Additional file 10 of Application of convolutional neural networks towards nuclei segmentation in localization-based super-resolution fluorescence microscopy images

Additional file 10: Figure S9. Tiling method example with and without overlap. When the test image is divided into non-overlapping squares, or tiles, for segmentation, the borders between tiles will act as artificial borders between labeled detected instances upon image reconstruction, creating divided predictions (upper diagram). Overlapping allows for a post-segmentation program to sample the overlapping region in adjacent tiles and merge overlapping instances during reconstruction (lower diagram).