Mindpeak Gastric PD-L1

Mindpeak Gastric PD-L1 is an automated image analysis software module developed by Mindpeak for detecting and quantifying programmed death ligand 1 (PD-L1) in gastric cancer tissue. It is designed for use in digital histopathology workflows and targets researchers and pathologists working with immunohistochemically (IHC) stained whole slide images. The tool is intended for Research Use Only and is not for use in diagnostic procedures.

The software operates as a 0-Click solution, meaning that with background image pre-calculation, all detected and classified cells are instantly visualized upon loading a digital slide. It was developed with lab-specific variations in mind — including differences in slide preparation, staining, and imaging — to provide consistent results across different laboratory environments.

Output and Scoring

• Detects and classifies IHC-stained tumor cells and mononuclear inflammatory cells within whole slide images.
• Tumor cells are categorized as either not stained (Tumor negative) or stained (Tumor positive).
• Mononuclear inflammatory cells are classified as stained (Positive mononuclear inflammatory cell).
• Calculates a Combined Positive Score (CPS), representing the proportion of PD-L1 positive cells relative to the total number of PD-L1 positive and PD-L1 negative tumor cells.

Workflow Steps

1. Retrieve or select a whole slide image with IHC PD-L1 staining suited for automated analysis.
2. The results of the whole slide analysis are displayed automatically.
3. Review, adjust, and correct the automated analysis results using an integrated interactive viewer.
4. Make a final analytic decision based on the reviewed results.

Key Capabilities

• Accurately identifies PD-L1 positive and negative tumor cells as well as positive mononuclear inflammatory cells.
• Background pre-calculation enables instant visualization of all detected and classified cells upon slide loading.
• Developed to handle typical lab-specific variations, supporting reliable results across different preparation, staining, and imaging conditions.
• Compatible with a wide range of scanners and designed for integration into existing digital pathology workflows.

Clinical Reference Data

• In a published study, the AI demonstrated higher agreement with pathologists than pathologists showed with each other.
• The AI showed higher correlation and better concordance (+15%, p<0.05) in classifying patients compared to pathologist-to-pathologist agreement.
• The AI detected over 50% more positive cases (46 vs. 30.3), suggesting potential to identify more patients eligible for targeted PD-L1 therapy.
• Results were published via ASCO (https://ascopubs.org/doi/10.1200/JCO.2024.42.16_suppl.2633).

Technical Requirements

• Mindpeak Gastric PD-L1 is a stand-alone software module that must be integrated into an interactive image viewer that supports manual selection or correction of regions of interest and cell-level review and adjustment of results.
• Integration is supported via the Mindpeak AI Interface Definition API.
• The software runs as a web service, available either in the cloud or on a dedicated on-premise server within the laboratory network.
• The cloud variant is specified and provided by Mindpeak or its distribution partners and requires no additional hardware from the user.
• For on-premise deployment minimum hardware requirements, direct contact with Mindpeak is required.

Mindpeak GmbH, founded in 2018 and headquartered in Hamburg, Germany, also maintains a presence in Cambridge, MA, USA. The company builds AI and deep learning-based automation tools for visual diagnosis in oncology pathology, with a focus on reproducible quantification of cells and biomarkers. Explainable AI (xAI) outputs are incorporated to support pathologist trust in the results.