MAE: Mean absolute error, MACE: Major adverse cardiac events
Bai et al. showed that Fully Convolution Network (FCN) matches expert
like performance in speed, scalability and accuracy over CV magnetic
resonant (CMR) image segmentation and clinical measurements [127].
In another study on fatty liver disease (FLD) tissue characterization,
Kuppili et al. [131] observed accuracy of 96.75% using
SymtosisTM, an extreme learning machine (ELM).
Further, Biswas et al. [130] developed a DL-based system and
exhibited 100% accuracy on a database of 63 samples. Meanwhile,
Hemalatha et al. [132] used DL for analysing different grades of
synovitis. The study observed an increase in false positives, a decrease
in false negatives and an average accuracy of \(95.02\%\ \pm 2.78\) in
276 images. Wang et al. developed a CNN-based detection of calcification
using mammograms and achieved human-like perfection [134]. Google
used DL to analyse retinal fundus images and predicted multiple
cardiovascular risk factors with good accuracy [133]. The team
observed that retinal fundus images could accurately predict
cardiovascular diseases and found method comparable to traditional risk
calculators. Benchmarking studies and outcomes are summarized in Table
III. Further, the favourable outcomes inspire the application of DL in
stiffness computation.