2138. Effects of Iterative Reconstruction Algorithms on Coronary CTA Stenosis Measurement and FFR-CT Analysis
Authors * Denotes Presenting Author
  1. Fides Schwartz *; Duke University Health System
  2. Juan-Carlos Ramirez Giraldo; Siemens Healthineers
  3. Daniele Marin; Duke University Health System
  4. Lynne Hurwitz Koweek; Duke University Health System
To assess the impact of different reconstruction algorithms on luminal modeling of the coronary arteries and on computed tomography-derived fractional flow reserve (CT-FFR).

Materials and Methods:
Nine coronary cardiac computed tomographic angiography exams (CCTA) performed on a 3rd generation DS MDCT scanner (Siemens Somatom FORCE) containing a total of 50 unique coronary artery stenoses were reconstructed with filtered back projection (FBP) and two strength levels of iterative reconstruction (IR-2 and IR-4, ADMIRE, Siemens Healthineers), resulting in 27 CCTA renderings and analysis of 150 stenoses. Image quality was assessed for each dataset by calculating signal to noise ratio (SNR) and contrast to noise ratio (CNR). Centerlines were created for the three main epicardial vessels (left anterior descending LAD, right coronary artery RCA and circumflex artery CX). Area of stenosis was measured on cross sectional slices along the centerline and plaque features were recorded. CT-FFR was performed by Heartflow Inc. Descriptive statistics, two-tailed student’s t-tests and ANOVA analyses were applied (jmp Pro V15.0, SAS statistics).

The study population was 33% female (61.8 ± 10.8 years) with an average of six stenoses/patient (range: 0-9). CNR and SNR was significantly higher for IR-4 than for IR-2 and FBP (P<0.05); no significant difference between IR-2 and FBP (P=0.6705 and 0.2644, respectively). A majority of stenoses was = 50% in FBP (30/50) and IR-2 (31/50) but =50% for IR-4 (23/50). 23/50 and 18/50 stenoses on FBP changed by >10% in area for the IR-2 and IR-4 renderings, respectively. ANOVA analysis showed a significant difference for degree of stenosis between FBP and IR-2 (Prob>[t]=0.0236, Intercept 0.586) and IR-4 (Prob>[t]=0.0283, Intercept 0.514) reconstructions. There was no significant difference in area of stenosis measured on FBP compared to IR algorithms for solitary stenoses (IR-2: P=0.613, IR-4: P=0.953), tandem stenoses (IR-2: P=0.752, IR-4: P=0.132), proximal vessel stenoses (IR-2: P=0.682, IR-4: P=0.986), distal vessel stenoses (IR-2: P=0.725, IR-4: P=0.475), the right system (IR-2: P=0.950, IR-4: P=0.744) or the left system (IR-2: P=0.833, IR-4: P=0.413). A majority of stenoses had FFR values = 0.81. FFR values in both IR-2 and IR-4 were not significantly different from FBP for the 50 stenoses.

Difference in area of stenosis measured on CCTA occurs for iterative reconstruction compared to FBP reconstruction but FFR values were not significantly different between the 50 stenoses in three different reconstructions. This shows that the use of iterative reconstruction algorithms in clinical CCTA is valid for stenosis analysis using FFR-CT.