DAB staining in IHC: what this analyzer does
In immunohistochemistry (IHC), DAB produces a brown signal where the staining reaction occurs; hematoxylin provides a blue counterstain. This tool measures image pixels after staining, not antigen concentration or the number of positive cells. Upload a cropped hematoxylin + DAB field to inspect the stain-separation overlay before interpreting the reported values.
What this tool measures right now
This browser-based analyzer is designed for hematoxylin + DAB images that have already been cropped to a representative region. It separates brown DAB signal from the blue counterstain, detects tissue pixels, and reports positive tissue percentage, mean DAB optical density, H-Score, and optional DAB intensity summaries such as IntDen and CTCF. Tracking DAB intensity on the same control slide across staining runs for QC or CRO work? Measure each image here with the same settings, then compare many images at once in IHC batch analysis.
- Ruifrok deconvolution for stain separation, plus an HSV fallback for quick brown-signal checks.
- Threshold-based quantification of positive tissue within the detected tissue mask.
- Pixel-based H-Score using relative, fixed-OD, or IHC Profiler-style grading modes.
- Live overlay preview so you can see whether the thresholding agrees with the visible stain pattern.
How to read the results
The tissue mask determines which pixels enter the calculations. Check its coverage and the DAB overlay before comparing values: changing the crop, white reference, or threshold can change the result even when the underlying slide is the same.
- Positive tissue (%) is the number of tissue pixels above the DAB threshold divided by all detected tissue pixels. It is an area fraction, not a positive-cell percentage.
- Mean DAB OD averages the separated DAB optical-density values across all detected tissue pixels, including pixels below the positivity threshold.
- H-Score groups tissue pixels as negative, weak (1+), moderate (2+), or strong (3+), then calculates 1 × weak % + 2 × moderate % + 3 × strong %. Its possible range is 0–300; it is a pixel-based score, not a pathologist's cell-based score.
- IntDen and CTCF summarize total DAB signal within the tissue mask. IntDen sums DAB OD there; CTCF subtracts the estimated background signal based on pixels outside the mask. They depend on the selected area, so do not interpret them as stain concentration.
Worked example: positive area versus H-Score
Suppose a field has 10% negative, 50% weak, 30% moderate, and 10% strong tissue pixels in fixed-OD mode. Positive tissue is 90%, while H-Score is (1 × 50) + (2 × 30) + (3 × 10) = 140. A field with 90% weak and 10% negative pixels has the same positive area but an H-Score of 90. These are illustrative calculations, not measured samples or diagnostic cutoffs.
Recommended workflow
- Crop first. Use a representative field or ROI before uploading. This tool does not draw ROIs or navigate whole-slide images.
- Start with Ruifrok. For standard hematoxylin + DAB slides, colour deconvolution is usually more stable than HSV.
- Check the overlay. Make sure the highlighted pixels match the brown stain you intend to quantify.
- Only then tune threshold or white reference. If the background is not neutral white, use the eyedropper before comparing results.
- Export your settings with the result. Threshold and scoring mode directly affect the numbers.
When the result looks unexpected
If the overlay highlights labels, borders, or blank background, crop a cleaner field and recheck tissue detection. Sample a blank area with the white-reference eyedropper when the background is tinted. If an automatic-threshold warning appears, inspect the overlay and try another threshold method or a manual threshold before using the percentage. If the tool reports downscaling, crop from the original image to inspect small strongly stained regions.
For several fields, keep the same acquisition and analysis settings and use IHC batch analysis to apply one parameter set. The CSV records the threshold, scoring mode, algorithm version, and warnings alongside the measurements.
Interpret with care
- This is not cell segmentation. The reported H-Score is calculated from pixel intensities inside the detected tissue mask, not from counted cells or nuclei.
- Comparison is only meaningful when acquisition is consistent. Keep microscope, illumination, exposure, and stain protocol as constant as possible.
- Check both the positivity threshold and the grading cutoffs. Relative mode derives its grading cutoffs from the current image's maximum tissue DAB value. Fixed-OD mode lets you set weak and moderate cutoffs, but an automatic positivity threshold is still recalculated for each image. For a shared numerical threshold, set it manually and inspect each overlay. The tool keeps grading cutoffs ordered and above the positivity threshold, so check the displayed cutoffs after changing it. Fixed mode alone does not make images comparable.
- Best suited to exploratory research workflows. This tool is not a clinical diagnostic system.
Frequently Asked Questions
Is any image data uploaded to a server?
No. The image stays in your browser. All stain separation and quantification are executed client-side.
Which stain combinations are supported?
The current implementation is tuned for hematoxylin + DAB. It does not expose a custom stain-matrix editor in the UI.
Can I use it on a full whole-slide image?
Not directly. Large slides should be cropped to a representative field before analysis so the preview remains interpretable and the browser stays responsive.
What does “positive tissue (%)” mean?
It is the fraction of pixels above the DAB threshold within detected tissue, not the fraction of the entire uploaded canvas.