← Back to Home

Finding Clues for Drug Selection in Tissue Slides: CellCarta and Imagene Advance Two Pharma Pathology AI Programs

The companies’ collaboration has expanded into companion diagnostic development and lung cancer biomarker prediction, seeking to connect image analysis with drug trials. However, the announcement provides no new performance data, and the lung cancer prediction tool remains limited to research use.

By SURL BioNews

A tumor tissue slide can reveal not only cell morphology but also potential clues for selecting patients for drug trials. But turning signals read by artificial intelligence into reliable testing tools still requires clearing the hurdles of validation and clinical deployment. On October 1, CellCarta and Imagene AI announced an expanded collaboration, disclosing that they are working on programs with two global pharmaceutical companies to bring pathology image analysis into companion diagnostic development and non-small cell lung cancer research.

According to CellCarta’s announcement and SelectScience’s report the following day, the collaboration includes two distinct tracks: a companion diagnostic development program using AI to analyze immunohistochemistry (IHC) staining, and another using Imagene’s lung cancer biomarker prediction panel in non-small cell lung cancer. The pharmaceutical companies’ names, associated drugs, and trial details have not yet been disclosed. The two companies said they would provide more information in the coming months.

In the IHC track, Imagene’s OI CDx technology is used to quantify the expression of biomarkers and drug targets. Its analysis covers cells and subcellular structures and provides continuous numerical values and spatial distribution scores to support patient stratification. This information is intended to turn staining signals in tissue slides into comparable metrics for evaluation in drug development. The announcement has not yet specified which markers or decision thresholds the program uses.

The lung cancer program uses routine hematoxylin and eosin (H&E)-stained whole-slide images. The company says its multigene lung cancer prediction panel can estimate key biomarkers in non-small cell lung cancer within minutes, with the aim of helping biomarker-directed trials identify potential study participants. The outputs here are image-based estimates; validation data are still needed to establish how closely they agree with actual test results.

CellCarta is responsible for analytical validation, global clinical trial deployment, and companion diagnostic development and regulatory support within the collaboration, connecting the imaging tools with laboratory and trial workflows. This expansion builds on the relationship established when Imagene joined CellCarta’s Digital Pathology and AI Alliance in July 2026, extending it into development programs that already involve pharmaceutical companies.

However, progress in the collaboration and test performance are two issues that need to be assessed separately. This announcement provides no new validation sample counts, dataset composition, sensitivity, specificity, or performance across centers, nor does it provide results sufficient to assess clinical benefit. The lung cancer prediction panel is explicitly labeled “for research use only” and must not be used in diagnostic procedures. Program details and validation evidence disclosed next will help determine whether these tissue-slide signals can reliably support patient selection and subsequent companion diagnostic development.

References

  1. CellCarta via PR Newswire
  2. CellCarta
  3. Imagene AI
  4. SelectScience