Skip to main content

Advertisement

Table 1 Demographic characteristics and disease categories of the study subjects in different datasets

From: Development and validation of an endoscopic images-based deep learning model for detection with nasopharyngeal malignancies

Characteristics All Training set Validation set Test set Prospective test set
Subjects, n 8306 5557 807 1587 355
Mean (± SD), years 45.9 ± 12.7 45.8 ± 12.7 45.9 ± 12.7 45.7 ± 12.7 47.8 ± 13.0
Sex, n(%)
 Female 2562 (30.9) 1681 (30.3) 250 (31.0) 507 (32.0) 124 (34.9)
 Male 5612 (67.6) 3783 (68.1) 540 (66.9) 1058 (66.7) 231 (65.1)
 N/A 132 (1.6) 93 (1.7) 17 (2.1) 22 (1.4) 0 (0.0)
Disease category, n(%)
Normal 5713 (19.7) 3763 (19.2) 584 (21.7) 961 (18.2) 405 (28.3)
Malignancies
 NPC 19,107 (66.0) 13,061 (66.7) 1749 (65.0) 3564 (67.6) 731(51.1)
 Othersa 335 (1.2) 252 (1.3) 22 (0.8) 54 (1.0) 7 (0.4)
Benign diseasesb 3811 (13.2) 2500 (12.8) 335 (12.5) 691 (13.1) 287(20.1)
Images, n(%) 28,966 19,576 (67.6) 2690 (9.3) 5270 (18.2) 1430 (4.9)
  1. N/A not available, NPC nasopharyngeal carcinoma
  2. aLymphoma, rhabdomyosarcoma, olfactory neuroblastoma, malignant melanoma and plasmacytoma
  3. bPrecancerous/atypical hyperplasia, fibroangioma, leiomyoma, meningioma, minor salivary gland tumour, fungal infection, tuberculosis, chronic inflammation, adenoids/lymphoid hyperplasia, nasopharyngeal cyst and foreign bodies