PiNK Test
Our goal is to build tools that improve the outcomes for patients using scientific principles and evidence-based approaches. Cansera holds the exclusive worldwide commercial rights to both the quantified human performance status technologies and all the PiNK Test™ related intellectual property rights for the use of artificial intelligence for simultaneous use of clinical and molecular readouts.
PiNK Test™
Towards realizing PiNK Test, Cansera has established unique partnerships with the Convergent Science Institute in Cancer (CSI-Cancer) laboratory at the University of Southern California (USC) that has developed the blood-based liquid biopsy test for early breast cancer detection (see https://early.usc.edu). Blood-based cancer screening is ideally suited for widespread distribution to the community, but neither the companies who promote these tools nor health systems have the information technology infrastructure for communicating results and tracking people who were screened this way. The product vision for the PiNK Test is a complementary approach together with current standard of care for women at average risk for breast cancer and those where the current standard of care has reduced accuracy such as women with dense breasts. We are working with CSI-Cancer to facilitate community-based cancer screening using a blood test.
PiNK Logic for Deep Learning in Early Breast Cancer Detection
This next-generation analytics platform, is being developed to automate rare-event detection and patient-level classification using probabilistic deep learning, with the goal of achieving equivalent or better clinical performance with minimal manual intervention. Preliminary retrospective studies demonstrated 93% accuracy, supporting technical feasibility.
Core Innovations:
- Comprehensive blood analysis: Analyzes the entire cellular blood compartment rather than a limited subset, enabling a more complete liquid biopsy approach to complement breast cancer screening.
- Novel AI image processing: Uses optimized image tiles instead of traditional cell segmentation, improving computational efficiency and scalability for machine learning.
- Unsupervised rare-event detection: Identifies and ranks rare cellular events without prior assumptions or manual cell segmentation, creating a platform technology applicable across multiple diseases.
- Patient-level predictive AI: Combines rare-event signatures to classify disease status and predict recurrence or progression, with applications beyond breast cancer (e.g., bladder cancer, multiple myeloma).
- Explainable artificial intelligence (XAI): Provides transparent, interpretable predictions by identifying which rare events drive each classification, supporting clinician confidence and continuous model improvement.
- Scalable AI architecture: Incorporates low-precision computing, transfer learning, and domain adaptation to accelerate analysis, reduce computational costs, and enable commercialization at scale.