Skip to main content

Table 2 The optimal combinational-factor predictors with AUC, sensitivity and specificity for three morphological indices using four machine learning methods

From: Predicting plaque vulnerability change using intravascular ultrasound + optical coherence tomography image-based fluid–structure interaction models and machine learning methods with patient follow-up data: a feasibility study

Index ΔLPI ΔCTI ΔMPVI
Method Predictor AUC (Spe, Sen) Predictor AUC (Spe, Sen) Predictor AUC (Spe, Sen)
RF LP
MeanCT
Critical WSS
Cap PWS
0.931 (0.971, 0.642) MinCT
Critical PWSn
Critical WSS
Cap WSS
0.826 (0.923, 0.555) MinCT
Plaque area
Critical PWS
0.847 (0.855, 0.583)
DA LP
Wall Thickness
Critical PWS
Cap PWSn
0.957 (0.935, 0.920) MinCT
MeanCT
Critical PWS
Cap PWS
Cap PWSn
0.836 (0.823, 0.676) MinCT
MeanCT
Critical PWSn
Cap PWSn
Cap PWS
0.812 (0.831, 0.511)
SVM LP
Critical WSS
Cap PWS
Cap PWSn
Cap WSS
0.963 (0.974, 0.777) MinCT
Lumen area
Plaque area
Critical PWSn
0.731 (0.926, 0.320) MeanCT
MinCT
Plaque area
Critical PWS
Critical PWSn
0.773 (0.862, 0.436)
EL LP
MeanCT
Lumen area
Critical WSS
Cap PWSn
0.861 (0.972, 0.607) MeanCT
MinCT
Cap WSS
0.781 (0.915, 0.464) MinCT
Plaque area
Critical PWSn
0.794 (0.870,0.508)
  1. Bold indicates that AUC value is the largest in this columns
  2. Sen sensitivity, Spe specificity