Skip to main content
. 2024 Feb 29;2024:3711123. doi: 10.1155/2024/3711123

Table 4.

Ultrasound image features for predicting malignant lymph nodes in the training and validation sets.

Characteristics Total (n = 1082) Training set (n = 760) Validation set (n = 322) P value
Long-axis (median [IQR]) 2.000 [1.500, 2.770] 2.000 [1.500, 2.752] 2.000 [1.555, 2.800] 0.602
Number of passes per node (median [IQR]) 3.000 [2.000, 3.000] 3.000 [2.000, 3.000] 3.000 [2.000, 3.000] 0.017
Short-axis (%) 0.390
 >1 cm 882 (81.5%) 614 (80.8%) 268 (83.2%)
 ≤1 cm 200 (18.5%) 146 (19.2%) 54 (16.8%)
Long-axis/short-axis ratio (%) 0.955
 ≥1.5 229 (21.2%) 160 (21.1%) 69 (21.4%)
 <1.5 853 (78.8%) 600 (78.9%) 253 (78.6%)
Echo intensity (%) 0.087
 Hyperechoic 32 (3.0%) 25 (3.3%) 7 (2.2%)
 Hypoechoic 1041 (96.2%) 726 (95.5%) 315 (97.8%)
 Isoechoic 9 (0.8%) 9 (1.2%) 0 (0.00)
Echogenicity (%) 0.980
 Heterogeneous 671 (62.0%) 472 (62.1%) 199 (61.8%)
 Homogeneous 411 (38.0%) 288 (37.9%) 123 (38.2%)
Margin (%) 0.712
 Indistinct 866 (80.0%) 611 (80.4%) 255 (79.2%)
 Distinct 216 (20.0%) 149 (19.6%) 67 (20.8%)
Blood flow (%) 0.646
 Rich 890 (82.3%) 622 (81.8%) 268 (83.2%)
 Lacking 192 (17.7%) 138 (18.2%) 54 (16.8%)
Fusion (%) 0.268
 Absent 822 (76.0%) 585 (77.0%) 237 (73.6%)
 Present 260 (24.0%) 175 (23.0%) 85 (26.4%)
Central hilar structure (CHS) (%) 0.237
 Absent 740 (68.4%) 511 (67.2%) 229 (71.1%)
 Present 342 (31.6%) 249 (32.8%) 93 (28.9%)