Skip to main content

Table 5 Multivariate analysis for the influencing factors associated with CNN’s classification on calcified plaques with motion artifacts

From: Classification of moving coronary calcified plaques based on motion artifacts using convolutional neural networks: a robotic simulating study on influential factors

  Inception v3 ResNet101 DenseNet201
Coefficient p value Coefficient p value Coefficient p value
Density 0.033 < 0.001 0.024 < 0.001 0.319 < 0.001
CT vendor 0.012 0.147 − 0.025 0.091 − 0.038 0.102
Velocity 0.027 < 0.001 0.017 < 0.001 0.015 < 0.001
Dose − 0.009 0.601 − 0.011 0.159 0.002 0.779
Reconstruction 0.009 0.126 0.010 0.112 0.012 0.099
  1. High, medium-1, medium-2, and low-density plaque were coded as 1–4, respectively, four CT systems (CT-A to CT-D) as 1–4; velocities from 0 to 60 mm/s coded as 0–6; dose level 40%, 80% and full dose coded as 1–3; recon method FBP, IR1 to IR3 coded as 1–4
  2. FBP filtered back projection, IR iterative reconstruction