Skip to main content

Table 2 A comparison of SSL pretraining studies for 2D and 3D CNNs that investigated lung nodule detection using the LIDC-IDRI dataset and the LUNA 2016 splits. The table gives test AUC as reported by the authors

From: A survey of the impact of self-supervised pretraining for diagnostic tasks in medical X-ray, CT, MRI, and ultrasound

Method

Initialization

First Author [ref]

Identifier

Random

ImageNet

SSL

Zhou [73]

Models Genesis (2D)

0.9603

0.9779

0.9745

Models Genesis (3D)

0.9603

-

0.9834

Haghighi [74]

Semantic Genesis (2D)a

0.9425

0.9750

0.9750

Semantic Genesis (3D)

0.9425

-

0.9847

Feng [75]

Parts2Whole (3D)

0.9425

-

0.9867

  1. aValues were estimated via visual inspection, since results were reported in a chart