The application of texture quantification in hepatocellular carcinoma using CT and MRI: a review of perspectives and challenges

Table 4 summary of studies showing the predictive performance of radiomics signature, clinical-radiological and the combined models

Study	Objectives	No. of subjects	The area under the ROC curve			Sensitivity			Specificity			Best model
Study	Objectives	No. of subjects	RS	CM	COM	RS	CM	COM	RS	CM	COM	Best model
Ma et al. [56]	Preoperative prediction of MVI	157 (T:110, V: 47)	0.793	0.761	0.801	0.656	0.944	0.889	0.944	0.655	0.759	COM
Yang et al. [41]	Prediction of MVI	208 (T: 146, V: 62)	0.837	0.759	0.861	0.842	0.737	0.895	0.744	0.674	0.814	COM
Xu et al. [29]	Prediction of MVI and survival	495 (T:350, V:145)	0.806	N/A	0.889	0.755	0.653	0.898	0.719	0.760	0.792	COM
Zhang et al. [57]	Prediction of MVI	267 (T:194, V:73)	0.820	0.721	0.858	0.692	0.269	0.808	0.809	0.936	0.861	COM
Zhu et al. [58]	Prediction of MVI	142 (T:99, V:43)	0.773	N/A	0.794	0.750	N/A	0.812	0.815	N/A	0.852	COM
Zhang et al. [59]	Prediction of early recurrence	155 (T:108, V:47)	0.728	0.814	0.841	0.696	0.783	0.913	0.708	0.833	0.750	COM
Zhou et al. [60]	Prediction of early recurrence	215	0.817	0.781	0.708	0.794	0.784	0.824	0.699	0.619	0.708	COM

T training cohort, V validation cohort, N/A not available, ROC receiver operating characteristic curve, RS Radiomics signature, CM Clinical model, COM Combined model

ISSN: 1470-7330