From: Application and prospects of AI-based radiomics in ultrasound diagnosis
Reference | Number of patient | Tumor characteristic | Imaging modality | Function and prediction result | Method |
---|---|---|---|---|---|
Tong et al. [27] | 558 | Pancreatic ductal adenocarcinoma, chronic pancreatitis | CEUS | AUC in internal validation is 0.978 (95%CI: 0.950–0.996) | Deep learning radiomics (DLR) model |
Chen et al. [28] | 221 | Breast cancer | CEUS | Sensitivity of 97.2% and an accuracy of 86.3% | Three-dimensional convolutional neural network (CNN) model |
Liu et al. [29] | 130 | HCC | CEUS | AUC (R-DLCEUS) = 0.93, AUC (radiomics-based time intensity curve of the CEUS model (R-TIC)) = 0.80, AUC (radiomics-based BUS image model (R-BMode)) = 0.81 | R-DLCEUS, R-TIC, R-BMode |
Liu et al. [30] | 419 | Very-early or early stage HCC | CEUS | 17.3% RFA patients and 27.3% SR patients should swap their treatment | Deep learning-based radiomics model |
Sun and Lu [31] | 156 | Diabetic nephropathy | CEUS | Experimental group kidney volume: 136.07 ± 22.16 cm3, control group kidney volume: 159.11 ± 31.79 cm3 | Poisson three-dimensional reconstruction algorithm |
Meng et al. [32] | CMLN dataset: 199, BL dataset: 146 | Metastasis cervical lymph nodes, breast lesion | CEUS | CMLN:Â 91.05% dice and 80.06% IOU; BL:Â 89.97% dice and 75.62% IOU | CEUSegNet |
Iwasa et al. [33] | 100 | Pancreatic tumors | CEUS | IOU:Â 0.77 | U-Net |