Supplementary Figure 5: Additional images for branching.

(a) Machine-learning based nucleus segmentation for signal quantification: The deep learning model was trained with a manually annotated dataset to enable automatic identification of nuclear contours to be followed by watershed segmentation. The image highlights nuclear contours (right side of the image) from DAPI staining (left side). Right panel: Watershed segmentation was used to segment (pink) the pixels corresponding to nuclei of each cell. (b-c) Images display a typical germinal center in human FFPE tonsil samples stained for Ki-67 (Alexa647, red) by Immuno-SABER. DAPI stain (blue) is shown for reference. Qualitatively similar amplification levels were obtained by long and short hybridization times (at 37 °C) for the primary concatemer and branching concatemer (75 min each). (d) Iterative SABER of SV2 (Alexa647) in a 40 μm mouse retina cryosection. (e) For comparison SV2 staining with TSA was performed using mono HRP conjugated secondary antibodies. (f) Zoom-out and zoom-in views of the high-magnification confocal images in Fig. 3f displayed at different scaling ranges for comparative visualization. 10 min TSA amplification is included for further comparison. (g) Application of tyramide-Alexa647 for the maximum recommended incubation of 10 min. The germinal center image on the left is scaled in the same range with Fig. 3d. Zoom-in on the right is included to display the significant blurring of the signal at 10 min incubation.