Effects of Motion and Tissue Weighting on PET Myocardial Blood Flow Estimates Benjamin C. Lee 1, Jonathan B. Moody 1, Venkatesh L. Murthy 2, James R. Corbett 2, Edward P. Ficaro 2 1 INVIA Medical Imaging Solutions, Ann Arbor, MI, 2 University of Michigan Health System, Ann Arbor, MI Disclosures: BC Lee, J Moody, Employee: INVIA; VL Murthy, Stock Shareholder: GE, Grant: INVIA; JR Corbett, EP Ficaro, Owners: INVIA.
Introduction Background Relative PET perfusion imaging provides suboptimal clinical information of diffuse or multistenosis coronary artery disease (CAD) Dynamic 82 Rb PET measurements of myocardial blood flow (MBF) and coronary flow reserve (CFR) may improve those prognostic and diagnostic yields Respiratory motion may have an effect on the estimates of MBF and CFR
Introduction Background (cont d) Dataset with motion with blood pool spillover into myocardium Severe Motion Uncorrected Manual Motion Corrected
Introduction Background (cont d) Time activity curves (TAC) affected by motion Severe Motion TAC Manual Motion Corrected TAC 4
Activity (kbq/ml) Introduction Background (cont d) Weighting tissue data may reduce motion effects LV Blood LAD Tissue Tissue Weight 200 150 100 50 0 0 60 120 180 240 300 360 Time (sec)
Introduction Objectives Identify prevalence of respiratory motion in dynamic PET myocardial perfusion images Study respiratory motion effects on MBF and CFR estimates Manual motion correction is labor intensive therefore investigate effect of preferentially increasing weight in tissue-phase frames to offset respiratory motion blood-phase frames
Methods Data Acquisition 53 sequential patients referred for dynamic stress/rest 82 Rb PET imaging Acquired in 3-D list mode on Siemens Biograph mct 40 30 serial dynamic frames over 6.7 minutes 16 x 5sec, 6 x 10sec, 3 x 20sec, 4 x 30sec, 1 x 80sec Reconstructed using 3D-OSEM with PSF Attenuation corrected, 3 iterations with 24 subsets, no post-filtering 7
Methods Motion Analysis Visually assessed motion by maximum shift btwn. Blood pool in first 2 min. (first 20 frames) LV surface generated from post 2 min. as reference Categorized respiratory motion as: Minimal motion (sub-pixel shift) Moderate motion (1/2 wall thickness shift) Severe motion (full wall thickness shift) Datasets with moderate or severe respiratory motion were manually corrected for comparison 8
Methods Kinetic Analysis ROI method sampled left ventricular (LV) blood pool and tissue time activity curves (TAC) Kinetic fitting using Corridor4DM software (INVIA) 1-tissue compartment with Lortie flow model Weighting models No weighting Weighting using frame duration, decay correction factor, and activity concentration Evaluated mean MBF, CFR, and relative changes 9
Results Prevalence of Motion Motion was observed in 85% of the datasets with majority 55% of those with bloodpool in region Motion by Severity Motion* by Region Severe Moderate Minimal LAD LCX 100% 75% 50% 25% 17% 13% 15% 40% 53% 46% 43% 34% 39% 85% 100% 75% 50% 25% 14% 18% 68% 22% 18% 36% 42% 27% 55% 0% Stress Rest All 0% Stress Rest All *moderate and severe only
Results Change Bias, LOA MBF and CFR Effects from Motion Compared to motioncorrected, mean MBF and CFR changed < 5%, except for stress MBF with severe motion of 6% Mean stress MBF with severe motion had mean bias of 0.1* and 95% limits of agreement (LOA)** of 0.7* 10% 5% 0% 1.0 0.0 Moderate Motion Severe Motion * p < 0.05 * * * Stress MBF Rest MBF CFR Stress MBF Rest MBF CFR *(ml/min/g), **95% LOA=1.96 Stdev -1.0
Results Regional Change Regional Bias, LOA Regional Stress MBF Effects from Motion Mean stress MBF changes were less than 6% except for with severe motion of 20% with severe motion had mean stress MBF bias of 0.5* and 95% limits of agreement of 1.6* *(ml/min/g) 25% 20% 15% 10% 5% 0% 3.0 2.0 1.0 0.0-1.0-2.0 Moderate Motion * p < 0.05 * LAD Stress MBF LAD Stress MBF Severe Motion * LCX Stress MBF LCX Stress MBF * Stress MBF Stress MBF
Results Regional Change Regional Bias, LOA Regional CFR Effects from Motion Mean CFR changes were less than 4% except for with severe motion of 18% with severe motion had mean CFR bias of 0.4 and 95% limits of agreement of 1.9 20% 15% 10% 5% 0% -5% 3.0 2.0 1.0 0.0-1.0-2.0 Moderate Motion Severe Motion * p < 0.05 LAD CFR LCX CFR CFR LAD CFR LCX CFR CFR * *
Results Stress MBF Change Stress MBF Bias, LOA Tissue Weighting Effects on Stress MBF Tissue weighting (nonmotion corrected) did not reduce mean stress MBF bias Especially for with severe motion, change was only -1.4% point 95% limits of agreement did not decrease significantly 25% 20% 15% 10% 5% 0% 3.0 2.0 1.0 0.0-1.0-2.0 * p < 0.05 No Weight Weight
Results CFR Change CFR Bias, LOA Tissue Weighting Effects on CFR Tissue weighting (nonmotion corrected) did not reduce mean CFR bias with severe motion increased with change of just 1.4% point 25% 20% 15% 10% 5% 0% 3.0 * p < 0.05 No Weight Weight 95% limits of agreement did not decrease significantly 2.0 1.0 0.0-1.0-2.0
Limitations Manual Motion Correction Possible miscategorization of severe, moderate, and minimal respiratory motion No Tissue Phase (>2min) Motion Correction Reference LV surfaces generated from tissue phase Longer frame durations are motion averaged Lower count statistics in later frames
Conclusions Respiratory motion is common (up to 85%) in the blood pool phase due to fast kinetics of the tracer and the short time sampling. Increases in MBF and CFR estimates were observed in studies with moderate and severe respiratory motion, and substantially in the. Tissue weighting did not have desired result in greatly reducing respiratory motion induced bias. Motion correction is necessary to reduce bias in MBF and CFR.
Appendix
Activity (kbq/ml) Appendix: Frame Duration Weighting Weighting tissue data with frame duration only may reduce respiratory motion effects further LV Blood LAD Tissue Tissue Weight Duration Weight 200 150 100 50 0 0 60 120 180 240 300 360 Time (sec)
Stress MBF Change Appendix: Frame Duration Weighting Tissue Weighting Effects on Stress MBF Compared to no weighting, stress MBF with frame duration only weighting changed -2% points, except for with severe motion of -0.8% point No Weight Weight Duration Weight 25% 20% * p < 0.05 * 15% 10% 5% * 0% -5% (Minimal) (Minimal)
Appendix: Frame Duration Weighting CFR Change Tissue Weighting Effects on CFR Compared to no weighting, CFR with frame duration only weighting changed 1% point, except for with moderate and severe motion of 3% and 5% points respect. 25% 20% 15% * p < 0.05 No Weight Weight Duration Weight * 10% 5% * * 0% -5% (Minimal) (Minimal)