MYOCARDIAL BLOOD FLOW WITH [15O]H2O PET: ERROR ESTIMATES

1
MYOCARDIAL BLOOD FLOW WITH 15OH2O PETERROR
ESTIMATES
Turku PET Centre Vesa Oikonenhttp//users.utu.fi/
vesoik/
TPC modelling workshop 2005-09-14
2
Model parameters

• MBFmyocardial blood flow in perfusable tissue
  volume
• aperfusable tissue fraction, PTF
• VAmyocardial blood volume fraction and
  spill-over from LV cavity

3
MBF model is fitted

1. to ROI curves drawn on LV cavity (lv) and on
   entire myocardial tissue surrounding the cavity
   (whole) to estimate virtual arterial blood
   concentration, CA(t)
2. To smaller myocardial ROIs using CA(t) as model
   input.

4
Results of MBF model fit

• fitmbf 2.1.3 (c) 1999-2005 by Turku PET Centre
• Date 2005-09-13
• Study s05851dy
• Data file S05851\s05851.dft
• Reference region lv All
• Data range 0 - 360 sec (N20)
• Beta 0.85
• Data was not weighted.
• Region flow alpha Va
  SS AIC
• lat . All 0.8120 0.6564
  0.2671 9.5215e001 3.8708e001
• ant . All 0.8613 0.6391
  0.2036 1.2060e002 4.3434e001
• sept . All 0.7151 0.5245
  0.4841 1.9348e002 5.2888e001
• whole . All 0.8353 0.6329
  0.2863 4.8313e001 2.5140e001

(Weighted) Sum-of-squaresdescribing
goodness-of-fit
Estimates of model parameters
Akaike information criteriafor comparison
betweenmodels
5
Error estimate for each model parameter

• 95 confidence limits and/or standard
  deviations
• With command-line options-CL and SD
• Bootstrap method (Efron and Tibshirani, 1993
  Turkheimer et al., 1998)
• Starting from fitmbf version 2.1.0,2005-07-15

6
Measured and fitted TACs
7
Results with error estimates

• fitmbf 2.1.3 (c) 1999-2005 by Turku PET Centre
• Date 2005-09-13
• Study s05851dy
• Data file S05851\s05851.dft
• Reference region lv All
• Data range 0 - 360 sec (N20)
• Beta 0.85
• Data was not weighted.
• Region flow alpha Va
  SS AIC
• lat . All 0.8120 0.6564
  0.2671 9.5215e001 3.8708e001
• SD . . 0.0456 0.0164
  0.0163 . .
• CL 95 Lower 0.7241 0.6228
  0.2364 . .
• CL 95 Upper 0.9012 0.6849
  0.2955 . .
• ant . All 0.8613 0.6391
  0.2036 1.2060e002 4.3434e001
• SD . . 0.0529 0.0193
  0.0191 . .
• CL 95 Lower 0.7701 0.6073
  0.1674 . .
• CL 95 Upper 0.9690 0.6768
  0.2376 . .
• sept . All 0.7151 0.5245
  0.4841 1.9348e002 5.2888e001

8
Larger errors with higher MBF

• BASAL study
• Region flow alpha Va
  SS AIC
• lat . All 0.7259 0.6577
  0.1498 6.3335e001 3.0554e001
• SD . . 0.0493 0.0176
  0.0176 . .
• ant . All 0.6751 0.5503
  0.2290 7.2341e001 3.3213e001
• SD . . 0.0596 0.0186
  0.0187 . .
• sept . All 0.5756 0.5703
  0.4679 3.4841e002 6.4653e001
• SD . . 0.1109 0.0401
  0.0370 . .
• whole . All 0.6823 0.5924
  0.2555 7.0570e001 3.2717e001
• SD . . 0.0627 0.0193
  0.0197 . .
• Stimulation
• Region flow alpha Va
  SS AIC
• lat . All 4.1351 0.7403
  0.0500 7.1618e001 3.3012e001
• SD . . 0.4198 0.0486
  0.0461 . .
• ant . All 2.8656 0.4734
  0.2891 1.3123e002 4.5125e001
• SD . . 0.8369 0.0922
  0.0906 . .
• sept . All 4.9285 0.7674
  0.2390 1.6528e002 4.9738e001
• SD . . 1.0812 0.1251
  0.1183 . .

9
with confidence limits

• fitmbf 2.1.3 (c) 1999-2005 by Turku PET Centre
• Date 2005-09-13
• Study s06003dy
• Data file S06002\s06003.dft
• Reference region lv All
• Data range 0 - 360 sec (N20)
• Beta 0.85
• Data was not weighted.
• Region flow alpha Va
  SS AIC
• lat . All 4.1351 0.7403
  0.0500 7.1618e001 3.3012e001
• CL 95 Lower 3.3282 0.6074
  0.0228 . .
• CL 95 Upper 4.9667 0.7831
  0.1776 . .
• ant . All 2.8656 0.4734
  0.2891 1.3123e002 4.5125e001
• CL 95 Lower 1.5818 0.3160
  0.0794 . .
• CL 95 Upper 4.7635 0.6941
  0.4426 . .
• sept . All 4.9285 0.7674
  0.2390 1.6528e002 4.9738e001
• CL 95 Lower 2.9679 0.5418
  0.0292 . .
• CL 95 Upper 6.9406 0.9908
  0.4528 . .

10
Reduce SD by constraining a

• a during stimulation study is constrained to the
  estimate from basal conditions
• With command-line options-fAlpha and
  -fbasal.res,when using fitmbf 2.1.0 and later
  versions
• Also VA can be constrained with option-fVa.

11
after constraints for a

• fitmbf 2.1.3 (c) 1999-2005 by Turku PET Centre
• Date 2005-09-13
• Study s06003dy
• Data file S06002\s06003.dft
• Reference file results\s06002mbf1.res
• Reference region lv All
• Data range 0 - 360 sec (N20)
• Beta 0.85
• Data was not weighted.
• Region flow alpha Va
  SS AIC
• lat . All 3.5534 0.6577
  0.1283 7.6282e001 3.1480e001
• CL 95 Lower 2.8580 0.6577
  0.1096 . .
• CL 95 Upper 4.3003 0.6577
  0.1494 . .
• ant . All 3.4219 0.5503
  0.2140 1.3465e002 4.2845e001
• CL 95 Lower 2.5938 0.5503
  0.1865 . .
• CL 95 Upper 4.6877 0.5503
  0.2379 . .
• sept . All 3.2420 0.5703
  0.4294 1.7751e002 4.8372e001
• CL 95 Lower 2.3396 0.5703
  0.4071 . .

12
constraints for a and VA

• fitmbf 2.1.3 (c) 1999-2005 by Turku PET Centre
• Date 2005-09-13
• Study s06003dy
• Data file ..\S06002\s06003.dft
• Reference file s06002mbf1.res
• Reference region lv All
• Data range 0 - 360 sec (N20)
• Beta 0.85
• Data was not weighted.
• Region flow alpha Va
  SS AIC
• lat . All 3.0499 0.6577
  0.1498 9.4997e001 3.3384e001
• CL 95 Lower 2.6776 0.6577
  0.1498 . .
• CL 95 Upper 3.4477 0.6577
  0.1498 . .
• ant . All 2.9320 0.5503
  0.2290 1.4408e002 4.1715e001
• CL 95 Lower 2.3330 0.5503
  0.2290 . .
• CL 95 Upper 3.6193 0.5503
  0.2290 . .
• sept . All 2.3688 0.5703
  0.4679 2.2999e002 5.1068e001
• CL 95 Lower 1.9069 0.5703
  0.4679 . .

13
Constraints or not?

• Calculate results without and with constraints
  (a, a and VA, VA)
• Select the method that gives constantly smaller
  AIC

14
References

1. Araujo LI, Lammertsma AA, Rhodes CG, McFalls EO,
   Iida H, Rechavia E, Galassi A, De Silva R, Jones
   T, Maseri A. Noninvasive quantification of
   regional myocardial blood flow in coronary artery
   disease with oxygen-15-labeled carbon dioxide
   inhalation and positron emission tomography.
   Circulation 1991 83 875-885.
2. Efron B, Tibshirani RJ. An introduction to the
   bootstrap. (1993) Chapman Hall, New York.
3. Iida H, Rhodes CG, de Silva R, Yamamoto Y, Araujo
   LI, Maseri A, Jones T. Myocardial tissue fraction
   correction for partial volume effects and
   measure of tissue viability. J. Nucl. Med. 1991
   322169-2175.
4. Iida H, Rhodes CG, de Silva R, Araujo LI,
   Bloomfield P, Lammertsma AA, Jones T. Use of the
   left ventricular time-activity curve as a
   noninvasive input function in dynamic
   oxygen-15-water positron emission tomography. J.
   Nucl. Med. 1992 331669-1677.
5. Turkheimer F, Sokoloff L, Bertoldo A, Lucignani
   G, Reivich M, Jaggi JL, Schmidt K. Estimation of
   component and parameter distributions in spectral
   analysis. J. Cereb. Blood Flow Metabol. 1998 18
   1211-1222.
6. Watabe H, Jino H, Kawachi N, Teramoto N, Hayashi
   T, Ohta Y, Iida H. Parametric imaging of
   myocardial blood flow with 15O-water and PET
   using the basis function method. J. Nucl. Med.
   2005 46 1219-1224.

15
More information

• Detailed report TPCMOD0005http//www.turkupetcent
  re.net/reports/
• Analysis instructionshttp//www.turkupetcentre.ne
  t/analysis/doc/tracer/mbf_h2o.html
• Bootstrap method in estimation of variance of
  model parametershttp//www.turkupetcentre.net/mod
  elling/workshop1/bootstrap.ppt