Tuesday 20 August 2013

comparisons between two MRV images using imagej programm


Image analysis using
Fast Fourier Transform
SNR
CNR
Image contrast
1-Fast Fourier Transform (FFT)
Theory (in progress).
Comparison between the two techniques using FFT in imageJ.
Figure 1: FFT in imageJ
1st case:
A B
Figure2: A. CE MRV, B. FFT.
A B
Figure 3: A.PC MRV,B.FFT.
Figures 2 & 3 above show that the two images contain components of all frequencies, but the magnitude gets smaller sharply for higher frequencies in CE MRV image, while it decreases gradually in PC MRV. Hence, low frequencies in CE MRV image contain more fine information than PC MRV image. The transform image also tells us that there is one dominating directions in each Fourier image (Y-direction for CE MRV, and X-direction for PC MRV) , one passing vertically and one horizontally through the centre. These originate from the regular patterns in the background of the original images.
2- Signal-to Noise Ratio (SNR)
Background (in progress)
SNR in imageJ
First, Choosing the same region of interest (ROI) in two images by click on analysis> Tools > synchronize windows as shown below in table1 and figure 4.
Table 1: Synchronize windows.
Figure 4: The way to choose the same ROI in imageJ.
2
Figure 5: ROI (sagittal venous sinus) in 1) CE MRV, and 2) PC MRV.
Table 2: Mean intensity (signal) and standard deviation (noise) for image 1 &2
From Table 1, mean intensity µ for sagittal venous sinus is obtained in each image.
Now, calculation of background standard deviation is needed too.
Figure6: 3 ROIs in CE MRV to calculate the Standard deviation (background noise).
Figure7: 3 ROIs in PC MRV to calculate the Standard deviation (background noise).
Background noise in CE MRV
Standard Error = 0.22
Standard Deviation (1), s = 2.021± 0.22
Standard Deviation (2), s = 1.311 ± 0.22
Standard Deviation (3), s = 1.435 ± 0.22
Background noise in PC MRV
Standard Error = 0.083
Standard Deviation (1), s =1.946 ± 0.083
Standard Deviation (2), s =2.167± 0.083
Standard Deviation (3), s = 2.216 ± 0.083
To calculate SNR
SNR = µ/s
µ = mean signal intensity for venous sinuses
s = standard deviation (noise background)
SNR in CE MRV image
SNR (1) = 137 / 2.021 = 67.788
SNR (2) = 137 / 1.311 = 104.5
SNR (3) = 137 / 1.435 = 95.47
Average SNR in CE MRV = 67.788+104.5+95.47 / 3 = 89.25 ± 11.0444
SNR in PC MRV image
SNR (1) = 85.75/ 1.946= 44.06
SNR (2) = 85.75 / 2.167= 39.57
SNR (3) = 85.75 / 2.216= 38.69
Average SNR in PC MRV = 44.06+39.57+38.69/ 3 = 40.77 ± 1.6629
SNR in CE MRV is higher
3- Contrast –to –Noise Ration (CNR)
Contrast to noise ratio (CNR) = differences in signal between ROI (Sinus and soft tissue / Noise of interest (soft tissue)
CNR in CE MRV
CNR (1 ) = 137- 63.417 / 2.021= 36.40
CNR ( 2) = 137 – 66.583 / 1.311= 53.71
CNR (3) = 137 – 67.33 – 1.435 = 48.55
Average CNR in CE MRV image = 46.22 ± 5.131
CNR in PC MRV
CNR (1) = 85.75 – 13.67 / 1.946 = 37.04
CNR (2) = 85.75 – 11.167 / 2.167 = 34.41
CNR (3) = 85.75 – 12 / 2.216 = 33.28
Average CNR in PC MRV image = 34.91 ± 1.1138
CNR in CE MRV is higher too
Image contrast
Image contrast = contrast_sinus – contrast_soft tissue) / contrast_soft tissue
CE MRV Image contrast
A = 137 – 63.417 / 63.417 = 1.160
B= 137 – 66.583 / 66.583 = 1.057
C = 137 – 67.33 / 67.33 = 1.034
Average contrast for CE MRV image = 1.083 ± 0.0387
PC MRV image contrast
A = 85.75 – 13.67 / 13.67 = 5.27
B= 85.75 – 11.167 / 11.167= 6.67
C= 85.75 – 12 /12 = 6.97
Average contrast for PC MRV image = 6.3 ± 0.5239
It is clearly seen that the image contrast for PC MRV is more because it has more background suppression which is considered as an advantage of PC MRV over CE MRV technique.

The previous results is only for one case (9 cases left)
There will be tables for the 10 cases together and the aim of this doc is to give you a general background about this chapter

No comments:

Post a Comment