TABLE 1.
Comparison of LUS image classification methods.
| Method | Author | Objective | Accuracy | Sensitivity | Specificity |
|---|---|---|---|---|---|
| Supervised feed forward ANN (2018) | Correa et al. | Pediatric Pneumonia | — | 90.9% | 100% |
| ANN (2016) | Barrientos et al. | Pneumonia | — | 91.5% | 100% |
| CNN (2020) | Born et al. | COVID-19 | 92% | 96% | 79% |
| CNN (2020) | Cristiana et al. | B-lines (presence) | 94% | — | — |
| CNN (2020) | Cristiana et al. | B-line (severity) | 54% | — | — |
| CNN (2019) | van Sloun and Demi | B-lines (in-vitro) | 91.7% | 91.5% | 91.8% |
| CNN (2019) | van Sloun and Demi | B-lines (in-vivo) | 89.2% | 87.1% | 93% |
| CNN (2018) | Kulhare et al. | Multiple Abnormalities | — | ¿85% | ¿85% |
| SVM Classifier (2020) | Carrer et al. | COVID-19 | 88–94% | — | — |
| RVM Classifier (binary) (2016) | Veeramani and Muthusamy | Healthy lung | 100% | 100% | 100% |
| RVM Classifier (multiclass) (2016) | Veeramani and Muthusamy | Multiple Abnormalities | 100% | 100% | 100% |
| Stochastic Method (2013) | Brattain et al. | B-lines | 100% | — | — |