Analysis of bone structure in brittle

bone disease

John JamesonALS User Meeting – Tomography Workshop

10/09/2012

Osteogenesis imperfecta (OI)

• Genetic fragility disordero 25-50,000 in US

• Mutations cause impaired collagen synthesis and assemblyo Limb deformity, reduced stature, respiratory problems,

connective tissue laxity…

• 8 known types*o Mild, Type I: 50% of all caseso Moderate-Severe, Types II-VIIIo High degree of variability

Byers 1994, Chiasson 2004

Bone hierarchical structure

CollagenMineralized triple helices

(<1 µm)

FibrilsStaggered bundles

of molecules(~1 µm)

Fiber arrays“Plies” at different

orientations(~5-20 µm)

OsteonsLamellae + vasculature

(100-300 µm)

Solid materialCortical bone(>3000 µm)

Trabeculae“Plates and rods”

(100-200 µm)

Solid materialTrabecular bone

(>3000 µm)

Whole boneFemur

BONE FRAGILITY -How does bone structure affect mechanical

performance?-How do changes to OI bone cause fragility?

Micro

Nano

MOUSE HUMAN200 µm

50 µm

25 µm

5 µm

<1 µm

10 µm

Trabecular (spongy): -decreased quantity -more isotropic-poor organization

Cortical (compact): -higher porosity-thinner canals

Cortical (compact): -higher porosity-larger pores w/higher

connectivity

Additional considerations

• Tomographyo Preventing tissue damage

• SAXS/WAXDo Collagen spacing, mineral thickness

• In situ mechanical testing?o Similar to hot cell, but much simpler experimento Notch and apply different loads

• Can do this in ESEM now, but only get 2D info => 3D with tomo after

Data collection pipeline• Preparing for beamtime• Beamtime

o Average and peak data rates• 10-15 scans/beamtime => 11 GB/scan => 100-200 GB/session• Only person doing data collection, analysis

o Real-time feedback would be helpful in assuring quality of images• Working on noise study to provide feedback to users on proper settings to use

• After beamtimeo Reconstruction (1 hr/dataset)o Filtering and segmentation (1-2 hrs first dataset)

• Bilateral filtering => Usually have to downsample images, convert to 8-bit so it doesn’t take forever

• Binary morphological operations: Opening/closingo Data processing (0.5 day/dataset)

• Sphere-fitting algorithms, skeletonization, surface meshingo Visualizations

• Fiji and/or Avizo => Requires downsampling so it doesn’t crash all the time

Quantifying “resolution”• Resolution is difficult to measure

o Noiseo Instrument alignmento Imaging parameters/setup:

• Lens, #angles, counts (exp. time/scintillator), dithering…

• Fourier ring correlation (FRC)o Cardone et al (2005), J Struct Biol 151:p.117-129o Shows contribution of a single image

• Can give you an idea about the quality of your data sets and how they change over time

• Example: #Angles513 => 6.7 µm1025 => 3.5 µm2049 => 2.1 µm

Tradeoff b/w resolution and data set size

The RERC Grant Objective0-2 years: -µCT -Mechanical testing-Gait analysis

2-5 years: -SIMM-FE mesh refinement

5+ years: -Patient-specific fracture risk-Guided surgery-Assessment of existing/new

therapies

Mechanical & Imaging Data: Closing the Gap

Total crack length

(c)

Residual indent

α tip geometric factorE modulusH hardnessP applied tip loadc total crack length

Koester et al 2008

Crack

Haversion

Canals

Notched mechanical testingMicrohardness indentation

Tomography-3D crack visualization

-Crack volume

NERSC areas of interest

• Reconstruction:o Feedback on image quality/FRC

• Filtering:o Reduce amount of downsampling

• Software:o Fiji –macro capabilityo Avizo – advanced visualization of

crack propagationo Matlab – FRC for noise/resolution

calculationso Igor Pro – SAXS/WAXD