Upload
colin-sullivan
View
213
Download
0
Embed Size (px)
Citation preview
Processing Micro CT Bone Processing Micro CT Bone Density ImagesDensity Images
NASA SHARP Student: Paul NelsonNASA SHARP Student: Paul Nelson NASA Mentor: John DaPonte Ph. D.NASA Mentor: John DaPonte Ph. D.
Team Members: Michael Clark, Elizabeth Wood, Thomas Sadowski, Paul ThomasTeam Members: Michael Clark, Elizabeth Wood, Thomas Sadowski, Paul Thomas
Southern Connecticut State University (SCSU), Computer Science DepartmentSouthern Connecticut State University (SCSU), Computer Science DepartmentSuny Stonybook University Suny Stonybook University NASA SHARP Program NASA SHARP Program
Project OverviewProject Overview
6 sets of high resolution (10.5µm), 8 sets of low resolution (17.5µm) image stacks of approximately 86 micro CT mouse images of trabeculae before and after exposure to weightlessness
Iterative deconvolution, no pre-processing and Gaussian blur.
Iterative deconvolution was qualitatively better then Gaussian blur and no pre-processing
Provide quantitative verification of the qualitative findings through numerical evaluation of data obtained from a variety of programs
Original Image After a few programs including deconvolution
Current WorkCurrent Work
Areas of focus for this bone density study include the fractal dimension, entropy enhancement (EME), bone area, bone thickness, SMI , and BV/TV
A majority of these parameters have already been collected for high/low resolution images
Calculations BackgroundCalculations Background
Bone Volume/Tissue Volume (BV/TV)= Total Bone Volume Total Tissue Volume Structure Model Index (SMI)= 6x(S’xV/S2)
S’=Change in Surface Area Caused by Dilation, S=Object Surface Area Before Dilation, V=Initial Volume
Bone Area Pixel Ratio (BAPR)= #of Bone Pixels In Threshold Image Total # of Pixels
Trabecular Thickness Pixel Ratio (TTPR)= # of Bone Pixels In Threshold Image # of Bone Pixels In Skelonized
Image
High/Low Res. Comparison DataHigh/Low Res. Comparison Data
5868(BL1)6125(AL1)
5870(BL2)6128(AL2)
5864(BM1)6119(AM1)
5866(BM2)6124(AM2)
Deconv
No Prep
Gauss0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
AVG Bone Area Ratio
Case(Before/After,More/Less,Pair#)
Low Res AVG Bone Area
5868(BL1)6125(AL1)
5870(BL2)6128(AL2)
5864(BM1)6119(AM1)
5866(BM2)6124(AM2)
Deconv
No Prep
Gauss0
2
4
6
8
10
12
AVG Bone
Thickness Ratio
Case(Before/After,More/Less,Pair#)
Low Res AVG Bone Thickness
2779(B1)2916(A1)
2964(B2)3106(A2)
2969(B3)3115(A3)
Deconv
No Prep
Gauss0
1
2
3
4
5
6
7
8
9
10
AVG Bone
Thickness Ratio
Case(Before/After,Pair#)
High Res AVG Bone Thickness
2779(B1)2916(A1)
2964(B2)3106(A2)
2969(B3)3115(A3)
Deconv
No Prep
Gauss0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
AVG Bone Area Ratio
Case(Before/After,Pair#)
High Res AVG Bone Area
High/Low Res. ConclusionsHigh/Low Res. Conclusions The average bone thickness for the low resolution
images showed a slight thinning for deconvolution and greater thickening by the Gaussian
The deconvolved data was closer to the data with no pre-processing
For the high resolution images this feature is further shown because the Gaussian was thickened greatly and the deconvolved averages to about the same as the no pre-processing
The average bone area for the low resolution image showed trends similar to average bone thickness
This difference is not as large for the Gaussian as in the large for deconvolution in the high resolution images.
BV/TV Graphs and ConclusionsBV/TV Graphs and Conclusions
5864(BL1)6119(AL1)
5866(BL2)6124(AL2)
5868(BM1)6125(AM1)
5870(BM2)6128(AM2)
Deconv
Not Deconv
0.00
0.50
1.00
1.50
2.00
2.50
3.00
SM
I #
Case(Before/After,More/Less,Pair#)
SMI # Pre Processing Method Comparison
5864(BL1)6119(AL1)
5866(BL2)6124(AL2)
5868(BM1)6125(AM1)
5870(BM2)6128(AM2)
Deconv
Not Deconv
0.000
5.000
10.000
15.000
20.000
25.000
30.000
35.000
BV
/TV
(%
)
Case(Before/After,More/Less,Pair#)
BV/TV % Pre Processing Method Comparison
For BV/TV the bone loss is lower for those mice that experienced more bone loss and higher for those that experienced less bone loss
After suspension occurs, BV/TV always decreases The SMI appears inversely related to the BV/TV
according to these graphs
Future WorkFuture Work
To date, the fractal dimension values have yet to be compiled. Theses parameters will continue to be analyzed to identify any trends that might allow for the separation of more from less bone loss and before and after experiencing weightlessness.
A planned future parameter of study is connectivity analysis.
The EndThe End