Taking Pathology AI Beyond the Demonstration Stage: Eight Companies Team Up to Enter the Drug Development Process

Pathology AI does not lack impressive algorithms. The real challenge is integrating models into regulated laboratories and clinical trials: the same result must remain consistent across specimens, staining batches, scanning equipment, and hospitals, while also allowing review by pathologists. To bridge this gap, precision medicine laboratory services company CellCarta has established a global digital pathology and AI consortium, aiming to integrate fragmented technologies into workflows that pharmaceutical companies can evaluate in practice.

The consortium consists of eight companies: CellCarta, Lunit, Mindpeak, Imagene AI, Dipath AI, Nucleai, Tivenix, and Indica Labs. Its design is not tied to a single algorithm. Instead, it allows biotechnology and pharmaceutical companies to select different image analysis, spatial biology, multimodal AI, or precision diagnostics tools based on the disease, stage of development, and regulatory requirements.

Specific applications can begin with tumor tissue sections. After pathology images undergo whole-slide scanning, AI can quantify immunohistochemical staining, identify cellular composition within the tumor microenvironment, or detect features associated with molecular status in routine H&E-stained images. If appropriately validated, this information can be used to explore biomarkers, screen patients who may benefit, and support clinical trial stratification and companion diagnostic strategies.

The capabilities provided by each member differ. Lunit said it will contribute immunophenotyping, marker analysis, immunohistochemistry quantification, and tumor microenvironment analysis; Mindpeak focuses on H&E, immunohistochemistry, and multiplex immunofluorescence imaging; Nucleai applies tissue image analysis to development programs including antibody-drug conjugates. Tivenix extends the scope to liquid biopsy, using methylation markers to assess organ-specific cell death signals; companies including Indica Labs provide pathology platforms and clinical workflows.

Within this framework, CellCarta serves as the integrator, providing pathologist oversight, specimen logistics, whole-slide scanning, analytical method development, and histology, genomics, immunology, and proteomics laboratory capabilities. Lunit and Mindpeak separately confirmed their status as founding members; South Korean media also reported that Lunit and CellCarta have collaborated since 2025 and expanded their work in 2026 to include the development of AI companion diagnostics and preparations for commercialization.

The consortium says it can serve oncology, autoimmune diseases, and neurodegenerative diseases, but it has not disclosed any partnered drugs, clinical trials, training or validation datasets, nor has it provided results for sensitivity, specificity, or interlaboratory reproducibility. Whether the members’ tools can be combined into compliant companion diagnostics remains subject to clarification regarding algorithm version control, data bias, division of responsibilities, and performance across different scanners and specimen-processing conditions.

The formation of the consortium therefore represents a drug development services model, rather than confirmation of a new biomarker. It seeks to place model selection, laboratory validation, and clinical deployment within a single framework; whether it can truly shorten development timelines will depend on whether it can subsequently produce prospective trials, multicenter validation, and evidence acceptable to regulators.