Table 4 Results of segmentation tasks for the Dataset.

From: A multicenter bladder cancer MRI dataset and baseline evaluation of federated learning in clinical application

Method \DSC

Test Average

Test Center1

Test Center2

Test Center3

Test Center4

Centralized

0.841

0.789

0.864

0.860

0.850

Center 1

0.770

0.747

0.759

0.814

0.762

Center 2

0.740

0.645

0.763

0.781

0.769

Center 3

0.728

0.611

0.749

0.794

0.758

Center 4

0.741

0.600

0.799

0.828

0.738

FedAvg

0.819

0.761

0.859

0.828

0.829

FedProx

0.840

0.785

0.850

0.879

0.847

FedBN

0.837

0.781

0.860

0.862

0.844

SiloBN

0.831

0.769

0.856

0.865

0.834

  1. In the ‘Method’ column, ‘Centralized’ indicates that the training sets from 4 centers are mixed to train the DL model, ‘Center1/2/3/4’ means that the dataset from this center is used for training the DL model, and then all the testing datasets from all for centres are used for testing. ‘Test Center1/2/3/4’ refers to the DSC performance on the test set from each single center, while Test Average denotes the average DSC results across all four centers. ‘FedAvg’, ‘FedProx’, ‘FedBN’, ‘SiloBN’ refers to the four FL methods.
Back to article page