Table 5 Performance of artificial intelligence algorithms measuring implants in total joint arthroplasty
Author | AI Technique | DCNN | AUC | Accuracy | Sensitivity/Recall | Precision/PPV | Specificity |
|---|---|---|---|---|---|---|---|
Jang et al., 2023 [28] | CNN Transfer Learning to Segment Relevant Landmarks | U-Net Model | NR | Zonal Mapping: Femoral-89% Tibial-91% All Zones-90% Cone Identification: Femoral-97.8% Tibial-100% Cone Placement: Femoral-95.7% Tibial-89.1% | NR | NR | NR |
Schwarz et al., 2022 [33] | IB Lab LAMA | NR | NR | HKA: 99% FCA: 99% TCA: 97% | NR | NR | NR |
Median (IQR) | NA | NA | NA | 97.3% (94.5%–99.3%) | NA | NA | NA |