Ultrasound-based deep learning radiomics model for differentiating benign, borderline, and malignant ovarian tumours: a multi-class classification exploratory study

Table 4 Overall and class-specific classification performance

		AUC	Precision (%)	Recall (%)	F1 Score (%)	Acc (%)
Clinic_Sig	Benign ovarian tumour	0.68	65.33	89.91	75.68	62.72
	BOT	0.40	0.00	0.00	0.00	93.49
	Malignant ovarian tumour	0.73	42.11	16.33	25.40	69.23
	micro-average	0.83	62.72	62.72	62.72	62.72
	macro-average	0.61	35.81	35.41	33.69	75.15
	weighted-average		54.35	62.73	48.81	66.61
Rad_Sig	Benign ovarian tumour	0.75	72.93	88.99	80.17	71.60
	BOT	0.62	0.00	0.00	0.00	93.49
	Malignant ovarian tumour	0.74	52.78	38.78	44.71	72.19
	micro-average	0.85	68.64	68.64	68.64	68.64
	macro-average	0.71	41.90	42.59	41.63	79.09
	weighted-average		62.341	68.64	64.67	73.20
DTL_Sig	Benign ovarian tumour	0.87	83.93	86.24	85.07	62.72
	BOT	0.82	36.36	36.36	36.36	91.72
	Malignant ovarian tumour	0.84	63.04	59.18	61.05	78.11
	micro-average	0.89	75.15	75.14	75.14	75.14
	macro-average	0.85	61.11	60.59	60.83	77.52
	weighted-average		74.78	75.15	74.94	69.07
DLR_Sig	Benign ovarian tumour	0.84	84.91	82.57	83.72	79.29
	BOT	0.85	42.86	54.55	57.14	93.31
	Malignant ovarian tumour	0.83	63.27	63.27	63.27	78.70
	micro-average	0.90	75.14	75.14	75.14	75.14
	macro-average	0.84	63.68	66.80	68.04	83.77
	weighted-average		75.90	75.15	76.06	80.03

Clinic_Sig: clinical signature; Rad_Sig: radiomics signature; DTL_Sig: deep transfer learning signature; DLR_Sig: deep learning radiomic signature; Acc: Accuracy; BOT: Borderline ovarian tumour

ISSN: 1471-2342