Table 1:

Studies evaluating radiomic tools for early diagnosis in hepatocellular carcinoma.

Author	CT/MRI	N (Train / Valid)	Extraction Tool	Specific Outcome Measured	Statistical Result	Clinical Model	RQS
Dankerl 2013	CT	372	CADx	Differentiation of benign vs. malignant lesion (nodule vs. HCC)	AUC 0.75 for textural features AUC 0.91 for texture + semantic	No	5
Song 2019	CT	84	Omni-Kinetic	Differentiation of benign vs. malignant lesion (HCC vs. HH vs. FNH vs. HA)	AUC 0.927 for textural features	No	9
Stocker 2018	MRI	108	Matlab	Differentiation of benign vs. malignant lesion	AUC 0.92 arterial phase	No	7
Li 2017	MRI	T: 112 V:50	Internal	Differentiation of HH from HCC	AUC 0.73 for GLCM Energy-mean	No	10
Oyama 2019	MRI	T: 50,50 V: 50	Matlab	Differentiation of HH from HCC	AUC 0.95 textural features	No	9
Wu 2019	MRI	369	Internal	Differentiation of HH from HCC	AUC 0.89 textural features	No	8
Mokrane 2019	CT	T: 142 V: 36	Internal	Categorize indeterminate nodule as high-risk or low-risk for HCC	AUC 0.74 for training cohort AUC 0.66 for validation cohort	No	10
Asayama 2016	MRI	84	Internal	Comparison of individual textural features of non-cancerous parenchyma between those with and without HCC	p = 0.0006 for kurtosis p = 0.0152 for skewness	No	6
Rosenkrantz 2015	MRI	20	Internal	Progression of hypovascular nodule to likely HCC on subsequent MRI	AUC 0.68 for skewness	No	7

CT: computed tomography; MRI: magnetic resonance imaging; AUC: area under the curve; HCC: hepatocellular carcinoma; HH: hepatic hemangioma; FNH: focal nodular hyperplasia; HA: hepatic adenoma