BMC Medical Imaging

Table 1 Performance of diferent models for predicting LNM in the training and validation cohorts

From: Lymph node metastasis prediction and biological pathway associations underlying DCE-MRI deep learning radiomics in invasive breast cancer

Model	Training cohort					Validation cohort
Model	Radiomics	GoogLeNet	ResNet50	DLRN		Radiomics	GoogLeNet	ResNet50	DLRN
AUC	0.95	0.91	0.85	0.98		0.83	0.77	0.71	0.87
95%CI	0.91–0.98	0.84–0.96	0.76–0.94	0.93-1.00		0.69–0.94	0.64–0.90	0.56–0.86	0.76–0.95
ACC	0.88	0.82	0.81	0.93		0.79	0.77	0.70	0.90
SEN	0.93	0.90	0.83	1.00		0.77	0.86	0.66	1.00
SPEC	0.83	0.74	0.79	0.73		0.82	0.64	0.77	0.85
PPV	0.83	0.76	0.77	0.87		0.87	0.79	0.82	0.85
NPV	0.93	0.90	0.84	1.00		0.69	0.74	0.59	1.00

Back to article page

ISSN: 1471-2342

Contact us

General enquiries: journalsubmissions@springernature.com