The measurement needs to be reliable
A biomarker on a slide is a measurable signal that tells a pathologist something specific about the tissue. In oncology, the most important ones tell whether a tumor is likely to respond to a particular treatment.
What is being measured
Immunohistochemistry stains tissue for a specific protein, and the pathologist scores how much of it is present and where.
The classic example. A breast cancer scored HER2 positive is a candidate for HER2 targeted therapy, and the scoring scale, from 0 to 3+, decides eligibility.
A more recent target with therapies now directed at it.
Marks cytotoxic T cells. CD8 density in and around a tumor is an indicator of immune activity that informs immunotherapy decisions.
Hematoxylin and eosin — the standard general stain that shows tissue structure and is the baseline for most morphological assessment.
Each marker has its own scoring convention. Some count the proportion of positive cells, some grade staining intensity, some combine both into a composite score. The conventions exist to make scoring comparable across pathologists, and they only work if applied consistently.
Why consistency matters
Inter observer variability is well documented across pathology. Two experts scoring the same HER2 slide will disagree at some rate, especially in borderline cases — and borderline cases are exactly the ones where the treatment decision hangs on the score. Fatigue, slide quality, and staining variation all add noise.
The consequence can be significant. A score on the wrong side of a treatment threshold can lead to the wrong therapy.
How software quantifies
Tissue is segmented from background, and tumor regions are separated from stroma and other tissue.
Individual cells are detected and their nuclei segmented.
For each cell, staining is measured — intensity for a membrane or nuclear marker, presence for a cell type marker.
The results are aggregated into the score the clinical convention calls for: proportion positive, intensity grade, density per area.
Every step has failure modes. Stain variation between labs and scanners can shift intensity measurements, so normalization is essential. Overlapping cells confuse segmentation. Artifacts, folds, and out of focus regions have to be detected and excluded. A quantification pipeline is validated against expert scoring on held out slides from multiple sources before it is trusted.
The role of the pathologist
Quantification does not remove the pathologist. It gives them a consistent, reproducible measurement to review, override where their judgment says otherwise, and sign off on.
The goal is to make routine scoring faster and more consistent, while giving specialists the cases that need closer review.
SequoiaAT's digital pathology platform uses computational models to analyze staining for HER2, Trop2, CD8, and H&E, with outputs that support high accuracy scoring and help detect and characterize cancer subtypes.
Slide scoring and validation steps are automated within the pipeline, and the measured effect was better consistency, faster turnaround, and fewer human errors in routine assessment.
Read the case study →