Now

Tutors Online