| TECHNICAL NOTE |
|
| Year : 2022 | Volume
: 13
| Issue : 1 | Page : 7 |
|
|
Histo-fetch – On-the-fly processing of gigapixel whole slide images simplifies and speeds neural network training
Brendon Lutnick1, Leema Krishna Murali2, Brandon Ginley1, Avi Z Rosenberg3, Pinaki Sarder4
1 Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, New York, USA
2 Department of Biomedical Engineering, SUNY Buffalo, Buffalo, New York, USA
3 Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
4 Department of Pathology and Anatomical Sciences, SUNY Buffalo; Department of Biomedical Engineering, SUNY Buffalo, Buffalo, New York, USA
Correspondence Address:
Prof. Pinaki Sarder
Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, New York
USA
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/jpi.jpi_59_20
|
|
|
Background: Training convolutional neural networks using pathology whole slide images (WSIs) is traditionally prefaced by the extraction of a training dataset of image patches. While effective, for large datasets of WSIs, this dataset preparation is inefficient. Methods: We created a custom pipeline (histo-fetch) to efficiently extract random patches and labels from pathology WSIs for input to a neural network on-the-fly. We prefetch these patches as needed during network training, avoiding the need for WSI preparation such as chopping/tiling. Results & Conclusions: We demonstrate the utility of this pipeline to perform artificial stain transfer and image generation using the popular networks CycleGAN and ProGAN, respectively. For a large WSI dataset, histo-fetch is 98.6% faster to start training and used 7535x less disk space.
|
|
|
|
| [FULL TEXT] [PDF]* |
|
 |
|
