Skip to main content

Table 2 This review includes a comprehensive summary of the data collected from the studies

From: Deep learning-based evaluation of panoramic radiographs for osteoporosis screening: a systematic review and meta-analysis

Study/ Year

Country

Study Design

Number of patients

Mean age

Center

Reference test

validation techniques

DL Methods

External validation test

Nakamoto et al./ 2022 [15]

Japan

Retrospective

100

> 50

single-center

MCI

Random sampling

AlexNet, VGG16,

GoogleNet

NO

Lee et al./2020[11]

South Korea

Retrospective

680

> 50

single-center

DXA

Random sampling

CNN, VGG-16,

NO

Sukegawa et al./2022[18]

Japan

Retrospective

778

> 50

single-center

DXA

5-fold cross validation

EfficientNet b3,

ResNet50,

ResNet152,

ResNet18,

EfficientNet b0, EfficientNet b7

NO

Lee et al./2019[14]

South Korea

Retrospective

200

> 50

single-center

MCI

Random sampling

AlexNet

NO

Tassoker et al./2022[16]

Turkey

Retrospective

1488

> 50

single-center

MCI

5-fold cross validation

AlexNet, GoogleNet, ResNet50, ShuffleNet, SqueezeNet

NO

Dias et al./2025[17]

Brazil

Retrospective

471

> 50

single-center

MCI

5-fold cross validation

EfficientNet b5

EfficientNet b6

EfficientNet b7

NO

Gaudin et al./2024[19]

Germany

Retrospective

500

> 50

single-center

DXA

Random sampling

Densenet201

NO

  1. Mean age: mean age of participants. Center: data gathering centers (Single center/Multicenter). DL method: deep learning method. MCI: mandibular cortex index. DXA: dual-energy X-ray absorptiometry. DL Methods: Deep learning methods. NA: Not Available