BMC Medical Informatics and Decision Making

Table 2 The distribution of the dataset in the detection and classification experiment

From: Artificial intelligence for non-mass breast lesions detection and classification on ultrasound images: a comparative study

Dataset		Pathology		Total
Dataset A		Normal	NMLs
Training Set	Patients, n, %	481	172	653
	Patients, n, %	60%	40%	653
	Images, n, %	702	608	1310
	Images, n, %	54%	46%	1310
Validation Set	Patients, n, %	66	31	97
	Patients, n, %	68%	32%	97
	Images, n, %	100	88	188
	Images, n, %	53%	47%	188
Testing Set	Patients, n, %	49	25	74
	Patients, n, %	66%	34%	74
	Images, n, %	201	174	375
	Images, n, %	54%	46%	375
Dataset B		Benign MLs	Malignant MLs
Training Set	Patients, n, %	2728	1249	3977
	Patients, n, %	69%	31%	3977
	Images, n, %	7260	3359	10,619
	Images, n, %	68%	32%	10,619
Validation Set	Patients, n, %	352	158	510
	Patients, n, %	69%	31%	510
	Images, n, %	863	426	1289
	Images, n, %	67%	33%	1289
Internal Testing Set	Patients, n, %	338	163	501
	Patients, n, %	67%	33%	501
	Images, n, %	862	477	1339
	Images, n, %	64%	36%
Dataset C		Benign NMLs	Malignant NMLs
External Testing Set	Patients, n, %	135	93	228
	Patients, n, %	59%	41%	228
	Images, n, %	505	365	870
	Images, n, %	58%	42%	870

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com