Global View of All Collected Data

1
Global View of All Collected Data

• Phantom Data from all sites
• Duke, Echosens, Langevin, Mayo, Philips, Siemens,
  SSI, UCSD, U Wisconsin, U Michigan
• Each sites data is imported and cleaned and
  combined on one Excel spreadsheet
• Analyzed in JMP 5.1

2
Global View of All Collected Data

• Phantom Data from all sites
• Duke, Echosens, Langevin, Mayo, Philips, Siemens,
  SSI, UCSD, U Wisconsin, U Michigan
• Each Data Point is Mean of 10 measurements
• Red quantiles 10/90, 25/75, Median
• Includes Hard and Soft
• Philips phantom damaged
• Phantoms confounded w/ site

3
Soft Phantom Analysis 3/3
4
Soft Phantom Analysis 1/3

• r

Level Minimum 10 25 Median 75 90 Maximum
Echosens 0.87367 0.875554 0.879361 0.890544 0.930949 0.93447 0.934496
S2000 0.772 0.7948 0.814 0.835 0.868 0.882 0.9
SSI 0.89 0.9 0.93 0.96 1 1.028 1.19
Verisonics-Duke 0.86497 0.869225 0.87389 0.91451 0.93839 0.940552 0.94221
Verisonics-Mayo 0.8687 0.87562 0.8921 0.9017 0.9158 0.92604 0.9282
5
Soft Phantom Analysis 2/3
SSI

• Need to recheck Echosens data

6
Soft Phantom Analysis 2/3
Duke
Mayo
7
Soft Phantom Summary

• To Do
• Appraiser is not a factor in all analysis
• ANOVA is missing DoF, so concern over providing
  estimates
• Even after eliminating Echosens data, residual
  plots and model are suspicious
• Continue analysis within site and within system
  for further insights

Source Nparm DF Sum of Squares F Ratio Prob gt F
PHANTOM 7 5 0.07667410 13.2524 lt.0001 LostDFs
DEPTH (cm) 2 2 0.04339614 18.7515 lt.0001
APPRAISER PHANTOM 16 16 0.01498435 0.8093 0.6743
System 3 1 0.30120535 260.3023 lt.0001 LostDFs
8
Hard Phantom Analysis 1/3
9
Hard Phantom Analysis 2/3
Level Minimum 10 25 Median 75 90 Maximum
Echosens 2.028042 2.034633 2.065677 2.128192 2.168532 2.232099 2.28903
S2000 1.91 2.0157 2.11525 2.1605 2.21575 2.259 2.371
SSI 1.95 2.01 2.05 2.08 2.1 2.1 2.15
Verisonics-Duke 2.00657 2.012296 2.0203 2.08477 2.1516 2.155544 2.16001
Verisonics-Mayo 2.025 2.03512 2.0504 2.1318 2.1835 2.20006 2.2131
10
Hard Phantom Analysis 3/3
SSI
Siemens
11
Hard Phantom Analysis 3/3
Duke
Mayo
12
Summary

• Summary
• Appraiser is not a factor in all analysis
• Seem to be a depth dependency
• To Do
• Finish ANOVA Analysis
• Contribute to understanding of results and
  outliers Do analysis for each system
• Determine how to analyze Duke Phantom Measure