Skip to main content
. 2021 May 18;6(3):e00028-21. doi: 10.1128/mSystems.00028-21

FIG 2.

FIG 2

Harmony bacterial image analysis workflow for Gram-negative rods (A and B) and Gram-positive cocci (C and D). (A) Using basic bright-field correction and maximum projection, images were segmented by filtering the images using texture properties based on the FM4-64 channel to remove any background. The image region was filled and resized, and border objects were excluded to include only whole objects. The image region was further calculated using FM4-64 and DAPI fluorescence, and individual bacteria were defined. (*, single planes were analyzed for S. Typhimurium; **, left and bottom border object are not displayed as the image is cropped for better visualization). (B) Bacterial morphology and stain intensity properties were calculated using DAPI, SYTOX green, and FM4-64 fluorescence. Finally, a linear classifier was used to train the software to define single bacterial cells and exclude any artifacts. (C) Using basic bright-field correction and maximum projection, the bacterial region was defined using a calculated image based on DAPI and FM4-64 channels. Individual bacteria were identified within the image region, and the bacterial regions were defined and resized into individual bacterial cells. Any artifacts were removed using size filters. (D) Bacterial morphology and stain intensity properties were calculated using DAPI, SYTOX green, and FM4-64 fluorescence. Scale bars shown on input images correspond to 20 μm.