Table 3

Intrastrategy and interstrategy comparisons of DLRE and 2D-SWE for their relationship of diagnostic accuracy versus the number of image/measurement acquisitions in assessing liver fibrosis stages in training and validation cohorts

n (P)AUCSensitivity (%)Specificity (%)PPV (%)NPV (%)LR+LR−
Cirrhosis (F4)
DLRE
 1 imageT266 (24.8%)0.94**
(0.91 to 0.97)
95.5
(94.2 to 96.8)
84.5
(79.6 to 89.4)
67.0
(60.4 to 73.8)
98.3
(96.1 to 99.6)
6.2
(5.8 to 6.4)
0.1
(0.0 to 0.1)
V132 (24.2%)0.84***
(0.81 to 0.87)
75.0
(72.6 to 77.4)
82.0
(77.3 to 86.7)
57.1
(50.4 to 63.4)
91.1
(85.4 to 97.2)
4.2
(3.9 to 4.3)
0.3
(0.3 to 0.4)
 3 imagesT266 (24.8%)1.00
(0.98 to 1.00)
100.0
(98.0 to 100.0)
100.0
(98.0 to 100.0)
100.0
(99.0 to 100.0)
100.0
(99.0 to 100.0)
0.0
(0.0 to 0.0)
V132 (24.2%)0.96
(0.93 to 0.98)
84.4
(76.9 to 91.9)
95.0
(91.2 to 98.7)
84.4
(78.5 to 90.1)
95.0
(90.1 to 98.9)
16.9
(14.9 to 18.6)
0.2
(0.1 to 0.2)
 5 imagesT266 (24.8%)1.00
(0.99 to 1.00)
100.0
(98.0 to 100.0)
100.0
(98.0 to 100.0)
100.0
(99.0 to 100.0)
100.0
(99.0 to 100.0)
0.0
(0.0 to 0.0)
V132 (24.2%)0.97
(0.94 to 0.99)
96.9
(94.7 to 99.1)
88.0
(82.8 to 92.2)
72.1
(65.8 to 79.1)
98.9
(94.8 to 99.7)
8.1
(7.9 to 8.3)
0.0
(0.0 to 0.1)
2D-SWE
 1 measurementT266 (24.8%)0.87******
(0.83 to 0.91)
84.9
(73.9 to 92.5)
75.5
(68.9 to 81.3)
53.3
(43.3 to 63.1)
93.8
(88.9 to 97.0)
3.5
(3.0 to 3.9)
0.2
(0.1 to 0.4)
V132 (24.2%)0.86
(0.78 to 0.91)
93.8
(79.2 to 99.2)
74.0
(64.3 to 82.3)
53.6
(39.7 to 67.0)
97.4
(90.8 to 99.7)
3.6
(3.1 to 4.2)
0.1
(0.0 to 0.3)
 3 measurementsT266 (24.8%)0.88******
(0.83 to 0.92)
86.4
(75.7 to 93.6)
74.5
(67.9 to 80.4)
52.8
(42.9 to 62.5)
94.3
(89.5 to 97.4)
3.4
(3.0 to 3.8)
0.2
(0.1 to 0.4)
V132 (24.2%)0.85*****
(0.78 to 0.91)
87.5
(71.0 to 96.5)
76.0
(66.4 to 84.0)
53.8
(39.5 to 67.8)
95.0
(87.7 to 98.6)
3.7
(3.1 to 4.3)
0.2
(0.1 to 0.4)
 5 measurementsT266 (24.8%)0.87******
(0.83 to 0.91)
80.3
(68.7 to 89.1)
79.0
(72.7 to 84.4)
55.8
(45.2 to 66.0)
92.4
(87.4 to 95.9)
3.8
(3.3 to 4.4)
0.3
(0.1 to 0.4)
V132 (24.2%)0.86*****
(0.79 to 0.92)
87.5
(71.0 to 96.5)
76.0
(66.4 to 84.0)
53.8
(39.5 to 67.8)
95.0
(87.7 to 98.6)
3.7
(3.1 to 4.3)
0.2
(0.1 to 0.4)
Advanced fibrosis (≥F3)
DLRE
 1 imageT266 (83.8%)0.91*
(0.86 to 0.95)
86.8
(81.2 to 92.4)
81.7
(76.4 to 87.1)
86.2
(80.6 to 91.5)
82.5
(77.1 to 87.5)
4.8
(4.2 to 5.3)
0.2
(0.1 to 0.2)
V132 (83.3%)0.82**
(0.75 to 0.89)
75.3
(70.3 to 80.3)
79.7
(71.4 to 87.7)
82.1
(77.0 to 87.6)
72.3
(67.1 to 77.9)
3.7
(3.5 to 3.9)
0.3
(0.3 to 0.3)
 3 imagesT266 (83.8%)0.96
(0.92 to 0.99)
90.1
(86.8 to 93.4)
88.7
(82.3 to 95.1)
91.3
(83.8 to 97.3)
81.2
(74.8 to 87.6)
8.0
(7.7 to 8.2)
0.1
(0.1 to 0.1)
V132 (83.3%)0.95
(0.91 to 0.98)
94.5
(90.8 to 98.2)
86.4
(81.2 to 91.6)
89.6
(82.5 to 95.6)
92.7
(87.5 to 96.8)
7.0
(6.5 to 7.4)
0.1
(0.1 to 0.1)
 5 imagesT266 (83.8%)0.99
(0.97 to 1.00)
97.4
(93.2 to 99.1)
95.7
(91.4 to 98.6)
95.6
(90.2 to 97.8)
71.9
(63.8 to 78.5)
22.4
(21.5 to 23.3)
0.0
(0.0 to 0.0)
V132 (83.3%)0.98
(0.96 to 1.00)
90.4
(84.2 to 94.6)
98.3
(95.3 to 99.3)
98.5
(95.8 to 99.8)
89.2
(83.5 to 94.5)
53.3
(51.0 to 55.3)
0.1
(0.1 to 0.1)
2D-SWE
 1 measurementT266 (83.8%)0.81******
(0.75 to 0.82)
72.2
(64.3 to 79.2)
75.7
(66.8 to 83.2)
79.6
(71.8 to 86.0)
67.4
(58.6 to 75.5)
3.0
(2.6 to 3.4)
0.4
(0.2 to 0.6)
V132 (83.3%)0.83
(0.75 to 0.89)
68.5
(56.6 to 78.9)
89.8
(79.2 to 96.2)
89.3
(78.0 to 96.0)
69.7
(58.1 to 79.8)
6.7
(5.6 to 8.0)
0.4
(0.2 to 0.8)
 3 measurementsT266 (83.8%)0.81******
(0.75 to 0.85)
80.79
(73.6 to 86.7)
68.7
(59.4 to 77.0)
77.2
(69.9 to 83.5)
73.1
(63.7 to 81.3)
2.6
(2.2 to 3.0)
0.3
(0.2 to 0.4)
V132 (83.3%)0.84*****
(0.77 to 0.90)
78.1
(66.9 to 86.9)
79.7
(67.2 to 89.0)
82.6
(71.5 to 90.7)
74.6
(61.9 to 84.8)
3.8
(3.2 to 4.6)
0.3
(0.1 to 0.5)
 5 measurementsT266 (56.8%)0.81******
(0.75 to 0.85)
81.5
(74.3 to 87.3)
70.4
(61.2 to 78.6)
78.3
(71.1 to 84.5)
74.3
(65.0 to 82.2)
2.8
(2.4 to 3.2)
0.3
(0.2 to 0.4)
V132 (55.3%)0.85******
(0.78 to 0.90)
79.5
(68.4 to 88.0)
78.0
(65.3 to 87.7)
81.7
(70.6 to 89.9)
75.4
(62.7 to 85.5)
3.6
(3.0 to 4.3)
0.3
(0.1 to 0.5)
Significance fibrosis (≥F2)
DLRE
 1 imageT266 (83.8%)0.95
(0.93 to 0.97)
89.7
(83.5 to 95.9)
93.0
(87.5 to 98.5)
98.5
(95.7 to 99.7)
63.5
(58.4 to 68.1)
12.9
(11.6 to 13.5)
0.1
(0.1 to 0.1)
V132 (83.3%)0.74
(0.68 to 0.79)
76.4
(71.3 to 81.5)
72.7
(68.1 to 77.3)
93.3
(96.7 to 98.1)
38.1
(35.5 to 40.5)
2.8
(2.7 to 2.9)
0.3
(0.3 to 0.4)
 3 imagesT266 (83.8%)0.97
(0.94 to 0.99)
97.8
(92.4 to 99.9)
95.3
(92.5 to 98.1)
99.1
(97.2 to 99.9)
89.1
(83.4 to 94.5)
21.0
(19.8 to 22.2)
0.0
(0.0 to 0.0)
V132 (83.3%)0.82
(0.78 to 0.86)
67.3
(62.6 to 72.1)
95.5
(91.8 to 99.1)
98.7
(95.8 to 99.7)
36.8
(34.8 to 38.4)
14.8
(13.4 to 16.0)
0.3
(0.3 to 0.4)
 5 imagesT266 (83.8%)0.99
(0.97 to 1.00)
100.0
(99.0 to 100.0)
97.7
(92.5 to 99.3)
99.5
(97.8 to 99.9)
51.2
(48.5 to 53.9)
43.0
(41.8 to 44.5)
0.0
(0.0 to 0.0)
V132 (83.3%)0.85
(0.81 to 0.89)
69.1
(64.3 to 73.5)
90.9
(83.6 to 96.7)
97.4
(96.8 to 99.7)
37.0
(35.8 to 38.1)
7.6
(6.9 to 8.4)
0.3
(0.3 to 0.4)
2D-SWE
 1 measurementT266 (83.8%)0.72******
(0.66 to 0.77)
59.2
(52.4 to 65.7)
79.1
(64.0 to 90.0)
93.6
(88.2 to 97.0)
27.2
(19.6 to 35.9)
2.8
(2.3 to 3.4)
0.5
(0.3 to 0.9)
V132 (83.3%)0.76
(0.67 to 0.83)
60.0
(50.2 to 69.2)
86.4
(65.1 to 97.1)
95.7
(87.7 to 99.1)
30.2
(19.2 to 43.0)
4.4
(3.5 to 5.5)
0.5
(0.2 to 1.4)
 3 measurementsT266 (83.8%)0.74******
(0.68 to 0.79)
54.3
(47.5 to 60.9)
83.7
(69.3 to 93.2)
94.5
(89.0 to 97.8)
26.1
(19.0 to 34.2)
3.3
(2.8 to 4.0)
0.6
(0.3 to 1.1)
V132 (83.3%)0.77
(0.69 to 0.84)
51.8
(42.1 to 61.4)
95.5
(77.2 to 99.9)
98.3
(90.7 to 100.0)
28.4
(18.5 to 40.1)
11.4
(9.3 to 14.0)
0.5
(0.1 to 3.5)
 5 measurementsT266 (83.8%)0.74******
(0.68 to 0.79)
50.7
(43.9 to 57.4)
88.4
(74.9 to 96.1)
95.8
(90.4 to 98.6)
25.7
(18.8 to 33.5)
4.4
(3.7 to 5.2)
0.6
(0.2 to 1.3)
V132 (83.3%)0.77
(0.69 to 0.84)
49.1
(39.4 to 58.8)
95.5
(77.2 to 99.9)
98.2
(90.2 to 100.0)
27.3
(17.7 to 38.7)
10.8
(8.7 to 13.3)
0.5
(0.1 to 3.7)
  • Statistical quantifications were demonstrated with 95% CI, when applicable.

  • Intrastrategy comparison: for either DLRE or 2D-SWE, AUCs obtained by analysing three images/measurements were compared with these obtained by analysing one and five images/measurements, respectively (*P<0.05; **P<0.01; ***P<0.001).

  • Interstrategy comparison: for using the same number of images/measurements, AUCs obtained by DLRE were compared with these obtained by 2D-SWE in each liver fibrosis classification, respectively (****P<0.05; *****P<0.01; ******P<0.001).

  • 2D-SWE, two-dimensional shear wave elastography; AUC, area under the receiver operating characteristic curve; DLRE, deep learning Radiomics of elastography; LR+, positive diagnostic likelihood ratio; LR−, negative diagnostic likelihood ratio; n, number of patients; NPV, negative predictive value; P, prevalence; PPV, positive predictive value; T, training cohort; V, validation cohort.