Skip to main content

Table 1 Imaging system and typical parameters

From: Fully automated pelvic bone segmentation in multiparameteric MRI using a 3D convolutional neural network

MR Vendor

Algorithm Development# (N = 264)

External Set* (N = 60)

Typical parameters (DWI sequence)

Training set (N = 208)

Validation set (N = 28)

Testing set (N = 28)

3.0 T Discovery (Ge healthcare, Milwaukee, WI)

N = 157

N = 20

N = 20

N = 41

B value: 0, 800 s/mm2;

Imaging matrix:256 × 256;

Echo time: 60 ms;

Repetition time: 3000 ms;

Field of view: 360 mm;

Section thickness: 4 mm

Number of slices: 25

3.0 T Intera (Philips Medical Systems, Best, the Netherlands)

N = 37

N = 5

N = 5

N = 11

B value: 0, 1000 s/mm2;

Imaging matrix: 240 × 240;

Echo time:78 ms;

Repetition time: 4959 ms;

Field of view: 360 mm;

Section thickness:7 mm

Number of slices: 28

1.5 T Avanto (Siemens Medical Solutions, Erlangen, Germany)

N = 14

N = 3

N = 3

N = 8

B value: 0, 800 s/mm2;

Imaging matrix: 156 × 180;

Echo time:54 ms;

Repetition time: 3300 ms;

Field of view: 360 mm;

Section thickness: 7 mm

Number of slices: 24

  1. #The data in algorithm development were collected between August 2018 and August 2019
  2. *The data in external set were collected between January 2020 and March 2020