Day 2 - Friday - 0830 AM - Nick Petrick
Day 2 - Friday - 0830 AM - Nick Petrick
Day 2 - Friday - 0830 AM - Nick Petrick
1
Division of Imaging and Applied Math, OSEL/CDRH/FDA
2
US Army Medical Research and Materials Command
Outline
• Clinical studies
• Coffee break experiment
• Repeat scans on same patient
N Petrick PCF Presentation 3
May 14, 2010
Quantify Sources of
Estimation Error
• Understand how specific parameters
influence bias/variance of measurements
• Development of standards for imaging protocols &
software tools
• Speed adoption of quantitative imaging in drug
trials and clinical practice
Data Collection
Gavrielides, SPIE 2010
5
FDA Phantom Experiment
• Goal
• To quantify sources of measurement error in nodule change
assessment and when possible, minimize their effects
• Method
• Well-controlled anthropomorphic phantom studies
• Measure effect sizes
• Identify influential imaging parameters
• Determine smallest measurable change
• Funding
• FDA: Critical Path
• NIH: NCI, NIBIB
N Petrick PCF Presentation 6
May 14, 2010
Factors Influencing Quantitative
Accuracy and Precision
• Scanner characteristics
• Acquisition/reconstruction parameters
• Nodule characteristics
• Analysis software
• Patient biology/genetics
• Clinician
N Petrick PCF Presentation 7
May 14, 2010
Anthropomorphic Thorax
Phantom
HU Range
[-709, +41]
5 mm
5 mm
-800 HU -630 HU
10 mm
5 mm
+100 HU -630 HU
• Philips 16:
• Siemens 64: 720
20 mm 10 mm sets
20 mm
• Total: 3360 sets
10 mm
• Siemens 64
• Collected
40 mm 5 mm 40 mm
5 mm
curtsey of
Washington
University
+100 HU -630 HU
• Siemens 64
• Collected
curtsey of
Washington
University
+100 HU -630 HU
40 mm
5 mm
40 mm
20 mm
5 mm
10 mm
8 mm
8 mm 20 mm
-10 HU -10 HU
10 mm 10 mm
5 mm 5 mm
5 mm 5 mm
8 mm
8 mm
8 mm
5 mm
+100 HU -630 HU
Standard High
Pitch
1.2 0.9
x10
Collimation 16x0.75 mm 16x1.5 mm
21
Public Access
(https://imaging.nci.nih.gov)
0.8 mm slices
16x0.75
100 mAs
• Acquisition
• 200 mAs
• 1.2 pitch
• 0.8 mm slices
• BF40 kernel
N Petrick PCF Presentation 29
May 14, 2010
Future Collection
• Nodule
• Mixed density
• 10mm, -10HU outer
• 5mm -630 HU core
• Acquisition
• 200 mAs
• 1.2 pitch
• 0.8 mm slices
• BF60 kernel
N Petrick PCF Presentation 30
May 14, 2010
Discussion
• >5000 scans completed
• Public access through National Biomedical
Imaging Archive
• https://imaging.nci.nih.gov
• Additional releases every 4-6 weeks
• Current focus on
• Completing solid nodule collection
• Heterogeneous nodules
Reference Standard
Kinnard, SPIE 2010
32
Study Purpose
• Investigate methods to estimate true
volume & shape of nodules
• How accurate is truthing process?
• Micro CT provides
• Very high resolution imaging
• Absolute volume
• Shape information
• Useful for understanding how software perform
with complicated shapes
• 26 synthetic nodules
• Shapes
• spherical, elliptical,
lobulated, spiculated
• Sizes (nominal diameter)
• 5, 8, 10, 20, 40 mm
• Densities
• -630, +100 HU
• Micro CT
• Scanco CT 40
• Imtek MicroCAT II
• Measurements courtesy
• NIST
• Washington University
40 mm lobulated nodule in MicroCAT II
• Chose (1,1,1)
• Highest
precision
• Estimates
• VWD Weight-density estimate
• VCT Micro CT estimate
• Blurred boundary
• Imaging & reconstruction process
• May lead to underestimates in volume
Latest Analysis
50
FDA Volume Estimator
• Purpose
• To develop a low bias/variance estimator
• Improves measurement of CT effects by
reducing software effect
20
• Legend
Normalized bias
10
0
• + -630HU -10
• О -10HU -20
5 8 10
• Δ +100HU Nodule Size (mm)
3mm, Attached, SAD
20
Normalized bias
10
• Acquisition 0
-20
• 1.2 pitch 5 8
Nodule Size (mm)
10
• BF60 kernel
N Petrick PCF Presentation 52
May 14, 2010
Non-attached Spheres
0.8mm, Non-attached, SAD
20
• Legend
Normalized bias
10
0
• + -630HU -10
• О -10HU -20
5 8 10
• Δ +100HU Nodule Size (mm)
3mm, Non-attached, SAD
20
Normalized bias
10
• Acquisition 0
-20
• 1.2 pitch 5 8
Nodule Size (mm)
10
• BF60 kernel
N Petrick PCF Presentation 53
May 14, 2010
Attached Elliptical
• Legend
• О -10HU
• Acquisition
• 200 mAs
• 1.2 pitch
• 0.8 mm slices
• BF60 kernel
• References standard
• Micro CT provide both volume and shape information
• Volume estimates consistent with weight-density for high
densities nodules and simple shapes
• Micro CT scans were obtained from both NIST and Center for Clinical
Imaging Research in the Mallinckrodt Institute of Radiology (MIR) at the
Washington University School of Medicine in St. Louis, MO. The
authors would like to thank Sheng Lin-Gibson from NIST and Richard
Laforest and Anne Stroncek from MIR for their assistance