AI Solution for PD-L1 Expression

Online Demo
  • 50 %

    Finds 50% more patients eligible for
    immunotherapy from the human-classified
    PD-L1(-) patient group.

  • 1M+

    Trained with large-scale training
    set including 1M+ cell annotations by 150+
    pathologists worldwide.

  • 96 %

    Demonstrated to perform TPS
    classification (above or below 50) with
    96% accuracy.

  • Lunit SCOPE PD-L1 TPS has
    received CE Mark.

  • Background
  • Our Vision
  • How It Works
Free Online Demo


What if AI can find
more patients
eligible for

  • Quantifying
    PD-L1 positivity
    with AI analysis

    Lunit lab in Samsung Medical Center, Seoul, Korea

  • Finding more PD-L1
    positive patients

  • Counting every PD-L1
    positive tumor cell

  • Complex is what
    AI does best

Our Vision

Quantification of
Expression Level
With AI

  • Trained on large
    datasets reviewed
    by 150+

    Lunit SCOPE PD-L1 analyzes every PD-L1 positive tumor cell and generates analysis results close to the ground truth

  • How AI Improves
    PD-L1 Assessment

    AS-IS (Only Human)

    • Discrepancy between pathologists
    • Non-reproducible

    TO-BE (With AI)

    • Objective result
    • Reproducible

How It Works

Enabling Better Immunotherapy
based on AI-powered PD-L1 Analysis

  • Objective
    analysis of PD-L1
    positivity using
    deep learning

    * Lunit SCOPE PD-L1 is currently applicable to NSCLC (non-small cell lung cancer) only.

  • Generated by
    Lunit SCOPE PD-L1

    • Detection

      Detects tumor cell PD-L1 +/- and inflammatory cell PD-L1 +/- from whole slide images.

    • Quantification

      Calculates PD-L1 positive score including TPS (Tumor Proportion Score) and CPS (Combined Positive Score).

    • AI Score and Report

      Provides AI analysis results including PD-L1 TPS score and map within a report.

  • Improving PD-L1 test efficiency through
    artificial intelligence

  • How Our AI Works

    • STEP 1

      Tissue slide (FFPE)
      slicing & IHC staining

    • STEP 2

      Digitization using a scanner
      Supports various scanner vendors

    • STEP 3

      File upload to server
      (both cloud or on-prem)

    • STEP 4

      Lunit SCOPE PD-L1 analysis

  • Lunit SCOPE PD-L1

    • Detection results for visualization
    • Quantification results of key features detected
    • Final score for response prediction

    *Lunit SCOPE PD-L1 is currently available for
    research use only

Studies featuring

Our solutions