Insights into Imaging

Table 2 The performance of MRI-based radiomics in predicting DMI and LVSI

From: Radiomics-based fertility-sparing treatment in endometrial carcinoma: a review

Factors and reference	Model and dataset type	Sensitivity	Specificity	Accuracy	AUC
DMI
Stanzione et al
	Model^R
	Training set	0.71 (10/14)	0.93 (27/29)	0.86 (37/43)	0.92
	Test set	0.67 (2/3)	1.00 (8/8)	0.91 (10/11)	0.94
Ueno et al
	Model^R
	Only one set	0.79 (46/58)	0.82 (65/79)	0.81 (111/137)	0.84
Ytre-Hauge et al
	Model^R
	Only one set	0.70 (53/76)	0.84 (83/99)	0.78 (136/175)	0.81
Kristine et al
	Model^WT
	Training set	NA	NA	NA	0.84
	Test set	NA	NA	NA	0.76
	Model^SS
	Training set	NA	NA	NA	0.85
	Test set	NA	NA	NA	0.77
Yan et al
	Model^CR
	Training set	1.00 (NA)	0.83 (NA)	0.87 (NA)	0.96
	Test set	0.72 (NA)	0.90 (NA)	0.87 (NA)	0.88
Zhu et al
	Model^R
	Training set	0.95 (NA)	0.93 (NA)	0.94 (NA)	0.93
	Test set	0.95 (NA)	0.93 (NA)	0.94 (NA)	0.92
Alejandro et al
	Model^R
	Only one set	0.81 (NA)	0.93 (NA)	0.86 (NA)	0.87
LVSI
Ueno et al
	Model^R
	Only one set	0.81 (55/68)	0.72 (50/69)	0.77 (105/137)	0.80
Luo et al
	Model^R
	Training set	0.83 (NA)	0.73 (NA)	NA	0.82
	Test set	0.78 (NA)	0.79 (NA)	NA	0.81
Long et al
	Model^R
	Training set	0.89 (32/36)	0.58 (59/102)	0.66 (91/138)	0.70
	Test set	0.86 (12/14)	0.63 (20/32)	0.70 (32/46)	0.75
	Model^CVF
	Training set	0.92 (33/36)	0.96 (98/102)	0.95 (131/138)	0.93
	Test set	0.93 (13/14)	0.63 (20/32)	0.72 (33/46)	0.81
Bereby-Kahane et al
	Model^R
	Only one set	0.70 (19/27)	0.59 (27/46)	0.63 (46/73)	0.59

MRI, magnetic resonance imaging; AUC, area under curve; DMI, deep myometrial invasion; LVSI, lymph-vascular space invasion
Model^R: Model constructed by radiomics features. Model^CVF: Model constructed by radiomics and computer vision features. Model^CR: Models constructed by clinical and radiomics features. Model^WT: Model constructed by whole-tumor radiomics features. Model^SS: Model constructed by single-slice radiomics features

Back to article page