TY - JOUR T1 - Simultaneous use of cardiac PET perfusion and viability measurements to predict cardiac electrophysiological tissue properties JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 225 LP - 225 VL - 59 IS - supplement 1 AU - Benjamin Gutierrez AU - Timm Dickfeld AU - Vasken Dilsizian AU - Mark Smith Y1 - 2018/05/01 UR - http://jnm.snmjournals.org/content/59/supplement_1/225.abstract N2 - 225Objectives: Cardiac PET perfusion (Rb-82) and viability (F-18 fluorodeoxyglucose, FDG) measurements have previously been used independently to predict electrophysiological (EP) tissue categories to potentially aid electrophysiologists prior to radiofrequency (RF) ablation procedures to treat left ventricular (LV) tachycardia. The goal of this work was to investigate whether using perfusion and viability measurements together would improve these predictions. Methods: The population in this retrospective study was 18 subjects (18 male, mean age 68±10 yr) with ischemic heart disease who underwent PET Rb-82 perfusion and F-18 FDG viability studies prior to a substrate-guided RF ablation procedure using electroanatomic voltage mapping of the LV endocardial surface. Cardiac PET images and discrete EP datapoints were registered and displayed in polar map format. The PET perfusion values were normalized to the peak segmental value in the AHA 17 segment model and the FDG viability values were normalized to the FDG value in the peak perfusion segment. Dataset triples (EP voltage, Rb perfusion, FDG viability) were formed for each EP datapoint from the polar map display. Standard EP voltage-derived classifications of scar (0-0.5 mV), border zone (0.5-1.5 mV), normal (> 1.5 mV), abnormal (< 1.5 mV) were employed. The Rb and FDG data were used simultaneously in two different prediction methods. In Method A linear combinations of the Rb and FDG values were formed with different weights. This is equivalent to rotation of (Rb, FDG) values in 2-D space by different angles followed by projection onto the x-axis. In Method B (Rb, FDG) values were projected onto a family of non-linear curves in (Rb, FDG) space. In method A the x-value after rotation was the decision variable for receiver operating characteristic (ROC) analysis and in method B the projected x-value along the curve was used. ROC curves were generated for prediction of scar (0-0.5 mV) and abnormal (0-1.5 mV) EP tissue categories. The areas under the curve (AUC) of the ROC curves from the simultaneous use of Rb and FDG were compared to those using Rb and FDG alone. Results: There were a total of 5586 dataset triples. For prediction of scar tissue the AUC for FDG alone was 0.704, for Rb alone it was 0.721, for a linear combination of FDG and Rb the maximum was 0.732 at a rotation angle of 30 degrees, and for projection onto non-linear curves the maximum was 0.736 for a curve above the Rb-FDG line of identity. For prediction of abnormal tissue the AUC for FDG alone was 0.710, for Rb alone it was 0.738, for a linear combination of FDG and Rb the maximum was 0.746 at a rotation angle of 25 degrees, and for projection onto non-linear curves the maximum was 0.747 for a curve above the Rb-FDG line of identity. In all cases the AUC value with simultaneous use of Rb and FDG data was slightly greater than that from the use of FDG or Rb alone and the difference was statistically significant (p < 0.05). The difference in AUC between the maximum values for the rotated data and projections onto non-linear curves was not significant (p = NS). Conclusion: The simultaneous use of PET Rb-82 perfusion and FDG viability data to predict EP-derived cardiac tissue category results in a small but statistically significant improvement compared with the use of Rb-82 or FDG data alone. Multitracer cardiac tissue characterization with PET has the potential to aid substrate assessment prior to RF ablation to treat left ventricular tachycardia. ER -