Advanced indices in osteoporosis research: Measuring femoral cortical thickness with wholebody CT

1
Advanced indices in osteoporosis research
Measuring femoral cortical thickness with
whole-body CT

• Sven Prevrhal, PhD
• Musculoskeletal Quantitative Research Group,
• Department of Radiology, University of California
• San Francisco

2
Introduction

• Our goal is not the measurement of BMD but the
  assessment of fracture risk
• Hip Fractures are a severe and common consequence
  of osteoporosis
• DXA Hip BMD is a very good predictor of hip
  fracture
• Odds ratio per SD 2.6 (Cummings 1993)
• The relative amount of cortical bone in a hip DXA
  scan is very high

3
Indicator Finite Element Modeling
Healthy Proximal Femur
Osteoporotic Proximal Femur
µstrain
1500
0
-1500
4
How to assess cortical bone at the hip in vivo?

• CT
• Three-dimensional
• Calibration to of CT values to BMD possible
• Highly standardized
• -
• Limited spatial resolution
• Concern of radiation exposure

5
Study goals

• Can cortical thickness be measured with CT?
• What are the dose restrictions?
• Do we need a high resolution (high dose)
  protocol?

6
Materials and Methods

• Can cortical thickness be measured with CT?

20 excised femoral specimens age 79 /- 12
years no known bone disease
5 specimens
Whole body CT scans Siemens Single-Slice Spiral
CT 2 protocols High-resolution (1/1/1) Clinical
Resolution (5/5/1)
Micro-CT scans SCANCO ??CT 80 1024 x 1024 x
1200 Nominal orthotropic resolution 75 ?m
7
CT measurements
120 kVp 200 mAs 1 mm Slice Thickness Pitch
1 Contiguous Reconstruction 1 mm In Air Neck
axis aligned
120 kVp 240 mAs 5 mm Slice Thickness Pitch
1 Overlapping Reconstruction 1 mm In
Water Anatomically aligned
8
?CT measurements
9
Image ProcessingSurface Extraction

• Binary Thresholding
• Exact threshold not critical!
• 3D Morphology
• Closing of femur
• opening of background
• Filling of femur

25 50 75
100 125 150
10
Image ProcessingAnatomical Coordinates

• ACS positioned manually

11
Image Processing Cutting

• Cutting the femoral neck with planes
  perpendicular to the femoral neck axis
• 3 planes, 5 mm apart

12
Image Processing Normals

• Calculate surface normals
• (every 22.5 degrees)

13
Image Processing Probing

• Probe the CT data along the surface normals
• Calculate cortical thickness by 50 relative
  thresholds

14
Why relative thresholds?
High res
Low res
Ct.Th (mm) Threshold 50 REL
1.8 1.3 1.4
0.9 2.0 1.55
2.0 1.61
15
In-plane partial voluming
Prevrhal et al, Phys Med Biol 1999. 44(3)
16
Across-plane partial voluming
mm
1 2 3 5 8 10
mm slice thickness
?
Angle ?
Prevrhal et al., Med Phys, 2003. 30(1)
17
Results
Cortical Thickness in mm
18
Results - ??CT vs High-res CT
19
Results - ??CT vs Low-res
20
Dose considerations
WinDose
1 yr background2.5 mSv
High-res protocol 2.5 mSv for men, 0.5 for
women Low-res protocol 0.5 mSv for men, 0.1 for
women
Kalender et al, Eur Radiol, 1999. 9(3)
21
Conclusions

• High-resolution protocol clearly needed
• Radiation exposure very high
• Limitation to smaller scan range may be necessary
• Cortical thickness measurements limited by slice
  thickness and in-plane spatial resolution