Tiling Artifacts and Trade-Offs of Feature Normalization in the Segmentation of Large Biological Images
Year: 2025
Type: conference-paper
Abstract: Segmentation of very large images is a common problem in microscopy, medical imaging or remote sensing. The problem is usually addressed by sliding window inference, which can theoretically lead to seamlessly stitched predictions. However, in practice many of the popular pipelines still suffer from tiling artifacts. We investigate the root cause of these issues and show that they stem from the normalization layers within the neural networks. We propose indicators to detect normalization issues and further explore the trade-offs between artifact-free and high-quality predictions, using three diverse microscopy datasets as examples. Finally, we propose to use BatchRenorm as the most suitable normalization strategy, which effectively removes tiling artifacts and enhances transfer performance, thereby improving the reusability of trained networks for new datasets.
Language: English
Subfield: Biophysics
Domain: Life Sciences
Open Access status: green
Awards: In situ Structural Analysis of Molecular Crowding and Phase Separation, EMBL Interdisciplinary Postdoc Programme 4, G2354866577, G2848826123, G3533715186, G584537252, EXC 2067: Multiscale Bioimaging: Von molekularen Maschinen zu Netzwerken erregbarer Zellen, Next generation imaging technologies to probe structure and function of biological specimen across scales in their natural context, G8727600513
OpenAlex ID: W7160054982
