-
Notifications
You must be signed in to change notification settings - Fork 0
RMINC
To use RMINC, you need a session where MincTools is activated.
I recommend using RMINC in the command prompt rather than in RStudio. If it's not already installed on your computer, make sure R is installed first, then follow the instructions on the RMINC website.
More complete guides for RMINC are available:
- Voxel-wise statistics
- RMINC Parallelism
- 2D visualization
- 3D visualization
- Hierarchical visualization
First, open a command prompt, source MincTools and RMINC (if using the lab's computer) launch R, and load the RMINC library:
source /opt/minc/1.9.17/minc-toolkit-config.sh
source /data/data05/vfonov/conda/etc/profile.d/conda.sh
conda activate RMINC
R
library(RMINC)
RMINC works by using a dataframe that includes paths to 3D .mnc image files (such as the MCSA datasheet). For example:
| Dx | Age | Group | Sexe | MOCA | ID | FEOBV_blur6 |
|---|---|---|---|---|---|---|
| 1 | 70 | Pre | 0 | 27 | MTL0101 | /home/minc/Desktop/ScansAll/AllScans/MTL0101-FEOBV_blur6.mnc |
| 1 | 69 | Pre | 1 | 28 | MTL0128 | /home/minc/Desktop/ScansAll/AllScans/MTL0128-FEOBV_blur6.mnc |
| 1 | 67 | Pre | 0 | 29 | MTL0308 | /home/minc/Desktop/ScansAll/AllScans/MTL0308-FEOBV_blur6.mnc |
| 1 | 62 | ctl | 0 | 27 | MTL0377 | /home/minc/Desktop/ScansAll/AllScans/MTL0377-FEOBV_blur6.mnc |
| 0 | 62 | ctl | 0 | 30 | MTL0304 | /home/minc/Desktop/ScansAll/AllScans/MTL0304-FEOBV_blur6.mnc |
| 0 | 60 | ctl | 1 | 28 | MTL0381 | /home/minc/Desktop/ScansAll/AllScans/MTL0381-FEOBV_blur6.mnc |
| 0 | 66 | ctl | 0 | 28 | MTL0442 | /home/minc/Desktop/ScansAll/AllScans/MTL0442-FEOBV_blur6.mnc |
| 0 | 68 | ctl | 0 | 29 | MTL0608 | /home/minc/Desktop/ScansAll/AllScans/MTL0608-FEOBV_blur6.mnc |
You may also use a mask to specifically perform your analyses within that region to make the multiple comparison correction less stringent.
Here is how the script for t-tests between preclinical (pre) and controls (ctl) looks like:
mask = "/home/minc/Desktop/PrevST_mask_ICV.mnc"
gf1 <- read.csv("/home/minc/Desktop/ScansAll/RMINC64.csv")
gf1$Group <- relevel(gf1$Group, ref="ctl")
#Setting the control group as the reference
vs1 <- mincLm(FEOBV_blur6 ~ Group + Age, mask=mask, gf1)
#Your model. The 3D image has to be the dependent variable. Here, we corrected for age.
vs1[is.na(vs1)] <- 0
#Setting the "0" as "missing"
summary(vs1)
#To get a summary of your linear model
mincFDR(vs1, mask=mask)
#perform FDR correction
mincWriteVolume(vs1,"/home/minc/Desktop/ScansAll/Results_24-07-17/t-test_6ctlvs3pre_agecor.mnc","tvalue-GroupPre")
#Creating a t-map - a 3D .mnc file with the t-values
The resulting file can then be used with MincTools.