Table 1 CT image characteristics.

From: Validation of musculoskeletal segmentation model with uncertainty estimation for bone and muscle assessment in hip-to-knee clinical CT images

Database

Inst.

Diagnosis

Patient

positioning

No. of

cases

Modality

Matrix

size

In-plane

resolution

[mm]

Slice

interval

[mm]

Internal training and testing (fivefold cross-validation)

DB#1

Osaka

Univ

Hosp.

Unilateral

hip OA

Supine

20

HiSpeed, GE

5122

0.7032 − 

0.7422

1.0–6.0*

30

Optima

CT660, GE

5122

0.7032 − 

0.8202

1.25

External validation (small-scale, with ground-truth labels of the GMED muscle)

DB#2

TCIA

Soft tissue

sarcoma

Supine

18

Discovery

ST, GE

5122

0.9772

3.75

DB#3

Hitachi

Medical

Care

Center

Colorectal

cancer

Supine

10

Supria, Hitachi

5122

0.6852

0.63

DB#4

Keio

Univ

Hosp.

Normal

Supine

20

Aquilion ONE,

Canon Medical

Systems

5122

0.6832

0.5

Standing

20

prototype

TSX-401R,

Canon

Medical

Systems

5122

0.6832

0.5

Large-scale predictive uncertainty analysis (without ground-truth labels)

DB#5

Osaka

Univ

Hosp.

Uni/

bilateral

hip OA

Supine

460

HiSpeed, GE

5122

0.7032 − 

0.7422

1.0–6.0 

2119

Optima

CT660, GE

5122

0.7032 − 

0.8202

0.675-

3.75

  1. *pelvis and proximal femur: 2.0 mm, femoral shaft region: 6.0 mm, distal femur region: 1.0 mm.