Title | ||
---|---|---|
Quantifying the effects of data augmentation and stain color normalization in convolutional neural networks for computational pathology. |
Abstract | ||
---|---|---|
•Combining color augmentation and color normalization achieves the best performance.•Using color augmentation is essential to reduce the generalization error.•The specific type of color augmentation (HSV or HED) and its strength is irrelevant.•Color normalization based on neural networks is superior to more traditional methods.•Skip color normalization to save computational resources at a negligible performance cost. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1016/j.media.2019.101544 | Medical Image Analysis |
Keywords | Field | DocType |
Deep learning,Convolutional neural network,Computational pathology | H&E stain,Pattern recognition,Stain,Computer science,Convolutional neural network,Stain tissue,Artificial intelligence,Generalization error,Artificial neural network,Color normalization,Pathology | Journal |
Volume | ISSN | Citations |
58 | 1361-8415 | 9 |
PageRank | References | Authors |
0.93 | 20 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
David Tellez | 1 | 21 | 2.36 |
Geert Litjens | 2 | 996 | 50.79 |
Péter Bándi | 3 | 9 | 0.93 |
Wouter Bulten | 4 | 14 | 2.76 |
John-Melle Bokhorst | 5 | 9 | 0.93 |
Ciompi Francesco | 6 | 837 | 39.53 |
van der Laak Jeroen | 7 | 22 | 5.41 |