DiaKwant™ PD-L1 CPS-TPS

Automated quantification

DiaKwant™ is validated across multiple tumor types, including NSCLC, SCLC, GC/GEJ, HNSCC, CC, and TNBC.

Automatically detects tumor regions and the surrounding microenvironment on IHC slides. DiaKwant™ algorithm identifies tumor and immune cells, quantifies PD-L1–positive cells, and computes CPS and TPS at the whole-slide level¹.

Scientific results showed that AI-assisted PD-L1 scoring improved accuracy (89% vs. 75% with routine methods), sensitivity (96% vs. 78%)^, and reproducibility across multiple carcinoma types, particularly in challenging CPS ranges¹.

Automated quantification

DiaKwant™ HER2

DiaKwant™ detects invasive tumor regions and performs cell- and membrane-level analysis on IHC slides. The algorithm quantifies staining intensity (0, 1+, 2+, 3+) and membrane completeness across tumor cells, providing an interpretable whole-slide HER2 score in line with ASCO/CAP and GEFPICS guidelines.

In a study presented at ESMO, DiaKwant™ HER2 scoring algorithm achieved 90% balanced accuracy, with 100% detection of HER2 3+ cases and 93% of HER2-Low cases, while improving inter-observer agreement from 57% to 75%².

DiaKwant™ Ki-67

Automated quantification

DiaKwant™ Ki67 automatically detects invasive tumor regions on IHC slides and quantifies proliferation at both the hotspot and whole-slide level. DiaKwant™ algorithm identifies tumor nuclei, exclude lymphocytes, classifies Ki67-positive versus total cells, and computes proliferation indices.

Results obtained showed that DiaKwant™ Ki67 model achieved high concordance with pathologist consensus, with an R² of 95% between predicted and ground truth proliferation indices. DiaKwant™ automatically detected hotspots and quantified proliferation at both hotspot and whole-slide levels³.

DiaKwant™ Mitosis

Automated quantification

DiaKwant™ Mitosis automatically detects invasive tumor regions on H&E slides and identifies mitotic figures.
‍
It highlights candidate mitoses, classifies them, and generates interpretable heatmaps and hotspot-based mitotic scores.

DiaKwant™ mitosis model reached a precision of 88% and a recall of 71%. Pathologists validated 80% of DiaKwant™ mitoses, leading to a 41% increase in consensus mitoses count and a 26% increase in individual pathologist counts⁴.

DiaKwant™ ER/PR

Automated quantification

DiaKwant™ ER/PR automatically detects invasive tumor regions and performs nuclear-level analysis on IHC slides. DiaKwant™ algorithm quantifies staining intensity and percentage of positive tumor nuclei, providing interpretable whole-slide scores.

Our evaluation study showed that the DiaKwant™ ER/PR model achieved high concordance with pathologist consensus, with an R² of 0.95 between predicted and ground truth nuclear positivity. DiaKwant™ automatically detected invasive regions and quantified ER/PR expression at both hotspot and whole-slide levels³.