Fig 1.
Example well images of DENV and ZIKV plaques and resulting Viridot plaques counts.
The type of camera used for acquiring the image, cell type for infection, and the virus (with location and year of isolation) is printed above each raw image. The image in which plaques have been counted is shown below, with the resulting plaque count printed in the top left corner. Well images reflect separate experiments done by different analysts.
Fig 2.
The Viridot plaque counter program.
The scrollable panel on the left contains the parameter settings for plaque counting and buttons to load and save parameter settings. Selection of plates and images for analysis is shown in the center. A well containing DENV plaques is shown, with counted plaques circled in red and the well area in which plaques were counted circled in blue. The plaque count is shown above the image. The panel to the right includes options to save plaque count tables, circled plaque images, and plaque sizes.
Fig 3.
Evaluation of the Viridot automatic parameter-identification mode for a training set and two test sets.
Total number (A-C) or proportion (D-F) of plaques per well image counted by the Viridot automatic parameter-identification program that are discordant with analyst-estimated plaque counts. A single analyst estimated the number of miscounted plaques by checking raw images for plaques not circled by the automatic counter.
Table 1.
Comparison of manual plaque counting methods within and between Analyst 1 and Analyst 2 (n = 72 wells).
Table 2.
Comparison of Viridot plaque counting methods within and between Analyst 1 and Analyst 2 (n = 36 wells).
Table 3.
Comparison of the standardized manual and Viridot plaque counting methods within and between Analyst 1 and Analyst 2 (n = 168 wells).
Fig 4.
Comparison of standardized manual and Viridot plaque counts and neutralizing antibody titers.
(A) Comparison of manual and Viridot counts of DENV3 plaques on C6/36 cells for six repeat titrations of a DENV3-positive antiserum by two analysts (A1 and A2). Linear regression lines are shown for each titration and analyst. The grey line indicates perfect correspondence. (B) PRNT50 titers with 95% confidence intervals for the six titrations counted by A1 and A2, manually and with Viridot. Resulting neutralization curves for A1 (C) and A2 (D) for the six titrations, counted either manually (solid lines) or with Viridot (dotted).
Fig 5.
The Viridot neutralization titer estimation program.
The scrollable panel on the left contains the parameter settings for neutralizing antibody titer estimation. In the top-center, users can import plaque count tables and the plate template (the user can switch tabs to see the imported templates and tables). In the center, summary statistics for control wells and the neutralizing antibody titration curves are shown, with 95% confidence intervals in blue and error bars (± 1sd shown) in gray. The estimated neutralizing antibody titer is printed above. The panel to the right includes options to save neutralizing antibody titer data to a table and to print the image of neutralization curves.
Fig 6.
Comparison of manual and Viridot counts of DENV1-4 and ZIKV plaques and neutralizing antibody titers on C6/36 cells or Vero cells.
(A) Comparison of standardized manual and Viridot plaque counts, with linear regression (solid lines) showing concordance: ρc values for each titration are indicated in the legend. The dotted black line indicates perfect correspondence. (B and C) Resulting neutralization curves for each antiserum titration, counted either manually (solid lines) or with Viridot (dotted). The geometric average ratio of the manual to Viridot for all PRNT50 titers is 0.94 (95% CI: 0.86–1.02). (D) Comparison of standardized manual and Viridot counted plaques in well images acquired using camera phones. Linear regression lines (solid lines), concordance (ρc) values for plaque data of each image type (legend), and perfect correspondence (black dotted line) are shown.
Fig 7.
Viridot counting of camera phone well images and spots acquired using other staining methods.
(A and B) Images of DENV4 plaques on C6/36 cells (peroxidase-labeled antibody and TrueBlue stain) acquired using the iPhone 7 plus of the bottom and top of the plate. (C) Images of ZIKV plaques on Vero cells on white plates, with images acquired using the Samsung Galaxy S7. (D) Images of ZIKV plaques on Vero cells acquired using the Zeiss Axio Imager.M1 microscope, using bright top lighting. (E) IFN-γ ELISPOT of the T-cell response to Influenza A nucleoprotein peptide 147–155, imaged with the Zeiss Axio Cam Imager.M2 scope with the AxioCam MRc camera, 5x objective lens. (F) Total (left) and DENV4-specific IgG+ memory B cells (right). Spots were detected with TRITC-labeled IgG-specific antibody or Qdot-conjugated fluorescent antibody using an ELISPOT based Fluorospot, with images acquired using the ImmunoSpot Analyzer. (G) LCMV plaques on Vero E6 cells resolved using crystal violet and imaged using the Alpha Innotech digital camera with the Computar H6Z0812M motorized zoom lens.