Menchón-Lara et al. [120]
|
Multi-layer perceptron binary classifier
|
N=60 (30 subjects)
|
Bland-Altman plots, mean absolute distance, poly-line distance and
centre-line distance.
|
Error in:
LI =37.03±18.57 μm, MA= 34.52±10.29 μm, IMT=37.63±25.18 μm
|
Menchón-Lara et al. [2]
|
ELM-based autoencoder
|
N=67
|
Mean absolute difference, correlation coefficient
|
Error in
IMT=5.79±34.42 μm
|
Biswas et al. [3]
|
Encoder and decoder based on DL.
|
N=407 (204)
|
Precision-of-merit,
Figure-of-Merit analysis, inter-operator variability, and ROC
analysis.
|
LD measurement accuracy=99.89%
Jaccard index and Dice similarity for lumen segmentation on 407 images
was 0.94 and 0.97
|
Tajbakhsh et al. [135].
|
DL-based frame selection, ROI localisation, IMT assessment
|
N=48 videos (12)
|
Bland–Altman plot
Localisation error
FROC curve
|
Error in LI = 2.68 µm,
MA = 3.69 µm,
IMT =2 3.4±17.3 µm
Localisation error = 0.19 mm (carotid bulb region) and 0.35 mm
|
Lekadir et al. [136] |
CNN-based plaque characterization |
N=90000
image patches (56) |
Correlation=0.90 with expert readings. |
78.5%
improvement over SVM method. Real-time processing. |